Micron reported accelerating memory demand and earnings growth driven by AI HBM (high-bandwidth memory) manufacturing ramp-up.
Micron Technology delivered an earnings report that stunned Wall Street. Revenue for fiscal Q3 2026 reached $41.46 billion, a 345% increase year-over-year from $9.3 billion. Adjusted earnings per share came in at $25.11, crushing analyst expectations. The stock surged 15% on June 25 in response. Perhaps most striking: profits nearly tripled from the prior year—a 15-fold increase—while gross margins hit approximately 85%, a figure that would make most chipmakers envious.
The extraordinary numbers stem from a single dynamic: DRAM and high-bandwidth memory, or HBM, have become the critical constraint in AI infrastructure buildout. Data centers are racing to acquire these chips at any cost. HBM, which sits directly adjacent to AI processors, is essential—without sufficient supply, even the most powerful GPU becomes an expensive paperweight. Nvidia's processors require massive quantities of HBM to function, creating intense demand pressure.
Micron, Samsung, and SK Hynix control nearly all global DRAM and HBM supply, placing them in an extraordinarily favorable market position. Micron has reportedly sold out its entire HBM capacity through 2026, with supply tightness expected to persist into 2027. Analysts forecast that the three major memory makers will maintain pricing power for at least two additional years thanks to their consolidated market control.
To contextualize the acceleration, Micron's Q2 FY2026 results—reported March 18—showed revenue of $23.86 billion (a 196% year-over-year jump) with gross margins near 75%. The Q3 expansion from that already-elevated baseline is drawing comparisons to Nvidia's own trajectory. Micron crossed a $1 trillion market capitalization threshold in May 2026, reputedly one of the fastest such milestones in corporate history.
The parallel to Nvidia is instructive but imperfect. Nvidia expanded margins dramatically as data center demand exploded and maintained pricing power due to the absence of real competition in high-end AI training chips. Micron operates within a three-player oligopoly—Samsung and SK Hynix are substantial competitors—but the trio collectively commands enough supply that competitive dynamics resemble oligopolistic pricing rather than a free market.
Broader market dynamics reinforce the trajectory. Bitcoin miners have been pivoting toward AI and high-performance computing workloads for the past two years; companies like Core Scientific and Hut 8 have repositioned themselves as AI-adjacent data center operators. When memory chip demand reaches this intensity, it signals the AI infrastructure buildout is accelerating, not plateauing. DRAM and NAND supply tightness ripples into pricing for adjacent components, affecting cost structures for ASIC manufacturers and GPU-based operations.
A 345% revenue increase and 15-fold profit growth suggest the AI memory cycle remains in expansion phase. The consolidated market structure—three companies controlling virtually all supply—supports sustained pricing power as long as demand holds, but it also creates risk: any demand slowdown would hit all three players simultaneously.