Micron just became a trillion-dollar company, and Michael Burry immediately bet against it

Micron was the stock nobody wanted to hold through earnings season, a boom-bust memory chipmaker that traders bought for a quarter and sold before the cycle turned on them. Not anymore. This is now a company whose product is sold out through 2027, whose stock crossed a trillion dollars in market cap in May, and whose earnings this year triggered a same-week pile-on from a class-action lawsuit and Michael Burry’s short book. That’s why we built Winvesta Crisps, to decode what’s actually driving the companies you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

🔔 Don’t miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Most investors still remember Micron as the stock that ate your patience: a commodity chipmaker whose price rose and fell with PC and phone demand, never holding a rally for long. That memory is now roughly two years out of date. Micron’s fiscal third-quarter revenue came in at $41.46 billion, up about 346% from a year earlier, and the company told investors its high-bandwidth memory chips, the specific product every AI accelerator needs, are booked solid through 2027 with demand stretching into 2028. Three days after that print, a class-action lawsuit accused Micron, Samsung, and SK Hynix of colluding to restrict supply. Two days after that, Michael Burry disclosed a short position, calling Micron a “destroyer of capital.” The stock has lost around a fifth of its value since its late-June peak while still sitting on a gain of roughly 700% over the past year. Both things are true at once, and that tension is the entire story right now.

---

🏭 Not one company, but three memory businesses under one roof

Micron’s headline revenue number hides how differently its three product lines actually behave. DRAM is the largest by far, bringing in $31.3 billion in the fiscal third quarter, about 76% of total revenue, and it is the category where AI demand has bent pricing the hardest. NAND flash storage contributed $9.9 billion, the remaining 24%, and it has ridden the same wave, though at a smaller scale. Sitting inside DRAM is the product line actually driving the re-rating: high-bandwidth memory, or HBM, the stacked chip that sits directly beside every AI accelerator and determines how fast that accelerator can actually run. Data-centre revenue across all of Micron’s product lines reached $25 billion in the quarter, with enterprise SSDs alone contributing $5 billion, about 20% of that figure, per the company’s earnings call.

Micron is one of just three companies in the world that make this category of memory at scale, alongside Samsung and SK Hynix, both of which are based in South Korea. That concentration matters because none of the three can meaningfully expand supply overnight. CEO Sanjay Mehrotra has said Micron can currently fulfil only 50% to two-thirds of customer demand for its products in the medium term, a structural, not seasonal, supply gap. To close it, Micron has broken ground on a $9.3 billion expansion of its Hiroshima, Japan facility, formally started construction on a New York DRAM megafab that could eventually represent up to $100 billion in investment over the next two decades, and completed a $1.8 billion acquisition of a Powerchip fab site in Taiwan. For Indian investors, there’s a direct local thread here too: Micron opened its first Indian assembly and test facility in Sanand, Gujarat, in February, a $2.75 billion combined investment with the Indian government that is expected to assemble and test tens of millions of chips this year, scaling to hundreds of millions in 2027.

---

⚡ The AI supercycle bet: why HBM is sold out through 2027

The mechanism behind Micron’s numbers is simple to state and hard to unwind quickly. Every large AI accelerator, the kind Nvidia and its rivals ship into hyperscaler data centres, needs HBM sitting next to the processor to feed it data fast enough. As AI training and inference demand has scaled, HBM capacity has become the bottleneck, not raw chip supply. Manufacturing capacity for AI-optimised memory runs three to four times that of traditional computing products, and the three companies that make it have not chased that demand with reckless capacity additions, as memory makers historically did during past booms.

Micron’s response has been to lock in customers rather than just chase spot pricing. The company has signed 16 Strategic Customer Agreements, deals with data-centre operators and automakers that fix pricing and volume for periods of three to five years, and management expects roughly half or more of company revenue to eventually sit under these contracts. Those agreements already carry about $22 billion in financial commitments, including nearly $18 billion in cash deposits, according to the company’s earnings call. On July 1, Micron added General Motors to that list, a supply agreement covering automotive memory and storage for future vehicles, a signal that HBM demand is starting to spread beyond hyperscalers and into industries that simply need more computing power embedded in their products. HBM4, the next generation of the product, began shipping in the first half of calendar 2026 for Nvidia’s Vera Rubin platform, and Micron says it is reaching mature yields on that generation meaningfully faster than it did with the prior one. Industry estimates now put the total addressable market for HBM above $100 billion in 2027, up from a fraction of that just a few years ago.

Want to add Micron to your portfolio? Trade $MU directly from India on the Winvesta app. No US bank account needed!

🚀 Join 60,000+ investors, become a paying subscriber or download the Winvesta app and fund your account to get insights like this for free!