Micron Technology just reported a quarter that would have been unthinkable three years ago. Revenue of $54.23 billion in three months, up 379 percent from a year earlier. Gross margin of 87 percent. Net income of $37.70 billion. For the full fiscal year, revenue reached $133.19 billion, up 256 percent, with $84.97 billion in net income. These are the largest numbers the memory industry has ever printed, as Tech Insider documented.
The stock did nothing.
That disconnect is the story. Micron's fiscal fourth quarter was its sixth consecutive record quarter. The company generated more revenue in 13 weeks than it did in all of fiscal 2025 combined, when full-year sales were $37.38 billion. Sequentially, revenue climbed 31 percent from $41.46 billion in the prior quarter. DRAM brought in $39.8 billion, about 73 percent of the total, while NAND contributed $14.1 billion. And yet shares barely budged on the print, because every one of those numbers was already in the price.
Pricing, Not Volume, Did the Work
The mechanics of the quarter matter more than the headline. Growth was led nearly entirely by pricing, not by shipping more units. Gross margin expanded by more than 40 percentage points year over year to 87 percent, which means the incremental revenue from higher prices flowed almost straight to profit rather than being absorbed by higher production costs. That is a meaningfully different story than the demand-volume growth chipmakers usually tell.
The pricing power comes from high-bandwidth memory, or HBM. Each AI GPU deployed in a hyperscaler data center requires multiple HBM stacks, manufactured in three-dimensional layers using through-silicon vias. Only three companies in the world have commercialized the process at scale: Micron, SK Hynix, and Samsung. With Micron's 2026 HBM supply sold out under long-term agreements at locked-in prices, the revenue is not speculative. It is contracted.
That contracted visibility is both the bull case and the bear case. On one hand, Micron's data center revenues exceeded $25 billion on an annualized run-rate basis, with the Cloud Memory Business Unit generating $13.77 billion in quarterly revenue, according to TechTimes' earnings coverage. CEO Sanjay Mehrotra described the results as reflecting the strategic value of memory in the AI era. On the other hand, when your entire growth story is pricing on a single product category, the market starts asking what happens when supply catches up.
The competitive picture adds another wrinkle. SK Hynix and Samsung are not standing still. Both are racing to expand HBM capacity, and each new fab that comes online chips away at the scarcity premium Micron currently enjoys. The three-player oligopoly has held pricing discipline so far, partly because the manufacturing is genuinely difficult and partly because all three are sold out. But oligopolies in semiconductors have a poor track record of surviving demand plateaus. The last memory supercycle ended with a 40 percent price collapse that took two years to work through. Nobody at Micron will say when this one ends. Nobody can.
There is also the customer side of the equation to consider. Hyperscalers are paying these prices because they have no choice, but they are not happy about it. Every dollar of HBM margin Micron captures is a dollar of data center cost for Google, Microsoft, Amazon, and Meta. Those customers are already designing their own silicon and exploring alternative memory architectures to reduce dependence on the merchant market. The pricing power is real today. It is also the incentive for its own destruction tomorrow.
The Bar Keeps Moving
Micron guided fiscal Q1 2027 revenue to about $61.5 billion at roughly 86 percent gross margins, above Wall Street expectations, signaling the momentum has not abated. But guidance that would have seemed absurd a year ago now reads as merely meeting expectations. That is the trap of a sixth consecutive record quarter. Each print resets the bar, and the bar is now at a height where only perfection clears it.
There is a broader warning buried in the numbers. Goldman Sachs analyst Ioannis Blekos flagged that paper gains inflated S&P 500 earnings last quarter, and that ex-paper-gains EPS growth told a fundamentally different story about the underlying cycle. Micron's quarter is the opposite of a paper gain. It is real cash, real pricing, real contracted demand. The question is whether real can stay this good for a seventh quarter, and an eighth.
The memory industry has a long history of punishing exactly this kind of confidence. Revenue swings of 30 to 40 percent in a single year were historically considered extreme. Micron just did 379 percent. The cycle has not been repealed. It has just been stretched by AI demand into a shape nobody recognizes. When it snaps back, the fall will be measured from a much greater height.
Topics technologysemiconductorsmemoryai
