What Would Buffett Pay for Micron?
A word on terminology. The piece below uses the term free cash flow throughout and readers unfamiliar with Buffet thinking may not be familiar with it. In general, free cash is money that you could return to the shareholders because the business does not need it now or in the future. This is not the same thing as profit.
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.