Further reading — related pieces of mine on this site:

• Understanding the Israeli–Palestinian Conflict, Part 1

• Understanding the Israeli–Palestinian Conflict, Part 2

• Understanding the Israeli–Palestinian Conflict, Part 3

• UN Resolution 181 for Palestine: The Stupidest Resolution Ever Passed

• No Going Back: What the Palestinians Must Accept Before It’s Too Late

• Iran and the Little Satan/Great Satan Con

• What the UN Is — and What It Is Not

Why There Isn’t Already a Palestinian State

There has never been an independent Palestinian state. Not once, not ever. The territory was Ottoman provincial land for four hundred years, from 1517 until World War I. Britain took the Mandate for Palestine from the League of Nations in 1922, administered it for a quarter century, and handed the problem to the newly formed United Nations in 1947 when it became unworkable.

The UN’s answer was Resolution 181: partition the Mandate into a Jewish state and an Arab state. On paper, the Jewish state got the larger share — 56% of the land to the Arab state’s 43% — despite Jews being roughly a third of the population. In practice, about 60% of that Jewish allocation was the Negev desert: arid, uncultivated, unsuited at the time for agriculture or urban development. The Arab state got the fertile highlands and the water table. Jewish leadership accepted the deal anyway and made the desert productive through decades of irrigation engineering most of the world thought was impossible. The Arab Higher Committee and the Arab League rejected it outright — not as an opening bid, but as a matter of principle. They chose war instead.

The man who spoke loudest for the Palestinian Arabs going into that vote was Amin al-Husseini, the Grand Mufti of Jerusalem. He hadn’t been elected to anything. The British appointed him Mufti in 1921 and made him president of the Supreme Muslim Council in 1922 — a colonial administrative decision, not a popular mandate. When the 1936–39 Arab Revolt made him inconvenient, the British stripped his titles and issued a warrant for his arrest. He fled to Lebanon, then Iraq, then landed in Berlin, where he spent the war courting Hitler, lobbying the Nazis to block Jewish emigration out of occupied Europe, and helping recruit a Bosnian Muslim division for the Waffen-SS. He was still angling for Axis recognition as leader of a future Arab nation as late as 1943. That was the leadership structure representing Palestinian Arabs when the UN put a state on the table in 1947: a fugitive Nazi collaborator with no institutional legitimacy and no interest in building one.

There was no leadership capable of accepting a state, and there was no unified Arab position wanting one — for the Palestinians. Five Arab armies invaded the day Israel declared independence, but they weren’t fighting for a Palestinian state. They were fighting each other for the real estate. Jordan’s King Abdullah had designs on the West Bank before the war even started, and he acted on them: Jordan occupied the territory in 1948 and formally annexed it in 1950, over the objections of Arab League members who still wanted an actual Palestinian state to exist. Egypt took Gaza and ran it as an occupied military zone, not a path to sovereignty — Egyptian authorities wouldn’t even let Gazans move freely into Egypt proper. The Arab League’s own fig leaf, the All-Palestine Government installed in Gaza in September 1948, was a facade with no money and no power, created mainly to box out Abdullah. Nasser dissolved it in 1959 and ruled Gaza directly from Cairo.

By the 1949 armistice, the land the UN had earmarked for a Palestinian state was split between two Arab governments — Jordan and Egypt — not handed to Palestinian self-rule. That arrangement held for nineteen years, until 1967. Nobody in Amman or Cairo was in a hurry to change it.

The pattern set in 1947–48 repeats for the rest of this piece: every time a Palestinian state has been on the table, the obstacle hasn’t been the absence of an offer. It’s been the absence of anyone — Palestinian or Arab state — willing to accept one on terms that didn’t include eliminating the other side first.

What Role Did Israel and the Palestinians Actually Play?

Almost none. That’s the part the standard narrative skips.

The Jewish Agency accepted Resolution 181. Once Arab forces attacked, Israel’s role in the war that followed was defensive — surviving an invasion by five Arab armies who intended to eliminate the new state entirely, not negotiate its borders.

The Palestinians didn’t get a vote either — not from the UN, and not from their own leadership. No referendum was ever held asking Palestinian Arabs whether they’d accept a state alongside a Jewish one. The decision to reject partition was made by the Arab Higher Committee, an unelected body under al-Husseini’s orbit, and ratified by the Arab League. The AHC rejected not just the partition plan but also the alternative minority proposal for a single binational state — meaning it foreclosed every option on the table, on behalf of a population it never consulted.

So the two parties actually named in the 1947 plan — a Jewish state and an Arab state — split into two very different postures. One accepted the deal and then fought to survive the consequences of the other side’s rejection. The other never got asked. A handful of unelected men made the call, chose war, and the Palestinian population absorbed the loss: no state, land partitioned between Jordan and Egypt, and a leadership vacuum that would define the next several decades.

This is the first instance of a pattern that recurs through every subsequent chance at statehood: the decision gets made above the Palestinians’ heads, by people who aren’t accountable to them and whose interests don’t fully overlap with theirs.

Lather, Rinse, Repeat

The 1947-48 sequence didn’t happen once. It happened on a loop, with the same three ingredients every time: maximalist demands from surrounding Arab powers, eventual Palestinian leadership emerging to inherit the wreckage, and a return to war rather than acceptance of a deal on the table.

1967. Israel won the Six-Day War and, within weeks, was reportedly waiting for a phone call offering peace in exchange for the captured territory. The call never came. Instead, eight Arab heads of state met in Khartoum and issued their answer: no peace with Israel, no recognition of Israel, no negotiations with it. The Three Noes weren’t a Palestinian position — the Palestinians still had no state and no seat at that table. It was regional powers foreclosing any settlement on principle.

1964-68. Actual Palestinian leadership finally showed up in the form of the PLO, founded at a Cairo Arab League summit in 1964. Its founding charter didn’t call for a state alongside Israel — it called for Israel’s elimination. The 1968 revision was explicit: armed struggle as the only way to liberate Palestine, not a tactic but the overall strategy. For the next quarter century, the PLO’s answer to “how do we get a state” was war, not diplomacy — plane hijackings, cross-border raids, the 1972 Munich massacre, the Achille Lauro murder of a wheelchair-bound American in 1985. It took until 1988 for Arafat to even claim he’d accept Israel’s existence, and until the 1993 Oslo Accords for the PLO to put that in writing.

2000. With a Palestinian Authority now in place under Oslo, Israel made its most concrete offer yet. At Camp David, Prime Minister Ehud Barak offered roughly 91-94% of the West Bank, all of Gaza, and a capital in East Jerusalem. Arafat walked away without a counteroffer. Weeks after Camp David collapsed, the Second Intifada began — a five-year campaign of suicide bombings that killed over a thousand Israelis and devastated the Palestinian economy along with it. Five months later, Clinton produced a further improved set of parameters, and in January 2001 Palestinian negotiators at Taba finally did put detailed counterproposals on the table for the first time — but Barak broke off the talks under pressure from an Israeli public radicalized by months of bombings. Even the accounts most sympathetic to Arafat place the first real Palestinian counteroffer after the violence had already started, not before.

2008. Prime Minister Ehud Olmert made an even more generous offer to Arafat’s successor, Mahmoud Abbas — Israeli withdrawal from roughly 94% of the West Bank with land swaps, a shared arrangement over Jerusalem’s holy sites, and a mechanism for a limited number of refugees. Abbas didn’t reject it outright. He simply never answered it. By his own negotiator’s account, the Palestinian position was to hold the 1967 lines “without budging an inch” — a maximalist floor, not a negotiating position.

2006-2007. Palestinians did get an actual vote this time — for a parliament, not for statehood, but the closest thing to a referendum on leadership the story has. They elected Hamas, a party whose founding charter calls for Israel’s destruction. Fatah and Hamas then fought each other for control; Hamas won Gaza outright in June 2007, killing Fatah members in the process, while Fatah held on in the West Bank. Palestinian politics has been split into two governments — one that negotiates, one that doesn’t recognize Israel’s right to exist at all — ever since.

The shape doesn’t change. Regional or factional maximalists set terms no Israeli government can accept, or Palestinian leadership rejects terms without a serious counteroffer, and the interval between offers gets filled with rockets, bombings, or invasion. Every generation gets a new leadership, a new war, and the same non-outcome.

The Neighborhood

Every actor discussed so far — the Arab League, Jordan, Egypt, the regimes that set the terms Palestinians were expected to accept or reject — comes out of the same political mold. There isn’t a democracy in the bunch.

Saudi Arabia and Oman are absolute monarchies, ruled by decree with no legislature that can override the king or sultan. Jordan, Morocco, Bahrain, Kuwait, Qatar, and the UAE are constitutional monarchies — a king, emir, or sheikh sits above the system regardless of what any parliament does, and in most of them the ruling family also dominates the government ministries and the economy. Egypt is a republic on paper and a military autocracy in practice — its current president took power via a 2013 coup and has run the country through security-state methods ever since. Iran is a theocracy, with a Supreme Leader who outranks any elected president. Syria’s post-Assad government is a transitional military authority, not a democracy. Iraq holds elections, but Freedom House still rates it Not Free, given the militias, sectarian patronage networks, and Iranian influence that shape outcomes regardless of vote totals. Yemen has been a failed state and civil-war battleground since 2014.

None of these governments answer to their own populations the way a democracy does, and that matters for the Palestinian question specifically: every regional actor that has weighed in on Palestinian statehood — funding it, blocking it, using it as leverage — has been an unelected government pursuing its own interests, not an electorate expressing one.

They don’t agree with each other, either. Broadly, the Gulf monarchies — Saudi Arabia, the UAE, Bahrain — along with Jordan and Egypt have aligned with the United States and, increasingly, normalized or quietly cooperated with Israel. Iran sits on the other side, backing Hezbollah, the Houthis, and until Assad’s fall, the Syrian regime, as part of a rejectionist axis explicitly built around Israel’s destruction. Qatar plays both sides — a U.S. military-basing partner that also bankrolls Hamas. Turkey, nominally a democracy with actual elections, has drifted into its own semi-authoritarian lane under one-man rule and swings between NATO obligations and open hostility to Israel depending on the week.

That’s the board the Palestinians have had to play on for eighty years: not a unified Arab world with one coherent position, but a scattered set of monarchs, generals, and clerics, some friendly to the West and some explicitly organized against it, none of them accountable to a ballot box, all of them treating the Palestinian cause as leverage for their own legitimacy rather than a project they were actually trying to finish.

Strip away the century in between and the map looks familiar. Before the Ottomans, the region was ruled by sultanates, caliphates, and dynasties — power concentrated in a ruling family or a court, no ballot box, no accountability to the governed. The Ottomans were simply the largest version of that same model. Four hundred years of empire, a world war, a British mandate, and a wave of independence movements later, the region has settled back into essentially the same arrangement it started with: monarchs and strongmen ruling by inheritance or force, not by consent. The furniture changed. The structure didn’t.

Iran Is a Different Animal

Every other regional actor in this piece treats Israel as leverage for domestic legitimacy. Iran is the exception, and it’s worth separating out, because for Tehran, Israel isn’t actually the point.

Persia was a great empire long before Islam existed — Cyrus, Darius, a civilization that governed a hundred peoples centuries before Muhammad was born. Arab armies conquered it in 651 CE, but unlike Syria or Egypt, Persia never became Arab. It kept its language, its culture, its identity, and its memory of having run the region once. In 1501, Persia’s rulers made Shia Islam the state religion — partly theology, mostly strategy, a way to mark Persia as distinct from the Sunni powers pressing in on both sides. From that point on, Persian imperial ambition and Shia religious identity fused into one project.

The 1979 regime inherited both halves. It also inherited a problem: a Persian, Shia theocracy, isolated, sitting in a region that’s overwhelmingly Arab and Sunni. It needed a cause no Sunni government could publicly oppose. It found one in Israel. Khomeini declared an annual Jerusalem Day within his first year in power, and Tehran spent the following decades building Hezbollah and bankrolling Hamas — turning a Persian outsider into a self-styled pan-Islamic champion, and turning every Sunni government quietly coexisting with Israel into an implicit traitor by comparison. Under the banner of erasing Israel, Iran spent those same decades embedding proxy armies across Lebanon, Syria, Iraq, and Yemen — infiltrating the entire region proxy army by proxy army, with a common enemy providing cover no one could openly object to.

The declared goal — destroy Israel, destroy America — is the cover story. The actual project is regional dominance: a nuclear program as the hard-power expression of restored Persian primacy, and a proxy network across Iraq, Lebanon, Syria, and Yemen as the expression of Shia religious leadership. Israel is roughly the size of New Jersey. It isn’t rational for a country to spend fifty years and untold billions trying to erase a speck that small — unless the speck is the vehicle for something much larger, not the actual destination. (The fuller argument — including the Persian-Arab and Sunni-Shia fault lines underneath all of this — is laid out in more detail here.)

What’s In It For Them

If none of these governments are accountable to their own people, the obvious question is why they’d keep fighting over a strip of land roughly the size of New Jersey. The answer isn’t sentiment. It’s leverage.

An unelected government needs a source of legitimacy that doesn’t come from a ballot box. “Standing with Palestine” has been one of the cheapest, most durable ones available in the Arab world for eighty years — a cause that costs a regime nothing in cash or territory, unifies a population that might otherwise be asking why it doesn’t get a vote either, and gives an external enemy to point to whenever the internal numbers look bad. The Arab League formalized this instinct into policy: as far back as the 1950s, member states adopted a deliberate rule that Palestinian refugees would not be granted citizenship anywhere in the Arab world, on the stated rationale that assimilating them would dissolve their claim to return and, not incidentally, would recognize Israel by implication. That’s not a policy designed to help Palestinian families rebuild their lives. It’s a policy designed to keep a stateless population in reserve as a permanent talking point.

The record backs up the cynicism. Lebanon still refuses Palestinians citizenship after three generations born on its soil, keeping hundreds of thousands in camps like Ain al-Hilweh that the Lebanese army itself periodically raids. Syria offered residency rights but never naturalization. Egypt ran Gaza for nineteen years and never so much as floated Palestinian self-rule there. And when Palestinian leadership stopped being useful — or actively inconvenient — the hospitality evaporated instantly: after Arafat publicly sided with Saddam Hussein during the 1990 invasion of Kuwait, Kuwait expelled roughly 400,000 Palestinians, most of whom had lived and worked there for decades, in a matter of months. It was, by proportional scale, one of the largest single displacements of Palestinians since 1948 — inflicted by a fellow Arab state, over a political dispute that had nothing to do with the people it punished.

None of this reads like a region that ever put Palestinian welfare above its own governments’ interests. It reads like a cause that’s been useful precisely because it never gets resolved — a grievance regimes can keep pointing at instead of a population they were ever in a hurry to actually see governed, housed, or free.

It’s also a convenient theme for occupying the subjects of these same regimes. Cast Israel as the villain and the Palestinians as the perpetual victim, and a king or a general gets something he can’t manufacture domestically: a righteous cause, a common enemy, and free sympathy — all without holding an election, building an institution, or answering for the state of his own country. It’s a lot easier to keep a population angry at Israel than to explain why a monarch’s family owns the economy, why a general who took power in a coup still runs the country a decade later, or why there’s no parliament worth the name anywhere in the neighborhood. The Palestinian cause does the work that a free press and a ballot box would otherwise be doing. Every regime in the region has figured out that it’s cheaper to rent a grievance than to earn legitimacy.

What Do You Propose?

Before getting to a proposal, ask the honest question: what has Israel actually gotten for the risks it’s already taken?

In 2005, Israel didn’t wait for a treaty. It unilaterally tore down every one of its 21 settlements in Gaza, forcibly removed 8,000 of its own citizens — some of them dragged out of their homes screaming — and handed the entire territory to Palestinian control with no Israeli soldier, settler, or checkpoint left inside it. Critics point to a Sharon aide’s since-disputed remark that the withdrawal was designed to freeze the peace process rather than advance it — but motive is a debate for historians. Outcome isn’t. Whatever Sharon’s internal calculus, the facts on the ground were a full, unilateral withdrawal with nothing held back. That’s about as close to “just give them a state and see what happens” as a country has ever tried. Within two years, Hamas had seized Gaza in a coup against its own Palestinian rivals. Within eighteen years, that same territory produced October 7 — the deadliest single day for Jews since the Holocaust, and the first large-scale invasion of Israeli territory since 1948. Rockets, tunnels, and a cross-border invasion force were the return on a unilateral land-for-peace bet nobody forced Israel to make.

That’s the actual track record being asked to repeat itself. So when the word “genocide” gets thrown around by people who couldn’t find Gaza on a map two years ago, consider the source doing most of the certifying. The UN Commission of Inquiry finding and the rest of the UN apparatus that laundered it into headlines are not neutral arbiters — as I’ve written elsewhere, the UN is a political body that packages its outputs as objective fact, run by a General Assembly stacked with the same autocracies and theocracies cataloged above, none of which answer to an electorate and several of which have spent eighty years treating the Palestinian cause as leverage. Their institutional “finding” isn’t a verdict. It’s a vote, dressed up as one. What actually happened is a war Israel didn’t start, against an enemy that hid its command structure under hospitals and its rocket stores under schools, fought in a strip of land nine miles across at its narrowest point, where every square mile is within range of the enemy’s own weapons. That doesn’t make the civilian toll anything less than a tragedy. It does make “genocide” — a word with a specific legal meaning involving intent to destroy a people, not a byproduct of urban warfare against an enemy using civilians as cover — a term doing propaganda work for exactly the governments with the most to gain from Israel staying occupied with an unwinnable war of optics.

So what’s the actual proposal? Not another summit that produces a signature and no enforcement. Every failed attempt in this piece — 1947, 1967, 2000, 2008 — had the same design flaw: it assumed good faith and asked for none of it to be verified. A state gets recognized, or funded, or armed, and everyone hopes the other side behaves. That’s not a peace process. That’s a bet placed with someone else’s chips.

The alternative is sequencing tied to verified performance, not promises. Concrete steps happen first — demilitarization of Gaza, dismantling of the tunnel network, an internationally verified end to the flow of Iranian weapons and cash — monitored by a mechanism with actual enforcement power, not a UN resolution nobody enforces. Statehood, if it comes, comes after that record is established and verified, not as a leap of faith that a fundamentally different outcome will emerge from the same unverified promises that failed in 1993, 2000, and 2008. Israel doesn’t owe anyone another unilateral experiment. The next one, if there is one, gets earned in the order that keeps Israelis alive first.

Before You Ask, Define It

So here’s the question back to anyone demanding “a Palestinian state” as if it were a simple, overdue fix: what does that actually look like?

Because it’s been offered. Repeatedly. In writing, with maps, with percentages attached. 1947: a state on 43% of Mandatory Palestine, rejected before it existed. 2000: roughly 91-94% of the West Bank, all of Gaza, a capital in East Jerusalem — rejected, no counteroffer. 2008: an even larger offer with land swaps and a mechanism for refugees — never answered at all. Every time, the state was on the table and the answer was war, or silence, or both. That’s not a people waiting for someone to finally propose a state. That’s a leadership structure, generation after generation, that has had the thing demanded of Israel and turned it down.

So when someone says Israel needs to “allow” a Palestinian state, ask them what they actually mean. A state that recognizes Israel’s right to exist, demilitarizes, and lives beside it — the thing offered three separate times and rejected three separate times? Or a state built on the actual charters of the actual parties who’d run it — the PLO’s founding document calling for Israel’s elimination, Hamas’s charter calling for the same — where “Palestine” is understood, in the plain language of its own leadership, to mean all of it, with Israel gone?

Because those are two entirely different proposals wearing the same three words. The first one keeps getting offered and keeps getting rejected. The second one isn’t a state — it’s the removal of eight million Jews from a country roughly the size of New Jersey, nine miles wide at its narrowest point, with nowhere else to go and no army in the world currently pledged to escort them out at gunpoint. Ask the people demanding a Palestinian state which of the two they’re actually picturing. Most of them have never had to answer that question, because nobody’s made them. Somebody should.

And worth considering: a country that’s had its neighbors try to erase it from the map in 1948, 1967, and 1973; that’s absorbed suicide bombings through the 2000s; that unilaterally handed over Gaza in 2005 and got Hamas and October 7 in return; and that just spent 2026 fighting a war against an Iranian regime that has spent decades and billions of dollars trying to finish the job — that’s a country that has some grounds for asking what’s actually different this time. Statehood has been offered before, more than once, on real terms, and rejected every time. So before the demand gets repeated again, it deserves a real answer: what, specifically, do you want Israel to do that it hasn’t already tried?

Check out my new app that unlocks the bias in articles: Bias Unlocker

From product to substrate

Search used to be a destination. You typed, you clicked a blue link, Google sold an ad against your attention. Now it’s a correctness input for AI models. A huge class of ordinary prompts can’t be answered from model weights at all — is so-and-so still the CEO, what does this cost, did the bill pass, who won last night. Without live retrieval the model isn’t slightly degraded on these; it’s confidently wrong about anything that moved after its training cutoff. That’s why every serious chatbot now searches the web before answering, silently, sometimes several times per prompt.

Those “several times” matter. Agentic workflows fan out — one question can trigger a dozen retrievals as the model decomposes it, cross-checks, and refines. On current Gemini models, Google bills grounding per search the model decides to run, not per prompt you type. Multiply across RAG pipelines, research agents, and coding assistants, and search stops being a feature. It’s the oxygen supply. So the question that matters isn’t whose model is best. It’s who owns the retrieval layer every model now depends on.

The physics of the index

Google dominates the index layer. The only other company that maintains a full-scale Western index today — Microsoft — holds roughly a tenth of Google’s scale. Everyone else rents, and has for a generation: DuckDuckGo never had an index of its own, Yahoo abandoned its own in 2009, and nearly every other “alternative” engine you’ve heard of syndicated Bing’s or Google’s. The retail search market wore a dozen masks over two wholesalers, one of them ten times the other.

And look at what even distant second place cost. Microsoft spent two decades and billions on Bing — with Windows distribution, Azure infrastructure, and an OpenAI partnership — and got an index in the low tens of billions of pages against Google’s hundreds of billions spanning more than a hundred petabytes. Twenty years and near-unlimited capital bought a tenth of the incumbent’s scale. That is the entry price for junior membership in the club. Nobody else has paid it, and nobody is trying — because the ladder has two cliffs, not one. Microsoft is a tenth of Google, but it still holds what no one below it has: decades of real consumer query traffic through Bing, Windows, and Edge feeding its ranking signals, and crawl infrastructure of a maturity no independent can match. Brave’s pages don’t come with twenty years of behavioral data. Second place is a distant second and still unreachable.

The gap never closed because an index isn’t built; it’s maintained. The moat is the perpetual recrawl: hundreds of billions of pages refreshed continuously, news revisited in minutes, a spam-filtering apparatus hardened by twenty-five years of war with the SEO industry, ranking signals distilled from trillions of human queries. And the index doesn’t sit alone — it’s fused to the Knowledge Graph, Maps, Shopping, YouTube, and the telemetry of Chrome and Android: a continuously refreshed model of the world. You don’t catch up to that with capital. You’d have had to be doing it since 2000.

Then, in 2025, the junior partner stopped wholesaling. Microsoft killed the open Bing Search API and pointed developers at “Grounding with Bing Search” inside the Azure Agents stack: $35 per thousand transactions, a steep increase over the old API for most workloads, and no longer a search API at all — raw results are never exposed, only a Microsoft model’s synthesis of them. Microsoft didn’t give up on search; it gave up on selling it. It adopted the dominant player’s playbook instead: never rent out the eyes raw — only rent your own brain wearing them, at a premium. When the distant second copies the leader’s enclosure strategy rather than undercutting it, the market isn’t being contested. It’s being farmed — one big farm and one small one, in its shadow.

The obvious rejoinder: maybe AI grounding doesn’t need Google-grade coverage — a curated, high-signal slice of the web plus a good model is “good enough.” But the general-assistant business lives in the tail curation excludes: the store hours, the spec sheet, the county ordinance, the forum thread where somebody already solved your exact problem. And the market has voted. The labs pay the toll today, and the ones rich enough to attempt independence — Perplexity with its partial stack, OpenAI with its reported in-house crawl — have spent fortunes without escaping it. If good enough were good enough, somebody with a hundred billion dollars would be acting like it.

Twelve months of receipts

August 2025: Bing’s API dies. January 2026: Google announces its own cheap Custom Search JSON API — $5 per thousand, the indie developer’s workhorse for nearly two decades — will return HTTP 410 Gone on January 1, 2027, steering everyone toward enterprise-priced Vertex AI. February 2026: Brave, now the only independent Western index at scale, kills its free tier and meters every developer at $5 per thousand. Brave’s own executives said it plainly: Bing’s exit made them “the only independent search API in the market at scale.”

The rate card

What it costs to give an AI model eyes, per 1,000 searches, July 2026:

Google — Gemini grounding (3.x) — $14
Google’s index, bundled to Gemini. Billed per search the model fires, not per prompt; model tokens extra.

Google — Gemini grounding (2.5) — $35
Same index, per-prompt billing.

Google — Custom Search JSON — $5
Raw Google results. Closed to new users; dead January 1, 2027.

OpenAI — web search tool — $10
Undisclosed index. Content tokens billed on top; some cheaper models charged a flat 8,000 input tokens per search.

Anthropic — web search tool — $10
Reportedly Brave’s index. Results billed as input tokens in that turn and every later turn.

Brave — $5
The last independent index (~30–40B pages). Free tier killed February 2026.

Exa — $7
Neural/semantic index. Smaller, curated coverage.

Tavily — $8–16
LLM-cleaned content over rented retrieval. Credit-based.

Microsoft — Grounding with Bing (Azure) — $35
Bing’s index, bundled to a Microsoft agent. Raw results never exposed; requires the Azure Agents stack; billed per tool invocation, and the model can invoke it multiple times per run.

Microsoft — Bing Search API — dead
Raw API decommissioned August 2025.

Two things jump out.

First, these prices are the entry toll, not the total. Retrieved content bills again as model tokens — often a multiple of the search fee itself. OpenAI charges some of its cheaper models a flat 8,000-token block per search. Anthropic’s results re-bill as input on every later turn of the conversation. On Gemini, the model decides how many searches to fire, invisibly, and you pay for each one. A modest product handling a million retrieval-backed prompts a month, at two searches per prompt, pays $20,000–$70,000 a month in search fees at the majors’ rates before a single token. Perplexity handled roughly 780 million queries in May 2025 — run that volume through any rate card and you understand why it spent a fortune building partial infrastructure of its own, and why almost nobody else can.

And the toll’s share is growing. Token prices collapsed an order of magnitude in two years — flagships repriced from $15 to $5 per million input tokens, budget models selling input at twenty cents. The search toll hasn’t moved: OpenAI and Anthropic at $10 since launch, Brave from free to metered, Bing from metered to enclosed. On a cheap model the search fee already exceeds the token cost of processing the results, and every model price cut shifts the ratio further. Inference is deflating like compute. The toll is priced like real estate.

Which hands Google a structural advantage at the model layer that has nothing to do with Gemini’s quality. Every rival’s live-world answer carries a real cash cost — a search fee paid to somebody. Google’s carries almost none: the index is already built, already running, already paid for by the ads business, and one more query against it costs a rounding error. As tokens race toward zero, the toll becomes the dominant marginal cost of a grounded answer — and Google is the only company that doesn’t pay it. In the endgame where models are commodities, every live-world answer costs Google nearly nothing and costs everyone else a fee that Google influences. You don’t need the best model to win that game. You need the cost floor, and only one company has it.

And the cost floor comes with a quality ceiling for everyone else. Nothing obliges Google to sell its best retrieval. What Gemini uses internally can be the whole crown jewels — the full ranking stack, the Knowledge Graph fusion, the freshness tiers, query understanding refined on a quarter century of traffic — while the grounding API exposes whatever slice Google chooses. Rivals can’t verify the gap, can’t buy the difference anywhere else, and can’t build it. And the incentive runs one way: every point of retrieval quality held back from the API is a point of advantage handed to Gemini. The toll road sells everyone else the service road and keeps the express lane for itself.

Second — my opinion, but the table argues it for me — these are monopoly prices. Google serves more than five trillion consumer searches a year for free, by its own disclosure, so the marginal cost of a query is a rounding error against these rates. Yet it prices grounding at $14–$35 per thousand, and the market arranges itself underneath: OpenAI and Anthropic, rivals who agree on nothing, both at exactly $10. Anthropic’s tool is reportedly Brave underneath — a clean 100% markup on a $5 input. Brave sits at $5 because with Bing gone and Google’s cheap tier dying there is nothing below it to fear. In a competitive infrastructure market, prices converge toward cost. Here they converge toward whatever Google charges, minus a courtesy discount. That’s a price umbrella, and the company holding it owns the index everyone else is approximating.

And the umbrella never gets tested, because nobody shops across it. At equal prices, everyone would choose Google’s index — it’s simply the best. But Google is also a competitor at the model layer, and that poisons the transaction twice over. A lab’s query stream is competitive intelligence of the highest grade — a live feed of what your users ask, what your agents are working on, where your model fails. And the queries are more than intelligence: they’re fuel. Google’s ranking quality was built on decades of human query traffic; chatbot traffic would hand it the equivalent signal for the AI era — how agents phrase requests, what they fan out into, which pages satisfy them. Route your retrieval through Google and you leak your product while paying to deepen the very moat you’re renting. So search is only sold bundled to each owner’s own models, no neutral market ever formed to discipline the prices, and the cleanest evidence is on the table above: Google is an investor in Anthropic and one of its cloud providers, and Anthropic still reportedly runs its search through Brave’s thirty-billion-page independent index rather than its own backer’s. When companies pay to not use the best product, the duopoly doesn’t need to collude. The structure of who-sees-whose-queries does it for them. And it explains the strangest number on the table: Microsoft charging Google’s $35 for an index a tenth the size. In a market where quality set prices, that would be laughable. In a market where half the customers can’t touch Google at any price — because Google is their competitor and their queries are its fuel — it’s simply what captivity costs. Microsoft isn’t pricing against Google. It’s pricing for Google’s refugees, and the $35 is its bet, written on a rate card: that the AI industry would rather pay a premium for the smaller index than hand its work to Google.

The stories everyone is talking about

Three adjacent stories get real attention. All three confirm the thesis.

Cloudflare has spent a year building tollbooths against AI crawlers — default blocks, a pay-per-crawl marketplace, now a “Pay Per Use” scheme — and said the quiet part loudly: Google’s crawler sees roughly twice the information available to any leading AI company, because Googlebot feeds search and AI through one pipe and no publisher will block it. They don’t want to. Being in Google’s index is where the traffic comes from, and an entire industry — SEO — consists of publishers doing free labor to be crawled better: sitemaps, schema markup, pages tuned to Google’s specifications. Every other AI crawler is an intruder to be blocked or billed. Google’s is a guest the web dresses up for. Nobody submits a sitemap to OpenAI. Even if regulators split Googlebot in two tomorrow, the search half keeps twenty-five years of index, query data, and freshness infrastructure that no ruling can redistribute — an index the world’s publishers improve daily, for free.

The licensing wave: as publishers wall off content, the premium head of the web is being carved up in direct deals — Axel Springer and Condé Nast with OpenAI, Reddit with Google at a reported $60 million a year. Some read this as a route around the crawled index. Two problems. Licensing covers the head, not the tail where the ordinary prompts live — no consortium of publisher deals covers the store hours and the spec sheets. And the bidding war is among companies whose content Google’s crawler still largely reads through the front door. Google bids when it wants to and abstains when it doesn’t. The closing of the web is an expense for everyone else and an option for Google.

The antitrust case: Judge Mehta’s September 2025 remedy declined to break off Chrome or Android, barred exclusive default deals while allowing paid placement, and ordered some sharing of index and behavioral data. Alphabet’s stock jumped eight percent on the ruling. The data-sharing provision hands competitors a snapshot, not a pipeline. The value is the machine that keeps the copy current, and the machine wasn’t on the table.

Free brain, no eyes

The open-source movement won a real victory: anyone can download a frontier-class brain, fine-tune it, run it on their own hardware. Model capability — the thing everyone spent two years treating as the moat — genuinely commoditized.

But a brain describing a world it stopped observing at its training cutoff is a brain in a jar. To be right about anything alive, it needs eyes — and the eyes were never open. The best pair belongs to a database in Mountain View that took a quarter century to build, that the only company to attempt a rival gave up matching and now farms alongside it, and that regulators declined to touch. Every AI company on earth now rents eyes by the query — from Google, from its junior partner, or from the one independent left — at rates set under Google’s umbrella.

Open weights got you a free brain. Nobody is handing out free eyes — and only one company owns a pair that sees the whole thing.

The core component of the AI that is setting the world on fire at the moment is the LLM.

Strip away the mystique and it is a simple program. It multiplies matrices against an astronomical amount of data, then does it again, billions of times.

The chip in your phone is running an operating system, a cellular modem, a camera pipeline, a GPU rendering a 3D interface, audio codecs, secure enclaves, neural accelerators for face ID, and a dozen background services — juggling interrupts and power states across all of them in real time. The computational complexity of what your phone does every second is orders of magnitude greater than what an LLM does.

What an LLM does is bigger, not more complicated. The workload is embarrassingly parallel matrix math on a scale nothing else in computing has ever required. That’s it. The magic isn’t in the operation. It’s in the amount.

This matters because moats are a protection requiring something hard to step over or go around. A workload this simple, this uniform, this well-understood, does not naturally support a monopoly on the silicon that runs it. You don’t need NVIDIA to multiply matrices. You need someone’s silicon to multiply matrices, at scale, with acceptable software support.

The question of who that someone has to be — and for how long — is the actual question.

The Architecture LLMs Actually Need

Because the workload is simple and uniform, the hardware to run it can be simple and uniform. A statically scheduled machine — one where the compiler decides in advance what happens on every cycle, and the hardware just executes — is enough. No branch prediction, no out-of-order execution, no speculative anything. The compiler knows the shape of the computation. It lays it out. The chip runs it.

For a workload this predictable, static scheduling is dramatically more efficient in power and performance than a GPU. NVIDIA’s chips are general-purpose parallel machines carrying decades of accumulated flexibility — graphics pipelines, gaming, scientific computing, sparse workloads, dynamic control flow. All of that costs silicon and watts. For matrix multiplication at scale, most of it is dead weight.

There’s a piece of folklore that static scheduling works for inference but not for training. It isn’t true. Google’s TPU is a statically scheduled machine. Google trained Gemini 3 end-to-end on TPUs and runs Search, Photos, Maps, and everything else on the same hardware. If static scheduling couldn’t handle training, Google would have discovered that at some point across a decade of shipping the world’s largest models on it.

The architecture the workload demands is simpler than what NVIDIA sells. It’s also cheaper to build, cheaper to run, and easy to design around. The general-purpose GPU is not the natural fit for this problem. It’s the incumbent that happened to be there when the wave arrived.

What NVIDIA Just Paid to Admit

In December 2025, NVIDIA paid $20 billion for Groq — its largest deal ever. Groq is a statically scheduled machine designed by Jonathan Ross, an ex-Google engineer who was on the original TPU team. The chip is built on the same principle as the TPU: the compiler lays out every operation in advance, and the hardware executes deterministically. No branch prediction. No dynamic scheduling. Silicon spent on math, not on hedging.

The deal was structured as a non-exclusive license plus a mass hire — a “reverse acqui-hire” designed to slip under Hart-Scott-Rodino review. Senators Warren and Blumenthal opened a Senate inquiry into the structure in March. Jensen Huang, at GTC 2026, framed Groq as completing NVIDIA’s GPU rather than competing with it — the same story he told about Mellanox. NVIDIA claims the combined system delivers up to 35× better tokens per watt than Blackwell alone.

Read that again. Thirty-five times better tokens per watt. That is not a companion. That is the admission.

The general-purpose GPU is not what the workload wants. The industry knows it. NVIDIA knows it — which is why they paid $20 billion to own the architecture that isn’t theirs.

They Don’t Know How to Build One

NVIDIA has no institutional experience with statically scheduled machines. Their entire history is dynamically scheduled parallel hardware — the GPU. Compiler-and-hardware co-design is a different engineering discipline, and it is not one you hire in a quarter.

Google has been shipping statically scheduled machines for roughly a decade, across seven TPU generations, with hundreds to perhaps a thousand engineers steeped in the discipline. That is only the public history — production silicon of that scope rarely appears without earlier internal experiments that never see daylight, and the TPU is almost certainly no exception. NVIDIA bought Groq. Groq is one branch of the TPU family tree, staffed by people who came out of Google. The rest of that tree is still at Google, or spread across the other statically scheduled programs in the industry. NVIDIA acquired one offshoot, not the discipline.

The Field NVIDIA Is Trying to Catch

Statically scheduled machines are the future of AI silicon. In Google’s case they are already the present, and have been for a decade. NVIDIA has no moat here. They came into this architecture late, and they are playing catch-up by buying someone else’s technology.

The field they are catching:

Hyperscaler in-house silicon. Google TPU (statically scheduled, powers Gemini and all of Google’s AI-serving infrastructure). Amazon Trainium and Inferentia. Meta MTIA. Microsoft Maia (architectural details less public, but designed in the same systolic-array family). Alibaba Hanguang, Baidu Kunlun, and Huawei Ascend on the Chinese side. Every one of these is a statically scheduled machine shipping at scale. NVIDIA has no moat in any of them.

Independent chip companies. Groq — now NVIDIA’s, but only after paying $20 billion for what NVIDIA didn’t build. MatX, compiler-first for LLMs. SiMa.ai at the edge. Intel’s Habana Gaudi, VLIW tensor cores, compiler-scheduled. Etched, transformer pipeline baked into silicon.

NVIDIA’s response to this field is not a superior general-purpose GPU. It is a Groq LPU sitting next to a Vera Rubin GPU in the same rack, with Huang selling the pairing as a completion story. The completion story is what companies tell when they have to buy the thing they should have built. It also rests on a false premise — that a GPU is still required for training, or for some class of workloads a statically scheduled machine can’t handle. The TPU has been doing all of it for a decade, without a GPU in sight.

The market is pricing NVIDIA as if the GPU wins the workload. The industry is spending tens of billions of dollars to make sure it doesn’t.

Four Trillion Dollars Buys You Competitors

A $4.66 trillion valuation is not a fortress. It is a signal flare. It tells every competent chip team in the world that the incumbent is charging monopoly prices for a workload that does not require monopoly technology. Capital, talent, and hyperscaler purchase orders follow that signal.

The switching cost story — the CUDA moat — is real for the long tail of legacy scientific computing, graphics, and mixed workloads. For LLMs it is thinner than NVIDIA’s price implies. Anthropic already runs across NVIDIA, Google TPU, and Amazon Trainium. OpenAI has a Cerebras deal for 750 megawatts. Every serious AI shop is multi-vendor by policy, because single-vendor dependency at these dollar amounts is malpractice.

The technical barrier to adopting a new chip is not what NVIDIA’s bulls think it is. You do not need a full PyTorch stack. You do not need decades of CUDA libraries. You need enough software to run your specific models — a compiler that lowers your graph to the target, a runtime, and a handful of kernels for the operations you actually use. That is a real engineering effort. It is not a decade-long moon shot. Everyone on the list above has done it, or is doing it, right now.

The workload is matrix multiplication. The bar for competing is: can your chip do the matrix multiplies at competitive tokens per watt, and can your compiler get the customer’s models onto it. That is the bar. That is not a $4.66 trillion bar.

Other Mousetraps

Statically scheduled machines are the sharpest bet against the GPU, but not the only one. There are more dynamic non-GPU architectures — dataflow, reconfigurable fabrics, wafer-scale integration — and a growing bench of special-purpose designs targeting specific model families. Beyond silicon there are photonics, in-memory compute, and other physical substrates being funded and taped out. The Korean programs, Rebellions and FuriosaAI, sit in this second front. The moat is being attacked from more than one direction.

Good Enough Beats Best

Hyperscalers do not buy silicon the way a gamer buys a graphics card. They buy on total cost of ownership per useful token, at the scale of a data center full of racks, powered by contracted megawatts, cooled by water they had to negotiate rights to.

At that scale, “best chip” is not the metric. “Cheapest way to serve the workload” is the metric. A hyperscaler chip that runs at a quarter of an NVIDIA GPU’s raw performance can still win — if it costs a quarter as much to buy, or uses a quarter as much power, or ships without a two-year lead time and an 80% gross margin markup going to Santa Clara.

The competitive threat is not a chip that beats the GPU on peak performance. It is a chip that is worse on peak performance and still cheaper to run the workload. It has to win the invoice, not the benchmark. And the customers writing those invoices are the same four companies that account for most of NVIDIA’s revenue.

The Software Moat Is Getting Cheaper by the Month

The one place NVIDIA’s bulls plant their flag is CUDA — the ecosystem, the libraries, the twenty years of accumulated software that runs on NVIDIA and nothing else. This is real. It is also shrinking as a moat, and it is shrinking fast. AI coders are collapsing the cost of building software stacks. Compilers, runtimes, kernel libraries, framework integrations — the work that used to require hundreds of specialist engineers over years is now being done by teams a fraction of the size, in a fraction of the time. Ports that used to be quoted in engineer-decades are quoted in engineer-months.

NVIDIA’s CUDA moat was built against expensive human software. That cost is falling. When the cost of the moat falls, the moat falls with it.

The Moat Won’t Last

The LLM workload is matrix multiplication at scale. The architecture the workload wants is a statically scheduled machine, and Google has been shipping one for a decade. Every hyperscaler and every serious startup is now shipping the same. NVIDIA came late, paid $20 billion for a branch of the TPU family tree, and is trying to integrate a discipline it has no history in. Its CUDA moat, meanwhile, is being drained by AI-assisted coding that collapses the cost of building competing software stacks.

The moat is real today. It seems unlikely to last for more than a few years. The math of what happens to NVIDIA’s valuation when it doesn’t is the subject of a follow-on piece.

NVIDIA may have invented the vacuum tube but the transistor is coming.

If that’s all you needed to hear, here it is:

https://biasunlocker.com/

For anyone new here, the short version of what it does. Modern news rarely lies outright. It frames. A loaded label, a missing counterpoint, a quote dropped in to anchor your conclusion before you knew you’d reached one — a hundred small techniques for shaping how you feel about a story you thought you were just reading. Catching that in real time takes training most of us never got.

Bias Unlocker hands that training to the AI you already use. Paste an article — or a transcript from whatever passed for a news show last night — and Claude, ChatGPT, Gemini, Perplexity, whichever you’ve got, walks through the framing techniques at work in it. The analysis shows up in your own chat. Same treatment for the outlets you trust and the ones you don’t. It tells you how a piece is framed, not whether it’s good or true. A method, not a verdict.

Two things I won’t budge on. It collects nothing — no accounts, no telemetry, nothing leaves your device. And it hides nothing — you see exactly what it sends before it sends it. No black box.

So now you’ve got options. Want to use it right now, on whatever device you’re reading this on, without a second thought? Open the browser version and go. Prefer it on your phone or your desktop, ready whenever you are? The apps are there too — iPhone, iPad, Mac, and Android, with Windows on the way. Same tool, same everything. Pick whichever suits you.

It’s free. It always will be.

Have a look:

https://biasunlocker.com/

And if you do put it through its paces, tell me where it gets things wrong. That’s how it gets better.

Wall Street has spent eighteen months working itself into a froth about Apple’s failure to “win in AI.” Apple Intelligence shipped late and thin, and the iPhone is now apparently a dumb terminal for somebody else’s models. I think this is mostly noise. Apple Intelligence will get to good-enough on its own clock, and most iPhone customers will never notice. The AI gap is a quarterly narrative looking for a catalyst. It is not the structural problem. The distinction is time horizon: on a five-year view, AI features may well move the needle more — but the AI gap is self-correcting, while the gaming gap compounds, and on a twenty-five-year view only one of them threatens the franchise.

The structural problem is that Apple has spent twenty-five years handing the PC gaming market to Microsoft. That miss constrains Mac TAM, locks an entire generation into Windows for life, and — the part nobody is discussing — quietly leaves the iPhone franchise undefended on a generational timeline. The questions these facts demand have never been put to Apple’s leadership in a public setting. They should be.

The head start Apple gave away

The clean start date is June 2000, when Microsoft acquired Bungie — the studio behind Marathon and Myth, both Mac-first franchises — to anchor the original Xbox. Halo, which Steve Jobs himself had unveiled onstage at Macworld a year earlier, shipped instead as an Xbox launch exclusive in 2001 and became the console’s killer app. Apple shrugged.

That shrug has compounded for twenty-five years. Microsoft spent roughly eighty billion dollars making gaming a strategic pillar — Activision Blizzard, Bethesda, Mojang, plus Xbox and Game Pass. Sony built PlayStation into a thirty-billion-dollar business with a stable of studios producing The Last of Us, God of War, and Spider-Man — works that win the cultural conversation, not just the sales charts.

Apple’s equivalent output: the Pippin (1996, dead on arrival), Game Center (2010, abandoned), Apple Arcade (2019, a family-friendly subscription explicitly aimed away from serious gamers), the Game Porting Toolkit (2023, a translation layer that handed developers a homework assignment: port your Windows game to our 2% user base at your own expense), and the Apple Games app plus Metal 4 (2025, a social launcher and graphics API update with no studio acquisitions, no gaming dock, and no first-party AAA titles behind them). Each chapter follows the same pattern: reactive, incremental, and carefully positioned to avoid genuine commitment. That is not a strategy. That is the absence of one.

The hardware is already there

This is what makes the miss inexplicable. Apple Silicon is competitive on raw GPU performance and dominant on perf-per-watt. Thunderbolt 5, which Apple ships on its M4 Pro and Max machines, runs at eighty gigabits per second bidirectional — enough bandwidth to run an external GPU at near-internal performance. The components for a category-defining product are already in Apple’s hands.

The product writes itself: a Mac mini-form-factor external gaming box that connects to any MacBook via a single cable. Mobile when you need mobile; drop into the dock at home and the laptop becomes a gaming workstation. Microsoft can’t build this because they don’t own the laptop. Nvidia can’t build it because they don’t own the OS. Only Apple can do the full vertical.

Instead, when Apple transitioned to Apple Silicon in 2020, they killed eGPU support entirely. No driver path for external GPUs. The stated reason was that Apple Silicon was “fast enough” — true for video editing and Xcode, emphatically not true for AAA gaming, where a single Nvidia RTX 4090 still embarrasses an M3 Ultra. Steam surveys put macOS at under 2% of its user base. That is the hardware gap in one number. And the retreat wasn’t limited to hardware: for years Apple’s App Store rules blocked native cloud gaming apps like Xbox Cloud Gaming and GeForce Now — the one path that would have brought the AAA library to Apple’s platforms at zero engineering cost to Apple. That was not a passive miss. That was an active retreat.

The economics of a self-imposed cap

Get the units-versus-dollars distinction right, because it is the whole story. The traditional PC hardware market — desktops, notebooks, and workstations, the IDC definition — runs about 260 million units and roughly $170 billion in revenue a year. Apple ships about 9.5% of the units. But because Macs sell at roughly twice the industry’s blended average selling price — about $1,370 per unit against an industry blend near $650 — that 9.5% of units translates into roughly 19% of the dollars, on Mac revenue of about $34 billion. Apple is fourth in units behind Lenovo, HP, and Dell, and arguably second in dollars. Mac grew 12% last year on the strength of the M-series — which only sharpens the point: the hardware is winning on its merits everywhere except the one segment Apple refuses to contest.

The gaming PC hardware segment — the machines, GPUs, and peripherals, separate from game software — was roughly $65 billion in 2025, growing at 13% annually, several times the broader PC market’s rate. The fastest-growing slice of the PC market is nearly twice the size of Apple’s entire Mac business, and Apple doesn’t compete in it. Yes, Apple collects a 15-30% toll on the world’s largest mobile gaming platform through the App Store — and a handful of prestige iPhone ports like Resident Evil Village and Assassin’s Creed Mirage notwithstanding, that’s not a gaming strategy. That’s a tax. Apple’s other answer, presumably, is that spatial computing via Vision Pro is the next gaming platform — a bet that requires a $3,500 headset to succeed where every previous VR device has failed, backed by a game library that doesn’t exist.

But the gaming PC dollar figure is only the visible part. The larger number is every productivity laptop, every kid’s first computer, every household desktop that gets bought as Windows because the gamer in the family decided. PC purchases lock in four to six years of platform commitment. And the mechanism is measured, not speculative: Pew finds 97% of American teenage boys play video games, and roughly half of teen gamers play on a PC. Explicitly badged gaming laptops alone run about a fifth of laptop shipments and roughly $31 billion in annual revenue — on the order of a quarter of worldwide laptop dollars, before counting the general-purpose PCs specced with gaming in mind. A single fourteen-year-old who wants to play with friends doesn’t just buy one Windows PC. He sets the household platform, which determines what mom replaces her laptop with, what dad gets for his home office, what the younger sibling unboxes at twelve. One gamer locks in three to five PC purchases over a decade, and Apple loses every one.

Now run the counterfactual, and be explicit about the frame. Suppose Apple held 25% of PC units instead of 9.5% — competitive with Lenovo, plausible if they actually tried. At Mac selling prices, 25% of units means roughly 40% of industry dollars: about $89 billion in annual Mac revenue against today’s $34 billion, or roughly $55 billion incremental — call it $40-45 billion if the expanded mix dilutes ASPs toward $1,100 — plus $15-20 billion in Services attach on the enlarged installed base. At Apple’s blended margins that stacks to something like $25-30 billion in incremental operating income, worth on the order of three-quarters of a trillion dollars at Apple’s current earnings multiple — before counting a single iPhone or wearable pulled along by each new Apple household. Count those, and a trillion dollars is the honest round number sitting on the table. And if you reject the upper bound entirely: even if serious gaming merely doubled Apple’s unit share into the high teens, the incremental revenue would still dwarf the AI narrative. The AI gap that analysts obsess about represents maybe $50-100 billion in five-year risk. The gaming gap is several times larger, and nobody is asking about it.

A moral judgment dressed as product strategy

If the financials are obvious and the hardware is sitting there, why hasn’t this changed in twenty-five years? The answer isn’t strategic. It’s cultural.

Jobs grew up in the Whole Earth Catalog / Bay Area counterculture, where “real” creative pursuits were music, film, graphic design, and serious literature. The original Mac was sold on Garamond and orchestral MIDI. Final Cut, Logic, GarageBand — the whole iLife suite — was built around his personal taste. Games were absent from that vision. When Microsoft bought Bungie in 2000, he didn’t think it mattered.

Cook is even further from gaming culture. Nothing in his public persona suggests he has ever engaged with games as a medium. The senior bench — Federighi, Joswiak, Williams, Ternus — same generation, same blind spot. Contrast Microsoft, where Phil Spencer reports near the top and is a visible, credible gamer. Or Sony, where PlayStation has institutional power. Or Nvidia, where Jensen Huang talks openly about games as the workload that built the company.

The clarifying test is the pitch meeting where someone proposes spending forty billion dollars on a top-tier game studio portfolio. At Microsoft that pitch wins. At Apple: “Is this really what we want to be known for?” That sentence is the whole problem. It treats the question as brand dignity rather than market opportunity. The charitable reading is capital-allocation discipline — plenty of non-gaming companies have bought studios and written them down. But discipline would look like a targeted alternative: selective first-party titles, funded ports, a hardware path for gamers. Apple’s alternative was nothing, for twenty-five years, which is how you know the objection was never financial. What Apple’s leadership believes, in the marrow, is that playing games is not a worthy adult activity. You can hear it in every keynote where games appear only as benchmarks — proof the silicon is fast — before the presentation pivots back to the “real” use cases.

This hierarchy is a generation out of date. Elden Ring, Baldur’s Gate 3, The Last of Us are works of comparable craft to the prestige television Apple is spending billions chasing on Apple TV+. The global games market hit roughly $197 billion in 2025 — larger than film and recorded music combined. None of this has penetrated Apple’s executive worldview.

The real cost: an undefended iPhone

Here is the part that should keep Cook up at night — and it requires being precise about what actually holds iPhone share in place.

It is not the Mac. American teenagers choose iPhone at rates near 87% while Mac sits in a small minority of their households; plenty of those teens run a Windows gaming rig and an iPhone side by side. What holds them is iMessage — the blue bubble, the group chat, the social network effect that lives entirely on the phone. Apple’s US iPhone dominance rests, for the majority of its customers, on a single social-network moat rather than on ecosystem depth.

Now look at what the Mac changes. In an Apple household, the iPhone is defended twice over: iMessage plus Handoff, AirDrop, Universal Clipboard, iPhone Mirroring, shared iCloud — integration that makes switching to Android genuinely painful. In a Windows household, the iPhone is defended once. Same phone, radically different switching cost. Mac → iPhone is nearly unswitchable. Windows → iPhone is one good reason away from Samsung.

That single remaining defense is under active siege. The Department of Justice’s 2024 antitrust suit made iMessage a centerpiece — the Attorney General called out the green bubbles by name — and the case survived Apple’s motion to dismiss in 2025 with every claim intact. Apple has already been pushed into adopting RCS, which closed most of the functional gap with Android messaging. The EU, tellingly, declined to regulate iMessage at all: outside America, the moat barely exists, which is one reason iPhone share runs from a third in Western Europe down to single digits in India. The blue bubble is a US-only network effect, and network effects — unlike integration lock-in — can collapse quickly when the friction sustaining them is removed, whether by court order or by a sufficiently good Android value play.

This is where the gaming miss compounds into franchise risk. The teenager whose games don’t run on a Mac defaults into a Windows household at twenty-two — an iPhone customer held by one thread instead of five. His kids grow up on Windows, and their phone choice is a coin flip. None of these transitions is inevitable; the point is that each one compounds the next. Apple’s gaming abandonment is not creating Android users today. It is manufacturing, at generational scale, the precise category of customer — iPhone-by-network-effect, ecosystem-free — that a legal remedy or a competitor can actually take. The market is pricing iPhone dominance as permanent. It is a single moat, currently in litigation, defending a customer base Apple declined to lock in when it had the chance.

The AI miss is a taste-of-the-week story. The gaming miss is a structural threat to the most valuable franchise in consumer technology, on a generational timeline, hiding in plain sight.

These are the questions Apple’s analysts should be asking on every earnings call. They won’t, because they share the blind spot. They too think video games are something kids do before they discover the things that really matter.

Profit is an accounting value. A company can report a $1 million profit and have nothing to return to shareholders, because staying in business next year requires spending that million on a new machine to replace the one wearing out. Technically it was profit. In reality it was money required to pay a future debt that does not show up in current accounting terms. Free cash is what remains after those obligations are met.

For readers unfamiliar with the concept, the most visible portion of free cash is the dividend the company pays, so for purposes of this article you can think of free cash flow as dividends, though in reality it can be more involved. I treat the concept at length in The Most Misunderstood Phrases in Buffett-Speak: Free Cash and the Float.

The Buffett Framework

When Warren Buffett buys a stock, he is buying a business — and the free cash that business will produce for as long as it exists. He does not plan to sell. He does not particularly care what the public market thinks the business is worth next year, next decade, or ever. He cares about one number: the cumulative free cash flow the business will deliver from the moment of purchase until it ceases to operate, discounted back to today. If that number exceeds the price by enough to make this a good investment return, he buys. If it doesn’t, he doesn’t.

He paid $25 million for See’s Candies in 1972. Over the subsequent five decades, See’s has handed Berkshire roughly $2 billion in free cash — eighty times the purchase price. He has never sold a share and does not intend to. He paid $1.3 billion for the Coca-Cola stake in 1988. Dividends received since: well over $10 billion. He has never trimmed the position. The return on either purchase has nothing to do with what someone would pay for those businesses today. It has to do with the cumulative free cash both have produced while Berkshire held them.

Applied to Micron

Apply this framework to Micron Technology, currently trading at $1,220 per share for a market capitalization of approximately $1.38 trillion.

If you bought all of Micron today for $1.38 trillion, and the stock market closed tomorrow and never reopened, what would you collect?

Micron’s all-time best year for free cash flow was fiscal 2018, the peak of the last memory boom: $8.5 billion. Assume the company repeats its best year ever, every year, forever, never declining. The undiscounted cumulative free cash flow takes 162 years to equal $1.38 trillion. Discount the perpetual stream at 10%, which is a reasonable cost of equity for a cyclical commodity producer, and the present value of forever-peak free cash flow is $85 billion — about 6.2% of the current market cap.

The through-cycle reality is harsher. Micron’s average annual free cash flow over the nine years from fiscal 2017 through fiscal 2025, capturing two full memory cycles, is $1.94 billion. Discount that perpetual stream at 10% and you get $19.4 billion of intrinsic value, or roughly $17 per share. The current price exceeds it by a factor of over seventy.

The arithmetic does not change much if you assume a generous future. Pick the bull case: HBM permanently elevates the floor, the three-supplier oligopoly cements, owner’s earnings stabilize at $7 billion annually and grow at 5% forever. The Gordon Growth model produces $140 billion of equity value, or $124 per share. That requires every assumption to break in Micron’s favor, simultaneously, indefinitely. It still leaves the stock roughly 90% overvalued.

Two Claims, Not Three

What this means is that the buyer of Micron at $1,220 is making one of two claims. Only two.

Claim one: Micron’s free cash flow will eventually reach $60 to $80 billion annually, sustained perpetually, growing from there. This requires the company to generate roughly eight times its all-time best year, every year, for the rest of time. It requires HBM gross margins to hold at levels memory has never sustained. It requires demand to grow without cyclical resets. It requires the oligopoly to never crack, no Chinese entrant to take volume share, and no architectural disruption to displace high-bandwidth memory. Anyone making this claim should be required to defend it explicitly. They almost never are.

Claim two: Micron’s free cash flows will never come close to justifying $1.38 trillion, but someone else will pay more than $1.38 trillion for the company before the current owner exits. This is the greater-fool trade. It is a legitimate trade. It is not an investment in the Buffett sense. It depends entirely on the future behavior of other market participants and has nothing to do with the underlying business.

There is no third claim. Every buyer at the current price is making one of these two bets, and they are not the same bet. The first requires belief in a future free cash flow trajectory that defies the entire empirical history of the memory industry. The second requires belief that narrative momentum will continue long enough for the buyer to exit. The first is rare. The second is common, and it is what most “investing” in late-stage bull markets actually consists of.

The Buffett framework does something other valuation methods don’t: it forces this distinction into the open. The fund manager who tells his clients he is “investing in the AI infrastructure thesis” by buying Micron at $1,220 is, under honest examination, doing the second thing while calling it the first. The retail trader watching the stock rip from $100 to $1,200 in less than a year and chasing it is doing the second thing without pretense, which is the more honest position. Both are making bets on future buyers, not on future free cash flows.

Conclusion

The intrinsic value of the stock, based on discounted free cash flow, is $17 to $124 per share. It trades at $1,220 — based on trying to guess what someone else will pay you for it later.

There is a simple test for whether a proposed law is about the problem it claims to solve: will its supporters accept any implementation timeline that doesn’t coincidentally (wink, wink) maximize political advantage in the next election? The SAVE America Act fails that test in a way that should make even sincere voter-ID supporters uncomfortable.

The bill, which the House passed in February and which is now stalled in the Senate, would require Americans to present documentary proof of citizenship to register to vote in federal elections, plus photo ID at the polls. These changes would take effect the moment the bill is signed. Not after the midterms. Not after a multi-year transition that would prevent mass disenfranchisement. It would take effect the moment the ink dries.

Even Republican senators have publicly broken with this timeline. Thom Tillis, a long-standing voter-ID supporter, called it “a waste of time” and “a distraction.” John Cornyn called the demand to eliminate the filibuster to pass it a “fantasy.” Mike Rounds, a co-sponsor of the underlying bill, conceded “the numbers aren’t there.”

The Problem That Isn’t

Federal law already requires that voters in national elections be U.S. citizens, has done so for decades, and is already enforced. The question is whether noncitizen voting occurs at a scale that justifies the most disruptive overhaul of voter registration in living memory. The answer, on every available piece of evidence, is no. Utah recently audited more than two million voters and found 27 noncitizens on the rolls and 13 noncitizen ballots cast since 2018. The rate is effectively zero. The Heritage Foundation’s own voter-fraud database, hardly an organ of the left, confirms the same pattern.

It is clear that the bill’s supporters believe disenfranchising voters with this bill — not because those voters are noncitizens, but because they are likely to vote for the other side — will help them win.

The Burden, and Whom It Falls On

Roughly 21 million American citizens lack ready access to a passport or birth certificate, numbers that skew Democratic. But the bill also lands hard on voters who go Republican: tens of millions of married women whose birth certificates don’t match their current legal name; rural Western voters facing round trips of several hours to an election office; some Alaska and Hawaii voters who would need to fly to register. These are not Brooklyn precincts.

President Trump himself, in a moment of unguarded honesty, said that if the SAVE Act became law, Republicans would “never lose a race. For 50 years, we won’t lose a race.” This reads less like voter integrity and more like how you put your thumb on the scale to win elections.

The Administrative Reality

Even setting aside who gets disenfranchised, there is the question of whether the bill can be implemented at all. Orange County, California estimated it would need 59 additional staff and roughly $6 million a year just to comply. Nationwide, you’re looking at a billion-dollar unfunded mandate dropped on local election officials in a midterm year. The bill provides no funding. It does, however, attach criminal penalties — including potential imprisonment — to election workers who register applicants without proper documentation. The rational response from a clerk facing prison time is to refuse registration in any ambiguous case. Primaries are already underway. Military and overseas ballots have been mailed. This will just be disruptive and undermine election confidence by asserting a clearly false emergency.

The Doctrine Has a Name

When the Supreme Court invoked the Purcell principle in 2006, it was constraining lower federal courts: don’t change election rules close to an election, because the disruption itself harms voters and administrators. What Purcell constrains for courts, prudence ought to constrain for Congress. Major election laws — HAVA in 2002, the NVRA in 1993, the Voting Rights Act in 1965 — all had phased implementation, because election systems are infrastructure, and you don’t rip out infrastructure two months before you need it. The SAVE Act’s “effective immediately” provision is therefore not a technicality. It is a tell.

They’ve Already Tried This. They’ve Already Lost.

After the 2020 election, the Trump campaign and allied groups filed more than sixty lawsuits alleging widespread fraud. Essentially every one was dismissed or lost on the merits, including by judges Trump himself had appointed. The Department of Justice under Trump’s own Attorney General found no evidence of fraud at a scale that could have affected any outcome. State-level audits and recounts in Arizona, Georgia, Michigan, Pennsylvania, and Wisconsin reached the same conclusion — the Arizona “Cyber Ninjas” review, run by election skeptics, actually found Biden’s margin had been larger than originally reported.

Since then, the legal losses have continued. Fox News paid $787.5 million to settle defamation claims for spreading fraud allegations it knew to be false. Rudy Giuliani was hit with a $148 million judgment over false claims about Georgia election workers. Sidney Powell, Mike Lindell, Newsmax, and OAN have all faced or settled defamation suits over the same allegations. The legal record on whether American elections are riddled with fraud is one of the most thoroughly developed factual records in modern American history, and it has reached a single conclusion: they are not.

The SAVE America Act is downstream of a claim that has been adjudicated to death. The people pushing it know this. They are not relitigating the question because they think they will win on the facts. They are relitigating it because losing on the facts has not stopped them.

The Steel Man

A serious approach to voter identification wouldn’t look anything like this. Last summer, in A Proof of Self: Why Every American Deserves a Portable, Universal Identity, I argued for a portable, secure digital identity — voluntary and free, on the model of Estonia’s national ID or India’s Aadhaar — that would give conservatives real voter ID without disenfranchisement and finally recognize the millions of Americans currently locked out of jobs, banking, and the polls. The SAVE America Act does the opposite. It makes proving who you are harder for tens of millions of citizens who already are who they say they are.

Even setting that aside, a January 2027 effective date would soften most objections. States would have time to staff up. Voters would have time to obtain documents. The courts could resolve the constitutional challenges before chaos hits. The fact that the bill’s supporters won’t accept a delayed effective date tells you what they actually want.

The Cleanest Argument

You don’t have to oppose voter ID to oppose the SAVE America Act. You only have to believe that rules shouldn’t change mid-game.

A bill that solved a real problem on a workable timeline would pass with bipartisan support, as HAVA did with 92 votes in the Senate. A bill that solves a phantom problem on an impossible timeline does neither. The unwillingness of its supporters to accept the simplest fix in the world — a different effective date — is the most honest thing about the whole exercise.

The American public is not stupid. Transparently mean policies — policies visibly designed to disenfranchise rather than to govern, that solve a phantom problem and break millions of citizens in the process — do not play in the middle. They energize the far ends of both parties and exhaust everyone else.

The tell is the timing. Everything else is theater.

Trying to pass off disingenuous behavior as principled reform is not a winning formula.

Reading the earnings release, you would never guess that 75% of Americans don’t trust the news, that the Reuters Institute just measured US trust in news at 25% — its lowest reading in eleven years of tracking — and that the establishment press is in the middle of the longest, deepest credibility collapse in its history.

How does a newspaper post record results during a record collapse of trust in newspapers?

The answer is that the New York Times Company is no longer primarily a newspaper. It is a lifestyle subscription bundle that includes a newspaper as one of its products. And the newspaper is not the part that’s growing.

What They Actually Sell

The Times reports its subscription numbers as one big aggregate, but the components are listed plainly in their own filings: news, The Athletic, Cooking, Games, Audio, and Wirecutter. Five of those six products are not journalism in any traditional sense. The Athletic is a sports site. Cooking is a recipe database. Games is Wordle, Connections, Spelling Bee, and the crossword. Audio is podcasts. Wirecutter is product reviews — essentially a comparison-shopping affiliate site.

The company’s own commentary on what’s driving growth tells the story without subtlety. CEO Meredith Kopit Levien describes the business as playing in “five spaces that marketers want to be in” — news, sports, games, cooking, and Wirecutter. Four of the five are not journalism. Their bundle pitch — the All Access subscription that most new subscribers now take — works because the non-news products are individually compelling. Wordle alone has the kind of daily-habit hook that the news product hasn’t had in decades. People open the Times app every morning to do the Spelling Bee, glance at a headline on the way out, and close the app.

The Times itself markets this. The bundle is “the essential subscription for curious people.” Curiosity, here, means puzzles, recipes, sports analytics, sleep tracker reviews, and twenty minutes of podcasts on the commute. The news happens to be in the same app.

Sold standalone at the bundle price, the news product would lose subscribers in months. The bundle carries it. As a newspaper, the Times sits in the basement of public trust along with the rest of the industry — the public has stopped buying it as news. Management cannot say this out loud and continue to claim what the institution claims about itself, so they don’t say it. But the financial statements say it for them.

Buffett Weighs In, Quietly

The most telling vote on what the Times actually is came from the one investor whose opinion on newspapers should carry more weight than anyone else’s. Warren Buffett spent four decades as the largest outside shareholder in the Washington Post Company. He understood the newspaper business cold, made a fortune on it, and watched it die in slow motion. The Graham family sold the actual paper to Jeff Bezos in 2013. In 2020, Buffett unloaded every remaining newspaper Berkshire Hathaway owned — dozens of regional papers acquired over the years — and pronounced the industry “toast.” He was done. He had no intention of ever owning a newspaper again, and he said so plainly.

Then in Q4 2025 — Buffett’s last quarter as CEO before stepping down — Berkshire disclosed a brand new $352 million position in the New York Times. Just over 5 million shares. The financial press treated it as a surprising reversal, a “vote of confidence” in journalism, a sign that maybe the news business wasn’t dead after all.

Then in Q1 2026, Berkshire tripled the position to 15.1 million shares, worth more than $1 billion. That’s not a vote of confidence. That’s a thesis.

Buffett did not get back into the newspaper business. He got into a games-and-lifestyle bundle that happens to ship with a newspaper. The man knows exactly what he is buying, and it is not what the Times claims to be selling. He’s buying recurring revenue from puzzle subscribers, recipe-database engagement, sports content with the highest paid retention rates in digital media, and an affiliate-commerce site sitting on top of one of the strongest consumer brands in America. The journalism is, for the investment thesis, an overhead line item.

Even the bullish Wall Street coverage of the Berkshire trade quietly acknowledges this. The pitch decks talk about “the moat” running through bundling. The CEO’s own quoted language in Q1 was “uncompromised journalism and premium lifestyle content.” The “and” is the entire story. Buffett doesn’t pay billion-dollar prices for prestige assets. He pays for predictable cash flow. The cash flow is in the puzzles, recipes, sports, and Wirecutter affiliate fees. He read the same 10-Q everyone else read and acted accordingly.

When the most disciplined capital allocator of the last fifty years looks at the New York Times and decides it is now a buy — after publicly declaring newspapers dead five years earlier — the conclusion is not that newspapers came back. It is that this isn’t a newspaper.

Is the Times “fake news”?

There’s a separate question worth addressing on its own. Does the Times deserve the “fake news” label that Donald Trump pinned on it and has been swinging like a club for a decade?

What does “fake news” even mean? The label is imprecise. The accusation behind it isn't.

When Trump arrived on the political scene with a loose-at-best relationship to verifiable fact, the press responded by building out a fact-checking apparatus. PolitiFact, the Washington Post Fact Checker, the Times’ own fact-check desk, dedicated annotated transcripts of speeches and debates. That was a legitimate response to a real problem, and it largely worked on its own terms. When a politician says a thing that isn’t true, somebody now says so on the record, in real time, with citations. That infrastructure didn’t exist twenty years ago and it should exist.

But fact-checking turned out to be necessary and nowhere near sufficient. A story can pass every fact-check on every individual sentence and still completely mislead the reader through what gets put in the headline, what gets buried in paragraph fourteen, which sources get quoted as experts and which get quoted as cranks, what context gets supplied and what context gets withheld, which adjectives attach to which actors, and which story angles get pursued and which get quietly dropped. And then there’s the ubiquitous use of anonymous sources — assertions no reader can ever fact-check because no one is told who said them.

All of it is editorial choice. And editorial choice, repeated story after story across a national platform, is how a worldview gets manufactured without anyone ever writing anything technically false.

Until recently there was no practical way to check this at scale. Identifying framing techniques required reading carefully, knowing what to look for, and having the time and training to do it across enough articles to see the pattern. Fact-checking scaled. Bias-checking didn’t.

AI changed that. I built a free app called Bias Unlocker that applies a structured framework to news articles and surfaces the specific techniques in use: selective omission, source asymmetry, loaded framing, headline mismatch, narrative laundering, false balance, and the rest of the catalog. The app grew out of an earlier op-ed of mine, The Modern Journalist’s Field Manual, which laid out the taxonomy.

There is no point now in arguing about whether articles are biased. You can feed it a Times story. It tells you, technique by technique, how the story is shaped. Then you decide for yourself.

A human doing this kind of work would themselves be open to charges of bias, and rightly so. The app isn’t — it’s run by AI with no political stake in the outcome, and the prompts driving the analysis are visible to the user. The check itself is checkable. Complete transparency.

The newspapers and mainstream media have been harping on fact checking.

Now the shoe is on the other foot.

The Question

When subscribers buy the bundle mostly for games, recipes, sports, and shopping advice — when 75% of Americans don’t trust the news product the bundle includes — when the institution itself reports that its growth is driven by non-news products and the news is the part that needs the bundle to survive — at what point do we stop calling this a newspaper and start calling it what it actually is?

A lifestyle subscription company that includes a politically aligned commentary product as a courtesy.

The Times will keep posting good quarters. The bundle will keep growing. And the institution will keep claiming the moral authority of a profession it has quietly stopped practicing.

Caveat lector.

Every few years, Washington discovers the Middle East is complicated.

However, the Middle East is not so complicated if you look at it the right way. The confusion arises from not understanding its history and the through lines, some of which are more than a thousand years old.

In respect to Iran’s relationship with the rest of the Middle East, there are two schisms.

Schism One: Persian vs. Arab

This one is ethnic and civilizational. It has nothing to do with religion.

Persia was a great empire before Islam existed. Cyrus the Great. Darius. The Achaemenid dynasty. A civilization that stretched from the Aegean to the Indus, that wrote law, built roads, and governed a hundred peoples — all of it centuries before Muhammad was born.

In 651 CE, Arab Muslim armies destroyed the Sassanid Empire, the last Persian dynasty, and overran the country. Persia was Islamized. It was never Arabized. The Persians kept their language. They kept their culture. They kept their identity. Every other nation the Arabs conquered — Syria, Egypt, North Africa — eventually became Arab in language and self-conception. Persia did not. It is the only one.

That resistance was not accidental. It was the expression of a civilization that remembered being great before the conquerors arrived. A civilization that absorbed the new religion on its own terms and built something the Arabs couldn’t dissolve. In the 8th and 9th centuries, Persian scholars, poets, and bureaucrats mounted a deliberate cultural resistance inside the Islamic Caliphate itself — staffing its administrative machinery while preserving their own language and literature. Ferdowsi’s Shahnameh — the Book of Kings — kept the memory of Cyrus and Darius alive for centuries. The Arabs brought the religion. The Persians kept the empire’s memory.

The ethnic rivalry — Persian pride and imperial ambition versus the Arab world — predates Islam. It existed before the conquest. It survived the conquest. It runs underneath every conflict in the region today.

This is the axis where “Iran wants to dominate the Middle East” lives. Permanent loss of regional dominance is not a settled issue for Iran.

Schism Two: Sunni vs. Shia

This one is religious, internal to Islam, and has nothing to do with ethnicity.

Muhammad died in 632 CE without clearly designating a successor. The community split immediately over who should lead. Most backed Abu Bakr, Muhammad’s close companion, selected by community consensus. A minority held that leadership belonged by right to Ali — Muhammad’s cousin and son-in-law — and to the line of descendants from Muhammad himself.

That minority became the Shia. The majority became the Sunni.

The split hardened into permanent fracture at Karbala in 680 CE, when the Sunni caliph’s forces killed Ali’s son Husayn. The massacre became the founding wound of Shia identity. It is commemorated annually. It is not forgotten.

Iran is Shia. Most of the Arab world is Sunni — but not all. Iraq is majority Shia Arab. Bahrain is majority Shia, ruled by a Sunni monarchy. Lebanon’s Hezbollah is Shia Arab.

Where They Fuse

In 1501, the rulers of Persia made Shia Islam the official state religion.

The move was strategic. Persia was squeezed between Sunni powers on both flanks — the Ottoman Empire to the west, Uzbek powers to the east. Adopting Shiism was a declaration of difference. It gave Persia a distinct religious identity that set it apart from every Sunni neighbor threatening its borders.

It also gave Persia something more durable: a civilizational claim. The rulers of Persia became simultaneously the heirs to ancient Persian imperial ambition and the champions and protectors of Shia Islam worldwide.

From that moment, the two schisms became one. They braided together inside the Persian state and have not separated since.

Two Engines, One Vehicle

Modern Iran is the product of both fault lines running at once.

The “Shia crescent” — Iran’s network of proxies and aligned forces across Iraq, Lebanon, Syria, and Yemen — runs on the religious axis. Shia solidarity. The defense of the faith against Sunni powers.

The drive for regional hegemony — the insistence on Iranian primacy, the resistance to any outside power dictating terms — runs on the ethnic-imperial axis. Persian memory. The empire that was. The regime calls it revolution. The map calls it empire.

Saudi Arabia fears Iran on both tracks simultaneously. It is Sunni facing a Shia power. It is Arab facing a Persian one. Two overlapping anxieties, neither of which disappears when you address the other.

Shia Arabs throughout the region feel the same tension in reverse. The religious axis pulls them toward Iran. The ethnic axis reminds them that Tehran is Persian — and Persian hegemony has historically come at Arab expense.

The conflict is overdetermined. Remove one engine and the other keeps running.

The Little Satan/Great Satan Con

The Iranian theocracy has a stated goal: destroy the Little Satan — Israel — and the Great Satan — the United States. It says so openly. It has said so for decades.

Cons employ misdirection. This stated goal is pure misdirection.

When Khomeini took power in 1979, Iran was isolated — a Persian Shia state in a region overwhelmingly Arab and Sunni. The regime needed a cause no Sunni leader could openly oppose. They found it in Palestine. Khomeini declared an annual Jerusalem Day that first year, and in the decades that followed Tehran built Hezbollah and backed Hamas — pivoting from Persian outsider to pan-Islamic champion. Sunni Arab leaders quietly accommodating Israel were suddenly the traitors. Tehran was suddenly the resistance. The con was running.

The real agenda is not the destruction of Israel or America. The real agenda is to dominate the Middle East and dominate Islam — on Iranian terms, with Iran’s brand of it. The two satans are a useful story. They unite Sunni and Shia against a common enemy, paper over the Persian/Arab rivalry, and dress up regional imperialism as righteous religious resistance.

Keeping Israel as the enemy of focus helps mask the con. Harassing Israel forever serves a greater purpose at this time than destroying it. Israel is roughly the size of New Jersey. It is not logical for a country to spend fifty years and all its money trying to erase a tiny speck on the Middle Eastern map. It is, however, worth every bit of that trouble if Israel is the vehicle for achieving something far larger.

The United States is the other half of the equation. It is the outside power whose presence in the Middle East directly obstructs Iranian dominance. Label it the Great Satan and you have a second villain that unites the faithful.

Strip away the billboard and you find the two schisms running underneath. That’s the real story. That’s what this piece is about.

The Arab world leadership itself uses the Israeli-Palestinian conflict for other purposes, which I have outlined in part earlier in op-eds on the conflict.

What Diplomacy Cannot Buy

Now look at what the US is demanding Iran give up.

No nuclear weapons. That kills the hard power instrument of Persian imperial ambition. Without it, Persia is a medium-sized country with oil and a long memory. With it, Persia is the regional hegemon its civilizational identity demands it be.

No proxy network. That kills the Shia crescent. Dismantle Hezbollah, the Iraqi militias, the Houthis, and the Shia religious axis collapses. Iran is no longer the champion and protector of Shia Islam across the Arab world. It’s just another country.

Washington is asking Iran to surrender both engines simultaneously. From Tehran’s perspective that isn’t a negotiation. It’s a civilizational extinction event. Not the regime — the entire 2,500-year project the regime exists to advance.

This is the real story unmasked.

Iran’s nuclear bomb is the ultimate declaration that “Persia is back.” They are not going to give that up — or any of the rest of their goal of Middle Eastern dominance.

The world waited for communist Russia to become a responsible global player, and when given the chance after the Soviet collapse, within twenty years it was back to the same place — nationalism, authoritarianism, and military aggression. Change in some parts of the world moves very slowly. Russia and the Middle East both appear to be in that pattern. Nobody making serious long-term infrastructure decisions can afford to bet otherwise.

What the War Did

Before the conflict, the Strait of Hormuz carried 15 million barrels per day of crude exports and 20% of global LNG. When Iran closed it, the Middle East lost more than 12 million barrels of oil equivalent per day. Iraq’s output fell 70%. Export losses averaged nearly $2 billion per day — for four months. North of $240 billion in lost revenue, before counting infrastructure damage and defense costs.

The IEA called it the largest supply disruption in recorded history. The world got a four-month preview of energy without the Middle East. Nobody liked what they saw. Nobody is going to forget it.

Three Forces, One Direction

They need to make up for what they lost. Every Gulf state is staring at a fiscal hole with one solution: pump every barrel possible at whatever price the market will bear. Saudi Arabia’s budget breakeven is north of $90 — they’re selling at $79. The UAE spent $5 billion intercepting missiles. Iraq lost 70% of exports for four months. This is financial survival, not strategy. The reservoir physics enforces it too — wells abandoned mid-operation don’t restart cleanly. Some capacity is permanently gone. The remaining wells must pump harder to compensate.

OPEC is broken. The UAE formally withdrew from OPEC and OPEC+ in May — ending 59 years of membership. Iraq will pump everything its damaged infrastructure can handle. Russia never complied with cuts anyway. There is no mechanism left to coordinate a cut that would actually hold.

The world is turning away. India doubled its oil supplier nations from 20 to 40 in weeks. The Americas quintet — the U.S. at 13.7 million barrels per day, Canada, Brazil, Guyana, and Argentina — accelerated during the war, not slowed. Nuclear is back on the table across Asia and Europe. The last time this happened was 1973, when the Arab embargo triggered the IEA, U.S. strategic reserves, and France going 75% nuclear within fifteen years. The Gulf states are reopening Hormuz into a market that spent four months replacing them — and that is now looking forward to permanently replacing them as a major source.

The Suppliers Know Exactly What’s Happening

The Middle Eastern producers see the picture clearly. The window to sell is open now and nobody can guarantee how long it stays open. For the next sixty days every Gulf producer will pump at maximum capacity — not because a committee decided to, but because there is no rational alternative. The fiscal damage demands it. The broken cartel permits it. And the awareness that their customer base is actively building away makes every barrel left in the ground feel like one that may never find a buyer.

What the financial models miss is the human reality underneath the numbers. These are not calm boardrooms making strategic production decisions. These are governments simultaneously facing empty revenue accounts after four months of near-zero exports, permanent defense budgets that just doubled overnight, sovereign wealth funds raided to cover wartime deficits, and the slow-motion realization that their customer base is structurally leaving. Vision 2030 — the entire project of building a post-oil economy before the oil money runs out — just got set back years. The patronage networks, the subsidies, the social contracts that keep these regimes stable all run on oil revenue. Below $90, Saudi Arabia cannot fund all of them at once. That is not a government that carefully manages supply to defend a price. That is a government under existential pressure that needs cash today, tomorrow, and every day after that. Desperation is a more powerful market force than any cartel mechanism, and every analyst pricing in Saudi discipline is missing it.

Nobody in Riyadh, Abu Dhabi, or Kuwait City is making plans based on the Iran situation improving in the next several years. The rational posture is to treat every period of open straits as a selling opportunity, not a return to normalcy.

The result is an uncontrolled glut. A race — every producer maximizing volume simultaneously, into a market already building alternatives, on top of a pre-war surplus the war temporarily masked. There is no mechanism left to stop it.

The Forecasts Are Wrong

The deal was signed June 19. Hormuz is still physically largely closed — tanker traffic 95% below pre-war, 800 vessels trapped, mines in the shipping lanes that will take weeks to clear. This will be a slow-motion car crash, not a sudden tidal wave. But the market isn’t waiting. Brent dropped below $80 on the announcement alone. When the oil physically starts flowing again, prices won’t recover. They will keep falling.

The restocking wave provides a temporary cushion — reserves were drained to the bone and the first returning supply gets absorbed by governments rebuilding buffers. That could hold prices near $75 through Q3. But there is a deeper irony nobody is discussing: every country filling strategic reserves is buying future oil consumption today at a discount, and then needing less Gulf oil tomorrow because the tanks are full and alternatives are being built simultaneously. The suppliers are selling forward at fire-sale prices into demand that will not repeat. The short-term revenue relief accelerates the long-term demand destruction.

Goldman’s $75 Brent forecast for 2027 will move lower. Every number on the street assumes a return to something like February 27. It won’t happen. The real floor is wherever new drilling stops being profitable — roughly $50-55 WTI for shale. Middle Eastern production costs run just $3-5 per barrel, so they can pump profitably at almost any price. But their governments cannot function below $90.

That gap is the engine of the crash. Normally, when prices fall too low, producers cut output to push them back up. The Gulf states can’t. Their wells stay profitable even at $40, so there is no reason to stop pumping — and their governments are so starved for revenue below $90 that they have to pump even more to make up the difference. Falling prices trigger more production, not less. The mechanism that normally puts a floor under oil now runs in reverse. And there is no OPEC left to break the cycle.

The Beginning of the End

The world is simply tired of the Middle East. Not just Iran — though Iran is the accelerant — but the entire half-century of dependency, manipulation, and geopolitical extortion. Saudi Arabia and OPEC spent fifty years treating oil as a weapon, engineering scarcity, extracting a premium the market would never have assigned on pure economics. The rest of the world went along because it had no choice. Now it does. The customers are walking out the door with the urgency of people who just got a very clear warning about what staying costs.

The discussion has turned into a comparison of the Obama JCPOA agreement and the Trump agreement. Looking at the full arc of the war shows how that is just shallow political posturing and ignores the full story. The implication is that the Obama deal was great and Trump is a fool who made a mess where there was none.

Set all of it aside. Read the document without the fake news framing.

The way to see what the MOU actually does is to walk the situation forward through twelve moments of the war to date and ask, at each one, where each side stands.

I think you will see that the war is being won gradually without US boots on the ground and with minimum collateral damage to the world’s economy, neighboring states, Iran itself, and commercial ships already trapped there. This runs counter to the chaotic narrative that we acted like a bull in a china shop and then handed Iran a win and gave them more than they had when we started.

As I stated in Why lift the blockade?, dealing with Iran was never going to be a clean simple operation. There is a reason why the US and the rest of the world has kicked the can down the road for 50 years as Iran became ever more dangerous and entrenched in the Middle East.

The war really began after the October 7 Hamas attack on Israel.

Phase One — Before October 7, 2023

Iran was a regional power with hostage doctrine fully loaded. The Ring of Fire was complete. Hezbollah at peak capability with an estimated 150,000 rockets and missiles aimed at Israel. Hamas armed and operational in Gaza with thousands of fighters and a tunnel network. Houthis active in Yemen with Iranian missiles and drones reaching Saudi Arabia and threatening Red Sea shipping. Syria under Assad as the land bridge connecting Iran to Lebanon, with Iranian forces and IRGC infrastructure throughout. Iraqi militias on retainer. Iran’s own missile and drone magazines built up over two decades — thousands of units. Air defenses dense and integrated. Nuclear program advancing — enrichment at high levels, stockpile growing, threshold capacity in reach. Oil exports running near pre-sanctions volumes through the shadow fleet, year after year. The revenue compounded.

Phase Two — October 7, 2023 and the Proxy War

Hamas attacked Israel on October 7, 2023. 1,200 killed, 250 taken hostage. The next day Hezbollah opened the northern front in solidarity. Houthis began launching at Israel from Yemen and attacking shipping in the Red Sea. Iraqi militias attacked US forces. Iran activated the entire forward layer simultaneously.

What followed was fourteen months of Israel grinding the proxies down at enormous cost. The international community demanded ceasefires, accused Israel of disproportionality, threatened sanctions, voted against Israel at the UN almost continuously. Israel kept fighting. Hamas’s military structure was dismantled in Gaza despite the urban combat being the hardest on earth. The Israeli northern population — 70,000 displaced — couldn’t go home while Hezbollah was active, and Israel made clear it would not accept a return to the pre-October status quo.

Phase Three — The Hezbollah Operation

September 2024. Pagers detonated across Lebanon. Walkie-talkies the next day. Hezbollah’s mid-level command structure decimated in 48 hours by an operation that had been seeded years earlier. The intelligence achievement was extraordinary. The strategic effect was greater. Hezbollah’s command-and-control was shattered before the kinetic phase even began.

September 27, 2024. Israel killed Nasrallah and most of Hezbollah’s senior leadership in successive strikes. Ground operation into southern Lebanon. The missile and drone capabilities Hezbollah had built over two decades — the load-bearing piece of Iran’s deterrent against any strike on Iran itself — heavily degraded. Tunnels destroyed. Storage sites hit. Launchers eliminated.

November 2024. Hezbollah ceasefire. Disarmament terms that Hezbollah would not fully comply with, but the damage was done. Iran’s premier forward force was hollowed out. The threat that had deterred Israeli action against Iran for forty years was no longer credible at the scale required.

Phase Four — The Direct Exchanges and Israel’s Strikes on Iran

Twice in 2024, Iran tried direct strikes on Israel. April: Operation True Promise I, 300+ missiles and drones. October: True Promise II, 200 ballistic missiles. Both were largely defeated by Israel’s layered air defenses with US, UK, French, and Jordanian assistance.

Israel hit back twice and that’s the part that mattered. On April 19, 2024, Israel struck an S-300 radar site near Isfahan defending the Natanz nuclear facility — a calibrated demonstration that Iran’s air defenses could be reached. On October 26, 2024, Operation Days of Repentance — a much larger campaign. Israeli strikes destroyed S-300 batteries across Iran, hit solid-fuel missile production facilities, drone production sites, and military bases. Iran’s air defenses and its ability to produce more long-range missiles were significantly degraded before the 12-Day War even began.

The strategic significance was clear in both directions. Iran’s last-resort deterrent — that it could rain destruction on Israel from its own territory — had been demonstrated to be inadequate. The defenses worked. The leakers caused minimal damage. And Israel had now demonstrated it could fly to Iran and hit what it wanted. The October 2024 strikes are why the 12-Day War in June 2025 went so well — Iran’s defenses were already weakened.

By June 2025, Iran was already structurally weaker than at any point in the regime’s history. Proxies mauled. Land bridge gone. Direct strikes defeated. Air defenses ripped open. Missile production capacity reduced. Yet the nuclear program was still advancing toward weapons-grade material, and the regime still refused to settle.

Phase Five — Syria Falls

December 2024. Assad collapsed. Rebel forces took Damascus in days. The Iranian and Russian backers couldn’t save him.

For Iran this was catastrophic. Syria was the land bridge from Iran to Lebanon. Without Syria, resupplying Hezbollah became enormously harder. The forty years of Iranian investment in Syrian infrastructure, IRGC bases, weapons depots, and logistics corridors evaporated. The Axis of Resistance was physically severed. The strategic depth Iran had spent two generations building was gone in a week.

This wasn’t directly Israeli action, but it was downstream of the Israeli campaign. Hezbollah had been Iran’s main expeditionary force in Syria for the entire civil war. With Hezbollah crippled, Iran no longer had the forward troops to send. Iran’s own air defenses had been ripped open by Israel’s October 2024 strikes. Iran’s missile production was degraded. Russia was tied down in Ukraine. When HTS moved, neither Iran nor Russia could respond at the scale needed. The proxy war’s secondary effects took down a regime that had survived fifteen years of civil war.

Phase Six — The Houthis

Early 2025. Trump took office and moved on the Houthis. Sustained US strikes (Operation Rough Rider) degraded Houthi capability — missile launchers, drone production, command nodes. May 6, 2025: Oman-brokered ceasefire ended US-Houthi attacks. The Houthis stopped targeting US vessels. They continued attacking Israel and would resume Red Sea attacks on other shipping later that summer, but the US had pulled itself out of the Red Sea fight on its own terms with Houthi capability significantly reduced.

The last functioning forward layer on Iran’s southern flank was no longer threatening US shipping. Hamas degraded. Hezbollah hollowed. Syria gone. Houthis pressed into a narrow US-only ceasefire. The Ring of Fire was now scattered embers.

Phase Seven — The 12-Day War

June 13 to 24, 2025. Israel attacked Iran. Israeli strikes hit nuclear facilities, killed nuclear scientists, killed senior military leadership, destroyed 70 air defense batteries, and eliminated more than half of Iran’s missile launchers in twelve days. The US’s role was narrow and limited to the nuclear program — on June 22, B-2 bombers and a submarine strike hit Natanz, Fordow, and Isfahan with bunker busters to destroy the underground enrichment infrastructure Israel couldn’t reach with its own arsenal. Iran retaliated with hundreds of ballistic missiles, hit Al Udeid in Qatar, ran symbolic strikes against Israeli cities. Ceasefire on June 24.

Air defenses thought impenetrable were ripped open in hours. Missile launchers thought hidden were found and destroyed. The regime’s senior security and scientific leadership was reachable. Forty-five years of carefully constructed deterrence collapsed in twelve days. Iran’s nuclear program was set back by US assessment one to two years. The proxies couldn’t intervene effectively because they were already mauled. Iran had spent forty-five years building threats meant to make any attack unthinkable. Most of those threats didn’t deliver when tested.

Phase Eight — The Interwar Pause

June 2025 to February 2026. Eight months of reload, not peace. Iran tried to rebuild what it could. Negotiations with the US continued and failed again. Iran refused to settle nuclear enrichment. The IAEA reported Iran continuing toward weapons-grade material. The economy strained further. The regime claimed victory while the strategic reality kept deteriorating.

Phase Nine — The Current War Resumes

February 2026 onward. Trump and Netanyahu launched the resumption. Sustained kinetic operations. Khamenei dead. The strategic depth that took fifty years to build, gone in months.

Iran’s response told the story. Its long-range strike capability had been significantly degraded in the 12-Day War and rebuilt only to roughly half of pre-war levels. Spread over three times the duration, what Iran could fire produced far less strategic effect than the 12-Day War. Shorter-range systems and drones went to US bases in Iraq and to Gulf states. The long-range tier that mattered wasn’t there.

Phase Ten — The April Ceasefire

April brought the original ceasefire. Two weeks. Iran came to the table. The US laid out demands: nuclear program dismantled, Hormuz neutralized, proxies off the board. Iran refused the nuclear point. Talks collapsed.

That is the empirical proof. Even after the 12-Day War, even after months of resumed strikes, Iran still wouldn’t deal when it had cards left to bargain with. The blockade started because nothing else was going to move them.

Phase Eleven — The Blockade and Final Push

By mid-June 2026, Iran was on the ropes. Missile and drone stockpiles significantly degraded. Hezbollah’s command hollowed. Other proxies mauled. Nuclear facilities struck repeatedly. Economy in free fall. Oil exports near zero. The Economist magazine’s verdict: Iran’s battered economy will take years to recover.

The US had Iran cornered and was paying to keep them there. Naval blockade burning political capital. Stranded shipping. Markets jittery. European allies grumbling. The Iranian hostage doctrine, what remained of it, was working against the US even while the US ground Iran further down. Both sides had reasons to stop. Iran’s leverage was bleeding out faster than the US’s political costs. That asymmetry produced the MOU.

Phase Twelve — The MOU

In short, the ceasefire is continued and the strait is opened, and a set of topics to be resolved within the next 60 days is presented. Nowhere does it say that additional conditions cannot be added. No guarantees that the war does not resume if the remaining topics are not to Trump’s liking. He has already said that. Where does it say that the blockade is forever gone? Nowhere except in the partisan posturing and support of their proxies in the news media and the credentialed class.

Read the document strictly. Five paragraphs do real work on signature. Paragraph 1 ends military operations. Paragraph 4 lifts the naval blockade within 30 days. Paragraph 5 commits Iran to reopen Hormuz, demine, free passage for 60 days. Paragraph 9 freezes the nuclear program at its current state and bars new US sanctions or force deployments. Paragraph 10 directs Treasury to issue oil export waivers.

Everything else is deferred. Paragraph 6’s $300 billion reconstruction — mechanism to be finalized in the final deal. Paragraph 7’s sanctions termination — schedule to be agreed in the final deal. Paragraph 8’s stockpile disposition — mechanism to be mutually agreed. Paragraph 11’s frozen assets — procedures to be negotiated. Paragraph 13 makes the deferral explicit: negotiations on these provisions only begin after the operational paragraphs are being implemented, and either side can walk away at any point.

This isn’t a deal that offers Iran nothing. For opening the strait for 60 days plus extensions, Iran can sell some oil. Permission to sell under Treasury-controlled waivers — not sanctions relief, because the entire sanctions architecture stays in force. Restoring exports takes weeks. Net regime revenue over the window is probably two to three billion dollars after discounts, costs, and IRGC skim. That’s real money. It pays salaries and keeps the lights on.

But measured against the pre-war baseline there’s no comparison. Pre-war Iran exported 1.5 to 1.7 million barrels a day through the shadow fleet, indefinitely, with the leverage of Hormuz backing the price. MOU Iran exports a fraction of that, on terms Treasury controls, for 60 days at a time, with no leverage backing anything. Iran didn’t just lose volume. Iran lost the calendar. The pre-war oil money was modest and permanent. MOU oil money is discretionary and temporary.

What the US gets is the shipping unstuck, markets calm, gas prices down, the hostage layer Iran built into the global oil supply neutralized as a pressure point. The political cost that was compounding against Trump every additional week — shed.

And every instrument that produced the leverage is preserved. Sanctions architecture intact. Navy in position. Kharg in range. The kinetic option preserved by paragraph 13’s walk-away clause. Paragraph 9’s restriction is narrow — no new sanctions, no new deployments — and leaves existing enforcement untouched. Interdiction authority on cargo moving outside the narrow waiver channel is unaddressed in the document and therefore intact. The waiver itself is a dial, not a switch — Treasury sets the scope, can narrow it at 30 days, can pull it at 60.

What’s missing from the MOU is the clause Iran most needed and didn’t get: any protection against blockade reimposition if the final deal isn’t signed. Paragraph 4 commits the US to lifting the blockade on signature. It does not commit the US to keeping it off if no final deal arrives. Iran took the lift unconditional on the front end and got nothing on the back end. They had no choice.

The MOU also asserts authority neither signatory has. Paragraph 1 binds “the United States of America and the Islamic Republic of Iran and their allies in the current war” to permanent cessation of operations including in Lebanon. But Israel isn’t signing. Hezbollah isn’t signing. The US cannot deliver Israeli restraint. Iran claims it cannot deliver Hezbollah.

The MOU is the Halftime Show of the War

The world gets relief from the immediate issue of the strait being closed. The US was opening it anyway, and with the calm, Europe and Japan will swoop in and can be ready after 60 days to keep it open without Iran’s blessing. The rest of the world is already busy making new routes. There is no commitment to give Iran any relief beyond short-term oxygen for 60 days.

Not exactly the US waving a white flag.

Hostages by Design

You don’t start a war without understanding the opponent’s playbook. Iran’s was always clear: the Strait of Hormuz and commercial shipping weren’t side issues—they were standing defense doctrine. Disrupt, mine, threaten tolls or closure, entangle neutral tankers and crews. Make any strike on the regime a problem for the entire world economy. Hostages by design. And it doesn’t stop at shipping—it’s the ability to disrupt the entire region by striking other countries. That’s the whole point of the ballistic missiles and drones. They have them in enormous numbers, and the Middle East is where a large percentage of the oil for the world comes from.

The hostages are layered. The most immediate ones to deal with are the ships caught out there and the neighboring countries, all of them within range. But behind those sit the bigger ones—global oil supply, the stock markets, the whole jittery machinery of the world economy. Iran built the doctrine so that touching the regime meant touching all of them at once.

There’s no clean way to solve everything at once, because these issues aren’t separate problems with separate fixes—the shipping, the missiles, the proxies, the regional reach are all part of one thing: Iran’s existential defense mechanism. Pull on any single thread and the whole web moves. That’s not a flaw in the design. That’s the design.

And remember where this regime started: taking American hostages, 47 years ago. Hostage-taking isn’t a tactic they picked up along the way—it’s the founding act, now scaled from an embassy in Tehran to the world’s oil supply. Nothing has changed in that way.

They Moved Anyway

The U.S. knew it. Intelligence and military planners saw the second- and third-order effects coming. Impose a naval blockade to squeeze revenue and force movement, and you activate exactly the trap Tehran built in advance. Stranded ships, stressed seafarers by the tens of thousands, global oil spikes, neighbors on edge, Europe complaining about energy and lack of consultation. Leaks would have killed surprise and given Iran time to harden. So they moved anyway.

Now the MoU framework lifts (or authorizes the end of) the blockade. Ships start moving. Hormuz reopens, supposedly toll-free. Oil flows. Markets cheer. Trump calls it done. The immediate hostages are released as best as possible.

Cleanup, Not Concession

This isn’t some grand concession or sudden weakness. It’s the unavoidable cleanup phase of a conflict where Iran’s pre-war mechanism made sustained pressure a shared liability. You can’t choke the regime indefinitely without owning the broader mess you helped create by engaging their doctrine. Iran took the revenue hit knowing this dynamic would eventually force a reset. They need the money badly enough to take the pause. The U.S. needs the shipping unstuck and a political/economic win to point to.

No Way Around It

The core demands remain mirror opposites, just as I wrote days ago. Trump wants the nuclear program dismantled, proxies off the board, the strait neutralized for good. Iran wants it all preserved with breathing room to rebuild. There is no middle ground anymore—and there never really was. No document can split that difference, because the difference is the regime’s survival. The blockade bought leverage, but the hostage situation it triggered set a clock no one could ignore forever.

Consensus with Europe and the neighbors? Good luck. Full buy-in would have leaked, diluted surprise, and let Tehran play divide-and-manage. Yes, the allies lined up to welcome it afterward—but applauding a done deal isn’t the same as being consulted before one. The statement came after the fact was already created. Trump isn’t telling anyone the plan publicly, and for good reason—Washington just can’t keep a secret. It works better to let the pundits chase the wrong story.

So the U.S. acted, accepted the coming entanglement, and now we’re mopping up the visible pain points while kicking nuclear verification and enrichment into a 60-day window.

The Tell

And here’s the tell nobody’s pricing in. The framework doesn’t just defer the hard stuff—it leaves the hardest stuff off the table entirely. Per Iran’s own state media, the final agreement is scoped to enrichment, sanctions relief, and reconstruction. The missile program and proxy support—”resistance groups,” in their language—aren’t even being negotiated. That’s not an oversight. It’s both sides quietly conceding those points are irreconcilable, so they walled them off rather than pretend to bargain over them.

There’s a reason this can has been kicked down the road for nearly fifty years. Every administration since 1979 has hit the same wall: there’s no way to actually deal with Iran without opening a can of worms—the regional blowup, the oil shock, the whole hostage cascade Tehran engineered—and nobody wanted to be the one to do it. The ceremony covers the shipping; the exclusions cover the war.

A Pause, Not a Resolution

Call it pragmatic. Call it sequencing. But don’t call it resolution. The war is far from over. This is a tactical pause dressed up with a signing ceremony. Iran sees the tactic, spins it as victory at home, and gets treasury relief to regroup. We get lower gas prices and optics.

As I said the other day, Nobody Is Negotiating Over Iran.

This latest drama is just the halftime show.

The thing worth understanding doesn’t require the document. It sits one level above it, in the shape of what each side is demanding.

The Two Demands

Trump’s demand is non-negotiable: Iran’s program dismantled. The uranium gone. Natanz, Fordow, Isfahan turned to gravel. The proxies finished — Hezbollah, Hamas, the militias on retainer — taken off the board. The neighbors left alone — no menacing the Gulf, no arsenal pointed across the water. The Strait of Hormuz open and untolled, no longer a valve Tehran can squeeze shut at will. Anything less and the war he joined bought nothing.

Iran’s demand is non-negotiable, and it is the exact mirror. The program kept — enrichment, capacity, the bomb within reach. The proxies funded and back in the field, Hezbollah and Hamas and the militias rearmed. The missile and drone stockpiles rebuilt — thousands of them, ranged on every capital within reach. The Strait of Hormuz under Tehran’s own hand, a valve it alone decides to open or shut. The frozen assets released, the money moving again. Every one of these is a card that makes the regime too dangerous to touch — the lesson every strongman took from Libya and re-learned from North Korea. Surrender them and survival becomes a favor others can withdraw.

Now hold those two demands next to each other. They are not far apart. They are opposite. One is the negation of the other. There is no version of the world in which Iran both keeps and surrenders its program. Whatever middle once existed — caps, inspections, enrichment held to a line — both sides have planted their flags well past it and called the flag non-negotiable.

Why No One Blinks

That alone wouldn’t freeze them. Incompatible demands get traded down all the time, when each side fears the cost of holding out. What freezes them is the second half: each is sure it can still get its maximum.

Trump is sure the blockade destroys Iran. So he waits to collect the dismantlement rather than bargain for it. And if the blockade leaks, Plan B is already drawn up: seize or flatten Kharg Island, the terminal that loads almost all of Iran’s crude. No island, no exports, nothing left for the shadow fleet to carry. And Hormuz he doesn't bargain for at all — he's already clearing it himself, with or without Iran's agreement, mines coming up and the Navy in the water. Commercial ships are moving through under escort, past everything Iran laid down to stop them.

Iran is sure the midterms break Trump. So it waits to keep the program rather than trade it away.

Put it together. Two demands that cannot coexist. Each non-negotiable to the side that holds it. And each side convinced it will get its own, whole — so neither has a reason to come down. A deal requires somebody to accept less than their maximum. Here nobody will, because nobody believes they have to.

That is why this is not a negotiation. It has the furniture of one — envoys, mediators, drafts, deadlines — and none of the substance, because substance would mean a number both sides could live with, and there isn’t one. There is only the waiting.

Could they sign something anyway? Of course. A ceasefire, a memorandum, a framework thick with the word progress — paper is cheap, and a signing ceremony makes everyone in the room look useful. But any such document runs into the same wall. It cannot dismantle Iran’s program; Iran won’t permit it. It cannot secure Iran’s program; Trump won’t permit it. So it touches neither demand, and a document that touches neither demand settles nothing. It is a photograph of a handshake over an empty table. The war goes on existing the moment the cameras leave.

Where I Could Be Wrong

There are other ways to be wrong, and the favorite is that both sides quietly lose their nerve and settle in the middle. But that one needs a middle to settle in, and there isn’t one. What’s left are two bets, each that a clock has already run out — and only one of them can be checked from where we sit.

The first. Maybe Iran is not waiting to lose but already beaten — the economy in freefall, the military hollowed, the regime nearer the street than the centrifuge — one shove from a surrender it can dress up as a face-saving deal.

The second — the United States is, in effect, out of ammunition. America has more than enough to blast Iran into the Stone Age several times over. The limit on Washington was never the size of the magazine. Set that one down.

We go to bed the way a child does on Christmas Eve. We’ll see what Santa brings.

The open question is: when does Iran realize it’s been played?

Around the first of October, or perhaps a little sooner. Not election day.

Tehran is watching November like a countdown, certain he panics into a deal before the vote.

Iran’s whole theory of leverage was the strait: close Hormuz, choke a fifth of the world’s oil, make him pay to reopen it. He isn’t buying it back. He’s clearing it himself — Navy in the water, mines coming up — calling it a favor to the world.

They cling anyway, because they’ve got the wrong man in their heads — Obama, who wanted his deal and could be bled toward it across two years. They keep waiting for that president to walk in. The ghost of Obama past.

It will start to dawn on them as November gets closer.

A negotiation that’s permanently “almost done” gives a newsroom nothing to run. Same three words every week — deal is close — no motion, no winner, nothing to photograph. The readers stop clicking. The thing gets filed under weather. And once the clicks dry up, editors do the only thing editors do: move the story off the front, then off the page, then out of the morning meeting.

The “deal is coming soon” story becomes the invisible man from the Monty Python sketch — the one who is so tedious and boring that after a while, he becomes invisible.

You don’t think Trump is being boring and repetitive on purpose, to kill the story?

The next two years after that, if the regime survives, will be the same.

Other drama from a desperate Iran could reopen the kinetic phase.

A negotiated settlement is not a likely outcome on any timeline.

This morning the Pentagon pushed people into corridors and locked down floors because a sensor reported anthrax. It was a malfunction. Nothing was there. But anthrax is still the word that empties a building on contact, and the reason runs back a quarter-century. Most people remember that the attacks happened. Far fewer remember how the investigation ended. Here is the whole sequence.

The Man Before the Case

April 22, 1946 — Bruce Edwards Ivins is born in Lebanon, Ohio, the youngest of three sons of a pharmacist.

Early 1960s — As an undergraduate at the University of Cincinnati, Ivins asks out a member of the Kappa Kappa Gamma sorority and is turned down. By his own later account, he becomes fixated on the sorority, a grievance he would carry for decades. He would tell investigators in 2008 that “they’ve been after me since the 1960s, and REALLY after me since the late 1970s.”

1968 to 1976 — Ivins earns three degrees at the University of Cincinnati, culminating in a PhD in microbiology in 1976.

1976 to 1978 — He breaks into the Kappa Kappa Gamma house at the University of North Carolina to steal the sorority’s ritual materials and a device used to decode them, and later breaks into the chapter at West Virginia University. A psychiatric panel reviewing his records years later would trace his problems to a “traumatic, damaging childhood.”

December 1980 — Ivins joins the U.S. Army Medical Research Institute of Infectious Diseases (USAMRIID) at Fort Detrick, Maryland, where he will work for nearly three decades as a senior anthrax-vaccine researcher. His employer never evaluates him for mental fitness during his career there.

1997 — Ivins creates and becomes the custodian of a large flask of liquid anthrax spores, designated RMR-1029.

2000 to 2001 — Ivins’s mental state deteriorates, documented in his own emails; in one from August 2000 he writes that he experiences “paranoid, delusional thoughts at times.” During this period the anthrax vaccine program to which he had devoted his career is in trouble, drawing criticism over potency and over alleged links to Gulf War Syndrome, and he fears it will be shut down.

The Attacks

September 18, 2001 — One week after the September 11 attacks, the first anthrax letters are postmarked. They are addressed to newsrooms: NBC, ABC, CBS, the New York Post, and American Media, a tabloid publisher in Florida. The letters are printed in block capitals with slogans built to read as jihadist: “Death to America,” “Allah is Great.”

October 5, 2001 — Robert Stevens, a photo editor in Florida, becomes the first to die. In all, the attacks cause 22 anthrax infections; five are fatal. The dead include two postal workers at the Brentwood facility in Washington, a hospital worker in New York, and a 94-year-old woman in Connecticut.

October 9, 2001 — A second batch of letters is postmarked, addressed to two United States senators, Tom Daschle and Patrick Leahy.

Mid-October 2001 — The anthrax from the Daschle letter is sent to USAMRIID for analysis. Ivins, one of the Army’s top anthrax experts, helps examine it, describing the material in his report as extremely pure and highly concentrated, not, in his words, “garage” spores.

The Attacks and the Case for War

The letters arrived in the rawest weeks after September 11, printed to read as foreign jihadist terror. The fear they produced did not stay confined to the criminal case. Over the next year and a half, the attacks were folded into the Bush administration’s argument for invading Iraq.

Late October 2001 — ABC News, citing four anonymous sources, reports that the Daschle anthrax contained bentonite, an additive it ties to Saddam Hussein’s bioweapons program, noting that as far as was known only Iraq had used it. The claim is the network’s lead story for five straight days. White House press secretary Ari Fleischer later says he had explicitly told ABC the report was wrong before it aired, and the network broadcast it anyway. The finding is false. No bentonite is present, ABC eventually corrects the story, and the sources are never identified.

A word on anonymous sources, here and always: when a claim rests on unnamed officials, two questions go unanswered that the reader should never stop asking — what does the source actually know, and why is he talking? A leak is a choice, and the choice serves someone. The bentonite story was false, traceable to no one, and it pointed at Iraq precisely when the case for war was being assembled. That the leak was deliberate cannot be proven. But it is fair to ask who gains when an unattributable, unverifiable, and false story leads a national newscast for five days, and whether the people who supplied it likely already knew it was false when they did.

January 30, 2003 — Deputy Secretary of State Richard Armitage tells the Senate Foreign Relations Committee that a single letter holding one teaspoon of anthrax had thrown the body into chaos in 2001 and killed two postal workers, then states that Iraq was believed to hold some 25,000 liters of anthrax, more than five million teaspoons, unaccounted for.

February 5, 2003 — Making the administration’s case for war at the UN Security Council, Secretary of State Colin Powell holds up a vial of simulated anthrax and reminds the chamber that a teaspoon of dry anthrax in an envelope had, in his words, “shut down the United States Senate in the fall of 2001.”

No link between the 2001 attacks and Iraq was ever established. The FBI’s eventual conclusion was that the anthrax had come from a flask in a U.S. Army laboratory in Maryland, and that the man responsible was an American government scientist.

The Investigation

2002 — Attorney General John Ashcroft publicly calls former Army scientist Steven Hatfill a “person of interest.” Hatfill is placed under extended surveillance and named repeatedly in press leaks. The same year, asked to submit a sample of his RMR-1029 flask to the FBI’s genetic repository, Ivins turns in one missing the genetic markers the flask carried (the same markers that linked it to the attack anthrax), which the bureau later characterizes as an attempt to mislead investigators. The investigation, opened after the 2001 attacks and codenamed Amerithrax, would become the largest in the bureau’s history: roughly 25 to 30 full-time investigators, more than 10,000 interviews, over 6,000 pieces of evidence, 5,750 grand jury subpoenas, and 5,730 environmental samples from 60 sites.

2005 — Investigators narrow the genetic trail to RMR-1029. Analyzing 947 samples, they find eight that carry the same four-mutation signature as the attack anthrax, all traceable to that flask. They interview Ivins, its custodian, as one of the people with access. An investigator also notes that the Princeton, New Jersey, mailbox the letters were dropped in, nearly 200 miles from Ivins’s Maryland lab, sits beside the local chapter of Kappa Kappa Gamma; to the bureau, his decades-long fixation on the sorority helps explain why the letters were mailed from so far away. The same year, investigators conclude Hatfill could not have had access to the flask at all.

2006 to 2007 — Ivins becomes the central suspect. As scrutiny intensifies, he is treated for depression, behaves erratically, and is stripped of access to sensitive areas at the lab.

The End

June 2008 — The government settles Hatfill’s lawsuit for 5.82 million dollars and issues a statement clearing him of involvement. No motive had ever been established against him.

July 9, 2008 — At a group therapy session, Ivins describes a plan to harm coworkers, according to his counselor, Jean Duley. The next day she has him committed and contacts the police.

July 24, 2008 — Duley testifies at a Maryland hearing in seeking a protective order, telling the judge that Ivins had obtained a gun and a bulletproof vest, intended to “go out in a blaze of glory,” and had been characterized by psychiatrists as a “sociopathic, homicidal killer.” The New York Times obtains and posts the audiotape, and it is widely broadcast.

July 29, 2008 — As the Justice Department prepares to seek an indictment, Ivins dies of an overdose of acetaminophen. He is 62. He has never been charged, and he never stands trial. His attorney maintains his innocence and attributes his death to the pressure of the investigation.

August 6, 2008 — At a Justice Department news conference, officials announce their conclusion that Ivins was responsible for the attacks.

Closure and Review

February 19, 2010 — The FBI issues its Amerithrax Investigative Summary, concluding that Bruce Ivins carried out the 2001 anthrax attacks and acted alone, and formally closes the case. The bureau’s theory of motive is that the failing vaccine program had put Ivins under severe professional and psychological strain, and that the attacks would revive urgency and funding for anthrax-vaccine work, including a newer vaccine he had helped invent. As for why Daschle and Leahy, officials offer two theories: that Ivins saw them as obstructing research funding, or as lapsed Catholics over their abortion-rights votes. His attorney rejects the second.

February 2011 — A National Research Council panel, convened at the FBI’s request, releases a 190-page report finding that the genetic evidence is consistent with the bureau’s conclusion but does not, by itself, definitively prove the attack anthrax came from RMR-1029. The same panel finds no evidence the anthrax was chemically weaponized with a silicon additive.

March 2011 — An Expert Behavioral Analysis Panel reviewing Ivins’s psychiatric records concludes he had severe, diagnosable mental illness and a profile that should have disqualified him from a security clearance and from access to pathogens.

2015 — The Government Accountability Office reviews the FBI’s genetic analysis and recommends improvements to its methods. Neither this review nor the 2011 panel identifies another suspect, and neither clears Ivins.

Today

June 11, 2026 — A malfunctioning sensor reports anthrax at the Pentagon, and part of the building is locked down before the alarm is found to be false.

The FBI investigated for nearly ten years, concluded that Bruce Ivins mailed the anthrax letters and acted alone, and closed the case. That is the government’s official and final conclusion, and it stands as the answer of record.

The attacks also reshaped American biodefense. Within two years the government launched BioWatch, a network of air sensors built to detect aerosolized anthrax and other pathogens across more than thirty cities, and committed nearly 6 billion dollars to Project BioShield for new vaccines and treatments. The detection equipment that emptied part of the Pentagon this morning descends from that buildout.

The attacks of 2001 were real, lethal, and never repeated. And a quarter-century later, the word anthrax alone is still enough to clear the Pentagon.

A note for the men in Tehran, still waiting on a signature.

Every story about Iran now arrives with a clock attached. Read closely and you’ll find it in the second or third paragraph, never the headline, smuggled in as a subordinate clause: with the midterms looming. As pressure builds ahead of November. Facing a restive electorate. The clock is presented as context — the thing everyone already knows.

Everyone does not already know it. Everyone has been told it.

Here is the part worth sitting with: the Iranian supreme leader is betting his country on the very same assumption — that all strategy fits into the midterm clock. When a Times analysis, an Axios scoop, and Tehran’s negotiating posture all run on the identical proposition, you are not looking at three independent reads of reality. You are looking at one model, repeated until it sounds like fact.

The model is simple. Presidents run on the electoral calendar. Pain before an election is intolerable. Therefore Trump must resolve Iran — deal or retreat — before November, or pay at the polls.

It’s a fine model. It describes most presidents most of the time. It does not describe this one, and the press cannot tell the difference, because the electoral calendar is the only calendar a newsroom knows how to keep.

The Clock Is a Narrative Engine

A countdown is a gift to an editor. It manufactures suspense out of stasis. Nothing has to actually happen — the blockade can hold for the ninth straight week, the talks can go nowhere for the ninth straight week — and the looming deadline keeps the story alive anyway. Can Trump hold the line as November approaches? The frame implies a cornered president, which is this outlet’s preferred posture for this president. And it lets every non-event be filed under rising stakes.

That single word, looming, does all the work. It asserts a causal link — between the calendar and the policy — that no one has demonstrated and no one is asked to. It is not reporting. It is a device for turning the absence of news into the appearance of momentum.

What the Man Actually Did

Set the model aside and look at the behavior.

He said, on the record, in a cabinet meeting, that he does not care about the midterms. Nobody believed him. That’s their problem, not his — because he then proceeded to act exactly like a man who meant it.

He declined to sign a deal that was, by his own description, largely negotiated. He left the blockade running through an indefinite extension. He went on national television and told Iran the frozen money comes after they behave, not before — slamming shut the one door that could have bought him a clean, photogenic, pre-election win.

A president optimizing for November books that win in October. He gift-wraps the ceasefire, stages the Rose Garden ceremony, and campaigns on it through the fall. Trump did the reverse at every turn. The revealed preference is not a man racing a clock. It’s a man who threw the clock in the drawer.

And why would he carry it? He is a second-term president. There is no re-election. The electoral whip the model assumes — the one that disciplines a first-termer — is not in his hand. What he has is two and a half years and a strategy built on attrition. Attrition has no November in it.

What’s Actually on the Calendar

There is a clock. The press is just reporting the wrong one.

The only date Trump answers to is January 20, 2029 — the morning he hands back the keys. Everything between now and then is runway, not countdown. Two and a half years in which nothing on the electoral calendar can compel him, because he has already run his last race.

And the plan for that runway is not a plan to reach a deal. It is a plan to reach the exit with Iran bled white and his signature on nothing.

The instruments are already set, and longtime readers know them. The three clocks do the work: the economic clock draining the regime by the day, the geological clock locking Iranian oil into the ground for good, the relevance clock building every pipeline and terminal that makes the Strait of Hormuz matter a little less each month. No second kinetic campaign — the harpoon is already in, and you don’t reopen the war once the hook is set. Just hold the blockade, keep the alternatives to Hormuz coming online, and let the economy bleed.

A deal would stop the bleeding. Trump does not want to stop the bleeding. So there is no deal — not this year, not next year, not before he walks out the door. The question the press keeps asking — will he sign before November? — was always the wrong question. The right one is whether he ever signs at all. The calendar answers it. No.

And he is right not to want one. Only a fool believes there is a deal to be made with the current regime that solves anything.

Forty years of evidence says so: every agreement is a pause the regime spends rebuilding, every concession is one it relabels rather than surrenders, every signature is the delay game in a fresh suit. The JCPOA was the masterclass — getting paid to slow down something it never had the slightest intention of stopping. A deal does not change what the regime is. It preserves it. And the regime is the problem. So the paper that “solves” Iran would solve nothing; it would only stop the bleeding before the bleeding had done its work.

So yes, there is a countdown. It does not read November, and it is not ticking toward a signing ceremony. It reads January 2029, and it counts down to an exit with the Iranian economy in ruins and nothing on paper to undo.

They keep checking his face for the blink. They are watching the wrong clock. The one that matters hangs on his wall, not theirs — and it is not counting toward anything they are prepared to report.

That person is not Steve Witkoff. Not Rubio. Not Vance. Not the military advisors. Not the hawks. Trump has not confided his game to anyone. That’s the strategy.

Washington cannot keep a secret. Trump learned this the hard way during his first term, watching every strategy briefing become a Times story and every negotiating position become an Axios scoop within 48 hours. Most presidents try to fix the leak. Trump worked around it.

If nobody knows the plan, nobody can leak it.

So Witkoff walks into Doha a true believer. His enthusiasm is real. His frustration when Iran pushes back is genuine. You cannot fake that at a negotiating table — Iranian diplomats who have been reading American negotiators for 45 years would detect a performance immediately. They can’t detect this one because there’s nothing to detect. Witkoff isn’t performing sincerity. He is sincere.

And Iran stays at the table. Waiting for a signature that isn’t coming.

Iran has run the delay game on every American president for forty years. String them along. Pocket whatever sanctions relief you can get. Survive another round. They did it to Bush. They did it to Obama. They turned the JCPOA into a masterclass in getting paid to slow down something you were never going to stop.

They think they’re running the same play now.

They’re not. Trump blocked their blockade. Now he’s stalling their stalling. He took their own game and ran it back on them with better cards, more patience, and an economy that isn’t collapsing.

Theirs is.

Every day of “negotiations” is another day of Iranian economic deterioration. The Strait stays closed. The blockade holds. The frozen assets stay frozen. The oil stays unsold. The currency keeps falling. Iran came to the table, moved on their demands, announced they were ready to sign. Trump said he needs a couple of days to think about it.

“I won’t sign a bad deal” does the rest of the work. Trump defines what’s bad. Nobody else does. There’s no external arbiter, no Senate ratification required for an MOU.

John Bolton thinks the midterms will break Trump. Jack Keane thinks the midterms will break Trump. The Iranian supreme leader thinks the midterms will break Trump. Trump said publicly, on the record, in a cabinet meeting: “I don’t care about the midterms.”

Nobody believed him. That’s their problem, not his.

The economy is the strategy. The stall is the weapon. And when Iran finally figures out there’s no deal coming — when the patience breaks and they do something that forces the issue — the kinetic response will be ready, fully justified, and nobody will be surprised except the people who spent months writing deal-imminent stories.

The stall isn’t the prelude to the war. The stall is the war. Kinetic is just the punctuation.

Previous pieces in this series: What’s Next in the Iran War? | The Hall of Mirrors | The Three Clocks Running Against Iran | How Will We Know When the Iran War Is Over?

The headlines keep coming. Talks. Progress. Frameworks. Envoys shuttling between capitals. Trump, we’re told, is close to a historic deal with Iran.

Don’t make me laugh.

There is no peace deal coming. There is no negotiation in any meaningful sense of the word. What there is: a pause of mutual convenience that both sides are using to buy time — for completely opposite purposes.

I called this in April. In Iran: Ceasefire or Pause of Mutual Convenience?, the argument was simple. Washington gathers intelligence, watches Iran from satellites, and works out the war plans for the next phase. Tehran catches its breath and hopes the American political calendar saves them. Call it diplomacy if it makes you feel better.

Let’s ask the obvious question nobody in the mainstream press bothers to ask: does anyone seriously believe Iran will negotiate away its nuclear program, its ballistic missiles, and its proxy network — and actually abide by the deal?

The only thing that stops Iran is running out of money to continue. I said that at the start of the war. Nothing I’ve seen since changes that conclusion.

So who believes a real deal is coming? The press, apparently. Democratic party leadership, who need Trump to fail badly enough that they’ll root for Iranian diplomacy. That’s the complete list. Every serious analyst, every intelligence professional, every person who has spent five minutes studying Iranian decision-making knows this is theater.

What the US Is Actually Doing

The United States and Israel are finishing the war plan and arming to the teeth. There have been no testy discussions between Trump and Netanyahu. Fake news. They are in lockstep.

There’s no rush. Every week of “negotiations” is another week of Iranian economic deterioration. I laid out the arithmetic in The Three Clocks Running Against Iran.

Washington has every incentive to let that clock run.

Trump is just playing their own games back on them. Iran creates a blockade — he builds one on top of theirs. They stall for time — he stalls on top of their stalling. While they’re congratulating themselves on the delay, their economy is collapsing underneath them.

What Iran Is Actually Doing

Iran’s hardliners may think they’re running the same play they’ve run for twenty years. It’s hard to know what’s going through their heads. String the Americans along. Pocket whatever sanctions relief or additional time to keep enriching uranium they can get. Survive another round.

They’ve done this to every American president. Obama. Bush. The Europeans for decades. They watched the JCPOA get negotiated — no dismantlement required, just delays and caps. They know the choreography.

The mainstream press may have convinced them they are winning — that the midterms will clear the slate and this all goes away. I wrote about how that happens in The Hall of Mirrors.

What the Press Is Actually Doing

The mainstream press is having a wet dream. Trump is negotiating a peace deal — they were crucial in making it happen! The self-congratulations are deafening.

What they don’t have is any idea what’s actually happening.

And by their wonderful investigative reporting and anonymous sources, they believe they have held the president accountable — exposing what he is really doing, or not doing, and the inconsistencies in his explanations about the war.

The mainstream press is so predictably dumb, and so convinced of its own sophistication, that Trump plays them like a fiddle. He knows what story they’ll write before they do. He hands it to them.

The Endgame

How will this end?

I wrote about this in How Will We Know When the Iran War Is Over? The war ends when Iran stops bargaining and starts begging. The real endgame isn’t a treaty. It’s the moment Iran can no longer afford to be Iran.

Here’s a simple test for whether this “peace deal” is real: watch for the arrival of the French. When there is no danger to them and no difficult decisions to be made, they will arrive and insist on having a major seat at the table. Haven’t seen this yet.

Don’t let the headlines fool you. The war plan is almost ready. The press will figure it out when the bombs drop.

Previous pieces in this series: Iran: The Bill Has Come Due | Will the US Seize Iranian Oil and Natural Gas Assets? | The Iran War: A Game of Three Card Monte | Iran: Ceasefire or Pause of Mutual Convenience? | Iran’s Economy: Conservatorship or Destruction | The Three Clocks Running Against Iran | The Hall of Mirrors | The Iran War: Why Now? | Iranian Currency Switches to Scientific Notation | How Will We Know When the Iran War Is Over?

Elon Musk is taking SpaceX public. The ask is $1.75 trillion.

Let’s talk about what that number means. And what it doesn’t.

SpaceX is a remarkable company. The Falcon 9 is the most-flown active orbital rocket and the backbone of modern launch cadence. Starlink has over 10 million subscribers and dominates space-based internet traffic. These are real accomplishments. Nobody serious disputes them.

But a business in an IPO is worth exactly one thing: the free cash flow it generates till the end of its life, discounted back to today. Not revenue. Not narrative. Not the founder’s Twitter following. Free cash flow, discounted.

Free cash is cash you could give out as a dividend if you wanted to. It’s never needed for future expenses or growth. It belongs to the owners, free and clear. Everything else — earnings, EBITDA, adjusted this, pro forma that — is accounting. Free cash is reality.

Discounted means adjusted for the cost of money over time. A dollar received ten years from now is worth less than a dollar today — because you could have invested that dollar and grown it in the interim. Inflation eats the value of money too. The longer you wait for the cash, and the riskier the bet that it ever arrives, the less it is worth today.

Think of an IPO — or any investment — as a gold mine. The gold is the product you make that generates free cash for you. Eventually you mine it all. The gold runs out, or nobody needs your gold anymore, or someone can make it for less.

The discounted free cash the business earns must be more than what the IPO was valued at.

No business lasts forever. A competitor can make something just as good or better, flood the market, and kill the price of yours. Technology can make your mine obsolete before you’ve dug halfway through it.

Your gold mine might have ten billion dollars worth of gold in it but if you spend a hundred billion dollars mining and processing it, you just lost ninety billion dollars. The point is not to just break even. You need to make a profit — and one that accounts for the money you could have earned with the cash you paid for the IPO over that same lifetime.

Wall Street does not like to show you this math. They prefer forward revenue multiples and TAM-based narratives — frameworks that sound rigorous but avoid the inconvenient question of how much free cash actually escapes the machine. Usually they also have a greater fool story — that momentum will make someone else buy it from you for more.

A prudent investor ignores that entirely. For purposes of value calculation, they assume they can never sell it again. In the end, all they will ever have is the free cash the business generated. That is the only number that matters.

When Warren Buffett considers buying a company, he first calculates its value and then looks at the stock price. If the value is much more than the stock price, he might buy it. Not the other way around.

When you run the free cash math on SpaceX, the number is not $1.75 trillion without a lot of help from the tooth fairy.

Mr. 10x

You might have thought I call Elon Mr. 10x because he has so many companies with the name X in it. SpaceX. xAI. Twitter, which he renamed X. A reasonable hypothesis. Nope. That’s not what the X means.

Musk has earned a nickname. Call him Mr. 10x. The 10x is the multiplier you attach to everything about the business — the timelines, the capabilities, the valuation. It is always there. It is always wrong in the same direction.

The pattern is consistent enough to be a physical constant. It takes ten times longer to deliver than he promises. He promises ten times more capability than he delivers. The valuation he gets is ten times the true valuation.

I covered the 10x pattern in detail for Tesla in The True Value of Tesla.

The SpaceX IPO is the 10x pattern applied to valuation — and most likely also to schedule and capabilities. He’s asking you to pay today for a company that needs to produce Apple-scale free cash flow — $100 billion annually, sustained — to justify the price. Free cash is cash you can pay as a dividend. Apple proves it — they do exactly that, billions every quarter. At $1.75 trillion, SpaceX would need to match Apple’s output for nearly twenty years just to return the purchase price. Undiscounted.

Smoke and Mirrors

AI is hot. The promise of transformational technology is in the air. SpaceX bundled rockets, satellites, a bleeding AI startup, and a cancelable compute contract into one whale’s tale and filed it with the SEC. The goal was a 10x valuation multiplier — tech multiples on an aerospace business. It worked. For now.

The IPO prospectus now pitches a combined entity with a $28.5 trillion total addressable market anchored on AI infrastructure.

The S-1 includes an Anthropic deal: Anthropic agreed to pay xAI $1.25 billion per month for compute capacity through May 2029 — a contract that could exceed $40 billion. One can assume it’s there to buttress the future earnings story.

But what are we really talking about here? SpaceX is renting excess capacity to Anthropic. There is nothing to stop Anthropic from building its own data centers or getting them from someone else. The contract carries a 90-day termination clause for either party. This is not likely a permanent revenue source. The implication that it is some kind of continuing revenue source is there (?) — since $40 billion as a one-time revenue item would be noise as part of a $1.75 trillion valuation.

The Math Nobody Is Doing

You can buy a piece of the future. That is what equity is. But just because the future is big does not mean it is a good investment at the price being offered. There has to be math. You have to estimate the size of the gold deposit, the cost to extract it, the timeline, the competition, the discount rate. Then you look at the price being asked and you either accept or reject the assumptions baked into it. Most SpaceX bulls skip that step. They see a big future and stop there. That is not investing. That is hoping.

At $1.75 trillion, with a 10% discount rate and modest terminal growth, SpaceX needs to generate roughly $120 billion in annual free cash flow. Essentially permanently.

Let’s look at the real business here — SpaceX minus the AI spin. The AI spin is just spin at this time. xAI is not Anthropic or OpenAI. xAI was carried at an implied valuation of roughly $250 billion in the merger. Whether that number is justified is its own leap of faith — Mistral, a genuine AI company with real enterprise revenue, is valued at $13.8 billion. Twitter — now X — is a perennial money loser. In 2025, xAI generated $3.2 billion in revenue and lost $6.4 billion. That is the AI business being bundled into a $1.75 trillion valuation.

Starlink at maturity — 30 to 40 million subscribers, $66 per month average (and falling as it expands into lower-income global markets), facing real competition from Amazon Leo and Blue Origin’s TeraWave (with a whole list of additional suitors on the way) — generates maybe $26 billion in revenue. 35 million subscribers at $66 per month is roughly $27 billion in revenue. At a 15% free cash flow margin — generous for a business that must continuously replace satellites — that’s $4 billion in free cash annually from Starlink.

Launch services at maturity: $12 to 15 billion in revenue. At a 20% free cash flow margin — better than satellite because reusable rockets don’t fall out of orbit — that is roughly $3 billion in free cash flow.

Government and defense contracts: $1 billion.

Total, at peak, on optimistic assumptions: $8 billion in annual free cash flow.

And that is at maturity — and assuming no significant competition compresses margins. Both are generous assumptions.

Now discount it. SpaceX doesn’t reach that state until 2032 at the earliest. Ten years out at 10% cuts the present value roughly in half. Assign a terminal multiple (FCF lingo) of 10x. You land somewhere between $50 and $150 billion in fair value, depending on how generously you treat the assumptions.

Let’s consider a somewhat comparable company, SiriusXM. After 25 years, one near-bankruptcy, and a forced merger between the only two competitors, SiriusXM generates $1.26 billion in free cash flow on $8.5 billion in revenue. The market values it at $9 billion. That is what a mature satellite subscription business actually looks like when the hype is gone.

The market is asking $1.75 trillion. That is Apple prices for a company with SiriusXM economics and xAI risk.

The Capex Treadmill

SpaceX is not See’s Candies. It’s Sun Microsystems with rockets. Sun reported billions in earnings for years. Every dollar was already committed to the next hardware cycle. Skip one cycle and you’re dead. Oracle bought the corpse for $7 billion. SpaceX’s constellation replacement cycle is that treadmill. It never stops.

Is There a Moat or a Puddle That Can Be Stepped Over?

Reusability is proven. Every serious launch company is now building reusable rockets because SpaceX demonstrated it works. Blue Origin has New Glenn. Rocket Lab is building Neutron. The head start is real. The exclusivity is gone.

The one genuine moat is orbital slot and spectrum allocation. SpaceX filed first, deployed first, and locked up premium low-earth orbital positions. Competitors face interference constraints that SpaceX doesn’t. That’s a real advantage.

It’s also a regulatory advantage, which means a government body can modify it. That has happened before in spectrum history.

Consider the benchmark. Tesla is Musk’s best execution — twenty years of manufacturing, iteration, and genuine engineering achievement. It survived only because a government tariff wall keeps Chinese competitors out of the American market. In the markets where the wall doesn’t exist, Xiaomi — a phone company — replicated Tesla’s core product in under two years and is now outselling it in China. Xiaomi didn’t invent anything. They assembled largely off-the-shelf components and applied their manufacturing expertise. Two years.

Mr. 10x is not some uber genius. Except for the rocket business, he is not even the best at anything I am aware of. Well, one thing — he is best at getting 10x valuations.

It is amazing how his fan boys seem to see something else.

If his strongest company, built over two decades, can be duplicated by a phone manufacturer in under two years the moment the market is open, what does that tell you about the durability of SpaceX’s advantages? Orbital slots are real. First-mover scale is real. Neither stopped Amazon, Blue Origin, and AST from entering the same market with serious capital. The moat is narrower than the price implies.

What the S-1 Actually Says

Read the filing and you find two things: a real rocket company, and an episode of the Twilight Zone.

The S-1 glossary — a legal document filed with the SEC — defines Kardashev Type II civilizations, lunar mass drivers, orbital AI data centers, and something called “Macrohard.” SpaceX filed under SIC code 7370: Computer Programming and Data Processing. Twenty-four years building rockets, and the moment they go public, they’re an AI company. The glossary isn’t accidental. It filters out investors who would ask uncomfortable questions about free cash flow and attracts believers willing to buy the dream at any price.

The investors lining up for this IPO are like the passengers in the Twilight Zone episode “To Serve Man” — boarding the rocket to a better life, certain they’ve been chosen for something extraordinary. “To Serve Man” turned out to be a cookbook.

Elon Musk’s S-1 isn’t a business plan. It’s a cookbook. And you’re the main course if you are buying into this valuation.

This also reminds me of something else — the people who waited in matching outfits for a spaceship that never came. Heaven’s Gate was also built on absolute conviction that the ship was coming. The ship did not come.

The S-1 also discloses that Musk retains 85.1% voting control through a super-voting share structure. You own the risk. He owns the company.

At $1.75 trillion, you’re not buying the business. You’re buying the dream — Mars colonization, point-to-point Earth transport, and orbital manufacturing.

Bon appétit.

Tesla’s market cap is approximately $1.4 trillion. That makes it one of the most valuable companies on earth — worth more than every traditional automaker combined, more than ExxonMobil, more than JPMorgan Chase.

There is a simple, honest way to value Tesla that nobody in the financial press wants to say plainly:

As a government-protected monopoly, it is worth roughly one-tenth of what it trades for. In a free market, it has no future.

Priced for vague future potential.

The business, as currently constituted, would not survive open competition.

The Invisible Tax on Everyone Who Isn’t Rich

The average new car in America costs $51,000. The average monthly car payment has crossed $700. Roughly 40% of Americans live paycheck to paycheck. Auto loan delinquency rates are at multi-decade highs. Affordable, reliable transportation is not an abstraction for most Americans — it is the difference between keeping a job and losing one, between making rent and not.

Into this environment, a fully-featured electric vehicle exists that sells for $10,000 to $15,000 in its home market. It has solid build quality, modern technology, and better real-world economics than most internal combustion vehicles thanks to lower fuel and maintenance costs. In Brazil it captures 60% of the EV market. In the UK it’s growing at triple digits year over year. In Southeast Asia it’s the default choice for first-time EV buyers.

Americans cannot buy it. A 100% import tariff — the functional equivalent of a ban — ensures that. That tariff was imposed by the Biden administration in May 2024 — the same administration that watched inflation gut working families and then erected a wall keeping a $10,000 car out of their reach. Democrats lost the 2024 election partly on the cost of living. They are now talking about affordability. They built the wall before they discovered a cause that might get them elected. (I ran the five-year cost comparison in The Price of Affordability — the numbers are not close.) And the tariff doesn’t just block one cheap car. It keeps EV prices artificially high across the entire market, slowing adoption for everyone who might benefit from lower fuel and maintenance costs.

What the Business Actually Is

Strip away the narrative and look at what Tesla is: a modestly profitable auto company with declining sales and an aging product lineup, trading at 362 times earnings.

It sold 1.64 million vehicles in 2025 — down 9% from the prior year, the second consecutive annual decline. Net income was $3.8 billion on $95 billion in revenue. That’s a 4% net margin. Toyota runs 6-8%. By any conventional metric this is a mediocre auto business with a spectacular story attached.

It is also a company that has never operated a day without government support. The DOE loan that kept it alive in 2010. The ZEV credits that were the difference between profit and loss in 2013 — $130 million in regulatory credits against $61 million in losses on actual car sales. The $1.3 billion Nevada incentive package for the Gigafactory. The federal $7,500-per-vehicle tax credit that subsidized demand until Tesla hit its cap in 2019. Billions in government support across its first decade. Tesla is a free-market success story built entirely on government assistance — and now protected by a government tariff. The throughline is consistent even if the story isn’t.

This mediocre performance is despite having essentially a government-protected monopoly in its largest market.

Apply a realistic multiple to Tesla’s actual earnings and you get a company worth $150 to $300 billion — roughly $40 to $80 per share, not $430. The trillion-dollar gap between those numbers and the current price is the capitalized value of the tariff wall, plus a large bet on robots and autonomous vehicles that have been two years away for about a decade.

The math is not complicated. Toyota trades at roughly 8 times earnings. GM and Ford trade at 6 to 7 times. Apply even a generous 10x multiple to Tesla’s $3.8 billion in net income and you get $38 billion for the auto business. Add a premium for the energy storage segment and whatever you think the Supercharger network is worth as standalone infrastructure — call it another $30 to $50 billion. You are at $70 to $90 billion. Tesla’s current market cap is $1.4 trillion. The remaining $1.3 trillion is the bet on autonomy, robotics, and AI — products that do not yet exist at commercial scale, promised by a CEO running four other organizations simultaneously.

Consider what the market actually pays for real autonomous taxis. Waymo operates 500,000 paid robotaxi rides per week across more than ten American cities with fully driverless vehicles. It is not a promise. It is a functioning commercial operation. Waymo’s most recent funding round valued it at $126 billion. Tesla’s autonomous taxi program — nearly a year into its Austin launch, still operating a handful of vehicles — is not in the same league as Waymo.

What is it worth? Not likely anywhere near $126 billion.

This matters because the entire $1.3 trillion premium above Tesla’s asset value rests on future promises — and its decade-long record on self-driving cars clearly demonstrates what those promises are worth.

The Wall

We know what Tesla looks like in markets where the wall doesn’t exist: it’s losing, badly, and accelerating.

In Europe, Tesla registrations fell 40% year-over-year in mid-2025. In China — where Tesla actually manufactures and has to compete — domestic retail sales crashed 16% in Q1 2026, with March alone down 24%. Tesla’s market share in China has collapsed from around 10% in 2023 to roughly 5% today. Xiaomi, a company that made phones until two years ago, is now outselling the Model Y — Tesla’s best-selling vehicle globally — in China with a car it launched 18 months ago.

The competitor doing this damage is not cheating. The Rhodium Group analyzed BYD’s $4,700-per-vehicle cost advantage over Tesla and found that state subsidies account for 5% of it. Five percent. The other 95% comes from vertical integration — BYD manufactures 80% of its own components — scale, and lower overhead. These are structural industrial advantages built over two decades. No tariff eliminates them. A tariff only hides them, and charges Americans for the concealment.

The Man Running the Company vs. The Man Not Running It

While we’re comparing Tesla and BYD, compare who is actually showing up to work.

Wang Chuanfu founded BYD in 1995 as a battery manufacturer with borrowed money and a chemistry degree. He has run one company for thirty years. He is a trained engineer who still walks factory floors. In 2008 he made a public prediction: BYD would be the world’s largest EV company by 2025. The press mocked him. He delivered — on schedule, within the decade he said it would take. He runs no other companies. He manages no government departments. He builds cars.

Elon Musk is simultaneously the CEO of Tesla, was until recently the operational head of the Department of Government Efficiency inside the federal government, controls SpaceX (which absorbed xAI in February 2026 ahead of a planned IPO), owns X, and founded Neuralink. During a period of accelerating Tesla sales declines in early 2025, he was attending White House cabinet meetings. Tesla investors publicly begged him to return his attention to the company he nominally runs. He eventually promised, on an earnings call, to reduce his government role.

There is also the matter of the self-inflicted wound. Tesla was built on a customer base of environmentally conscious, left-leaning early adopters — precisely the demographic Musk spent 2024 and 2025 alienating with $300 million in Republican political donations, a prominent role cutting federal programs, and a sustained public persona that repelled the people most likely to buy his cars. A Yale University study found the “Musk partisan effect” cost Tesla between one million and 1.26 million US vehicle sales from 2022 to 2025 — a 67% to 83% boost to sales that never materialized. Democratic-leaning buyers walked away. Republican buyers did not replace them. In Germany, Tesla sales fell 59% in January 2025 alone. The CEO of a consumer brand went to war with his own customers and is surprised the numbers are bad.

What Are We Actually Protecting?

The tariff gives Tesla a captive market. High-end EV buyers pay thousands more than they would for a comparable BYD. People who need an affordable car simply go without. The 60,000 to 70,000 Tesla jobs being protected are a real interest — but BYD would employ Americans too if we invited them to build here instead of banning them. Toyota, Honda, Mercedes, and BMW all built American factories when pressured to do so. Hundreds of thousands of American jobs followed. We haven’t asked BYD for the same deal.

What we’ve done instead is wall off the US market to serve a collection of special interests — a protected company, its shareholders, and politicians who get to call themselves tough on China — while the people who actually need cheaper transportation pay the bill. It is difficult to identify the American interest being served here. It is easy to identify who benefits.

Without the wall, Tesla’s trajectory is toward irrelevance and eventual acquisition. The volume business collapses first, as Europe and China have already demonstrated. At some point the stock falls far enough that the assets — the Supercharger network, the factories, the IP, $44 billion in cash — get picked up by an acquirer for $50 to $100 billion. A 90% haircut from today’s price. The wall is not permanent. Politics shift, and BYD will find its way in regardless — through Mexico, through joint ventures, through tariff workarounds. The question is when, not if.