Micron Technology reported fiscal fourth-quarter revenue that nearly quadrupled to $XX billion from $11.32 billion a year earlier, the company said in a statement, and issued fiscal first-quarter guidance of about $61.5 billion in revenue and $38.15 in adjusted earnings per share — figures that far exceeded analyst estimates.
The Boise, Idaho-based chipmaker's data center revenue jumped 11-fold compared with the prior-year period, driven by historic demand for high-bandwidth memory used in artificial intelligence systems. Micron is the only U.S.-based manufacturer of HBM, which is built from stacks of dynamic random-access memory, or DRAM.
The results sent Micron's stock, already up more than 500% over the past year, higher in after-hours trading. The worldwide supply crunch for memory chips has persisted as AI developers race to secure components for training and running large models.
CEO Sanjay Mehrotra told analysts on the earnings call that Micron has a "strong roadmap for future HBM products" and confirmed the company is working with Nvidia, the dominant supplier of AI accelerators. That relationship matters because Nvidia's chips require vast quantities of HBM, and supply constraints have limited how many units Nvidia can ship.
Micron's guidance implies continued acceleration. The $61.5 billion revenue forecast for the current quarter, if met, would represent growth of more than five times the $11.32 billion recorded in the year-ago quarter. Adjusted earnings of $38.15 per share would dwarf what the company earned in any prior period.
The memory market has historically been cyclical, with boom periods followed by steep downturns as manufacturers overbuild capacity. Micron and its competitors — South Korea's Samsung and SK Hynix — have so far been unable to produce HBM fast enough to meet demand, keeping prices elevated.
HBM differs from standard DRAM by stacking multiple memory dies vertically and connecting them with microscopic wires, allowing data to move faster between memory and processors. That speed is essential for AI workloads, where bottlenecks in data transfer can idle expensive GPU clusters.
Micron has not disclosed specific capacity expansion figures for HBM, but the company has said it is allocating a growing share of its DRAM wafer output to the high-margin product. Analysts have cautioned that adding HBM capacity reduces the supply of conventional DRAM, tightening that market as well.
Nvidia's demand for HBM has grown with each generation of its data center GPUs. The company's H100 and newer Blackwell chips each require multiple HBM stacks, and Nvidia has publicly acknowledged supply constraints on those components as a limiting factor on shipments.
Micron's fiscal fourth quarter ended in late August. The company's full-year results and detailed segment breakdown were included in the statement accompanying the earnings release.
