Micron Tops $1 Trillion; Memory Joins AI Bellwether

Micron Technology, founded in a Boise dental-office basement in 1978, reached a $1 trillion market capitalization after reporting quarterly revenue that rose 346% and gross margins of 84.9%. The company reported net income of $28.2 billion for the quarter. Micron’s shares have risen about 700% over the past year, with roughly 200 percentage points of that gain occurring in 2026.

The results reflect sharp increases in prices for high-bandwidth memory, or HBM, used in AI servers. HBM sits close to processors and helps move large data sets during model training and inference. Only three companies — Micron, Samsung and SK Hynix — manufacture HBM at scale, limiting supply.

Micron says it expects the HBM market it serves to expand to about $100 billion by 2028. The company is pursuing long-term supply contracts and is investing about $200 billion in manufacturing and research and development, including new fabrication plants in Boise, Idaho, and Syracuse, New York.

Nvidia reported revenue of $81.6 billion for its most recent quarter, an 85% increase from a year earlier, and net income of $58.3 billion. Investors had previously treated Nvidia’s results as a gauge of AI demand.

Analysts note risks tied to memory’s historical cycles. Analyst David Jagielski warned that if demand slows or production capacity expands to meet current prices, memory prices and Micron’s valuation could decline. Memory markets have in the past moved from tight supply to oversupply when capacity ramped up after price spikes.

Competition at the processor level is expanding as some cloud and AI firms develop custom chips and other suppliers secure cloud contracts. Producing next-generation HBM requires long lead times and complex manufacturing steps, the company says.

Market participants are watching Micron’s contract wins, capacity expansions and HBM price trends for signs of whether current margins and revenue growth persist. Sales and margin figures from Micron and Nvidia are being tracked as data points on AI infrastructure demand.