Baron Opportunity's Micron Add: The Cycle That May Not Turn

Baron Opportunity Fund added $35.8 million to Micron Technology during the second quarter of 2026, its largest new buy by dollar amount. The letter's language is not the language of a cyclical trade. The fund's stated thesis: memory bandwidth is the binding constraint in AI inference, Micron is transitioning from a commodity memory maker to a defensible business, and the transition is being locked in through long-term strategic supply agreements with hyperscale customers and proprietary controller logic embedded in its memory products. That is a structural claim. The financials, so far, agree with it.

The numbers the old story can't explain

For thirty years, memory has behaved like a commodity. DRAM and NAND prices swing with supply and demand, and earnings follow. At the top of a cycle the stock looks cheap on trailing earnings because earnings are about to collapse; at the bottom it looks expensive because earnings are about to recover. Investors who try to time the cycle get burned in both directions, and the market prices memory stocks accordingly.

Now look at what the financials show. Trailing-twelve-month free cash flow — the cash a company generates after covering its operations and capital spending, measured over the last four reported quarters — was roughly $121 million at the end of Micron's fiscal 2024. In the most recent reported quarter it stands at $26.2 billion. Operating margin, the share of revenue that ends up as operating profit, moved from 5 percent to 66 percent. Gross margin moved from 22 percent to 73 percent. Revenue more than doubled in a single year.

A cyclical uptick raises margins by ten or fifteen points and then reverses. What these numbers show is a step change in the shape of the business, not a blip in an existing pattern.

The gap in the multiple

The market has noticed, and has moved the price to match. Micron is up 260 percent year to date and trades at 23x trailing earnings, meaning the stock is priced at 23 times the earnings it generated over the last twelve months. The forward multiple, based on next-twelve-months consensus, is 163x — the price divided by the earnings analysts expect over the next year. The gap between the two is the whole argument in one number: consensus expects next year's earnings to fall by roughly 86 percent from the trailing level. The market is still pricing the old story. The cycle is at the top, and the turn is coming.

The fund's thesis is that the turn is not coming. If the supply agreements with AI customers are genuinely multi-year and take-or-pay — a contract structure in which the buyer is obligated to pay even if it does not take the full volume — the revenue floor is set, and the earnings decline the consensus is pricing in does not arrive. If the controller logic and high-bandwidth memory architecture give Micron a proprietary edge that competitors cannot replicate on a short timeline, the margin floor is set as well. Under that scenario, the 163x forward multiple is not a warning; it is the market still underpricing a new regime.

What would break the case

The break condition is specific, and a retail investor can check it. A customer renegotiates or walks from a supply agreement. A new memory supplier comes online with comparable high-bandwidth memory capability and takes share. A single reported quarter shows operating margin compressing meaningfully below 60 percent, a signal that the pricing power is cyclical rather than structural. If any of those happens, the 163x forward multiple stops being a mispricing and starts being what it looks like: a warning that the cycle is turning.

For a reader with no position, the honest framing is this. The "story change" is not the opportunity anymore; the tape already shows it. What is left to evaluate is whether the story holds, which is a different risk from the one the fund's letter is selling. The fund is positioned to be early and public on the structural claim, with a research team that can read the supply agreements. A retail investor without that access has to watch two indicators instead: does the next reported quarter hold operating margin above 60 percent, and do the multi-year supply agreements remain intact without renegotiation? If both hold, the "cycle turns down" pricing is wrong. If either breaks, the "cycle turns down" pricing is right, and the stock is at risk. I can be wrong about this; the structural claim is a thesis, not a fact, and the next two reported quarters will tell us which way it leans.

The fund's other Q2 moves are reasonable portfolio rotations and do not carry the same thesis weight. It added Fervo Energy ($18.5 million, enhanced geothermal), Cerebras Systems ($11.2 million, the recent IPO, with a reported multi-year OpenAI agreement worth over $20 billion for 750 megawatts of inference capacity), Alphabet ($16.7 million, added to an existing position), and Welltower ($10.7 million, senior housing). It trimmed Datadog by $7.9 million after the stock rose 121 percent in the quarter, and it exited CoStar Group ($15.3 million sold), citing slow revenue traction in the Homes.com residential platform, and ServiceNow ($10.5 million sold), redeploying capital into other software names. None of those moves carries the same kind of structural claim. The Micron add is the one worth understanding, because the operating transformation it is betting on is already visible in the financials, and the question of whether that transformation is permanent or temporary is the question that determines whether the next twelve months look like the last twelve.