Micron

Micron (NASDAQ: MU) makes DRAM and NAND — the memory a computer holds data in — and it is the one thing its scale peers are not: American. SK Hynix and Samsung, the other two names that matter, are Korean. So Micron is the accessible way onto the fourth wall of the AI buildout: after money, after power, after the chips themselves, the binding constraint that almost nobody had priced is memory, and Micron is the only US-listed pure-play on it. See the economics of AI compute for why HBM became the crunch, and the AI-capex loop for the spending it feeds.

The number that named the moment: Micron reported third-quarter revenue up 346% to $41.46 billion, above the $35.84 billion estimate — a single quarter larger than all of fiscal 2025 ($37.38 billion). In May 2026 the stock crossed a $1 trillion market value for the first time, weeks after it had passed $700 billion. This is DeadRisk’s second-order tier: a name the AI trade does not create, but reaches for when the input it depends on turns scarce.

Memory is a cyclical business, and the record remembers. TEXXR’s archive carries Micron back to 2015, and its coverage runs in the sawtooth memory has always run: quiet years broken by a spike, then a collapse. The last spike before this one came in 2023 — a glut, when memory makers were losing money and coverage jumped to 17 articles in a quarter, then fell back to two or three. The corpus holds that boom-bust in the same file as what is happening now, which is the whole tension of the name.

Because what is happening now reads as a break from the cycle, not another turn of it. AI’s appetite for high-bandwidth memory made the scarce part scarce, and the pricing power flipped to the makers: the Wall Street Journal ran the line that AI had made memory chips more valuable than oil, with the makers using their newfound leverage to lock in long-term agreements. That is a commodity behaving like a cornered resource, which commodities are not supposed to do.

Micron is spending as if it believes the break. It raised its US capital commitment to $250 billion through 2035, adding $50 billion for New York, Idaho and Virginia; it broke ground on a ~$9.3 billion expansion of its Hiroshima plant. Its constraint has moved from demand to its own capacity to build — the same shape The AI Capex Supercycle tracks one layer up.

But the tape has not made up its mind. Micron and Sandisk led a sharp selloff on June 23, each down more than 13% in what one outlet called a gut-check moment for AI stocks; in July, memory and storage makers fell sharply again. A stock that swings 13% in a session is a stock the market cannot decide is a supercycle or a top.

Two more forces sit in the record. Supply is coming: SK Hynix’s US listing seeks to raise ~$29.4 billion, and China’s CXMT is in the mix — memory vendors pushed back on the cheap-DDR5 claim, saying its prices match theirs. And politics is now inside the price: the chip group SEMI, Micron and Samsung among it, warned Washington against moves that would distort the memory market. New supply and policy are exactly the two things that end a memory cycle.

323 hyperedges touch Micron — a deep record for a company the tech press ignored for years, and the shape of it is the finding. The recent edges have turned. Where Micron’s older record was product — NAND densities, SSD launches — the 2026 cluster is capacity and policy: the $250 billion US capex commitment, the Hiroshima groundbreaking, a plan to ship HBM from summer 2028, and a regulatory edge for the SEMI memory-market warning. A memory company’s edges reading like a utility’s — capex, factories, policy — is the record registering the moment the commodity became strategic.

Positioning: how much of MU is sold short, on the record brokers file twice a month. It lags by design — read it as where the crowd stood, not where it stands.

Signals in the coverage and filing record, not trade ideas:

• Whether the long-term contracts hold. The bull case is that AI turned memory from a spot commodity into a contracted input. The bear case is that memory has mean-reverted every cycle in its history. The contracts the makers are locking in now are the test — whether they survive the next wave of supply, or melt the way memory prices always have.
• The HBM ramp lag. Micron says HBM shipments start summer 2028. The gap between demand today and supply years out is where a cycle either sustains (demand outlasts the wait) or breaks (supply floods in late).
• The China undercut. CXMT matching DDR5 prices is the classic cycle-ender — new low-cost capacity. Whether that stays a DDR5 story or reaches HBM is the structural question.
• Peer supply coming online. SK Hynix’s US listing and Samsung’s expansions add capacity to the same market. More supply is what has always broken memory pricing; watch whether demand absorbs it.
• Coverage against the cycle. The corpus caught the 2023 glut spike and its collapse. Whether Micron’s coverage stays elevated or falls back the way it did in 2024 is a read on whether this time is structural, tracked quarter by quarter.

Micron lives or dies by the same constraint as these names — small equal-weighted groups, one constraint each. Not investment products: see all categories.