Micron warns of persistent RAM shortages through 2028 as profits surge
Micron said demand will outpace supply of DRAM and high-bandwidth memory through 2028, while reporting a sharp rise in revenue and margins.
Micron Technology told investors that tight supply of random-access memory is expected to continue into 2028 and beyond. The company’s chief executive and chairman, Sanjay Mehrotra, said most of the memory it will produce next year has already been sold and that customers will face "much higher prices" than this year.
"In calendar 2027 as well as 2028, we see demand exceeding supply. In fact, we see greater tightness in the industry in 2027 and in 2028 versus 2026. Overall, supply-demand environment is only getting tighter," Mehrotra said on the FY 2026 earnings call.
He added that Micron does not know when the market will return to balance. New fabrication facilities are slated to start operations in 2028, supported by about $25 billion in capital spending in the first half of the new financial year, but the company cautioned that these plants will not immediately ease shortages or lower prices.
Demand for high-bandwidth memory (HBM) used in artificial-intelligence hardware is growing faster than demand for standard DRAM used in servers. Micron reported that its cloud-memory unit, which focuses on HBM, posted a fourth-quarter gross margin of 83 percent, up from 59 percent a year earlier. The core datacenter unit, which sells more DRAM, recorded a fourth-quarter gross margin of 90 percent, up from 41 percent. Sales of solid-state drives for datacenters reached nearly $10 billion, a rise of more than 1,000 percent year over year.
Quarterly revenue climbed to $54.25 billion, and annual revenue reached $133.2 billion, far exceeding the previous year’s $37.4 billion. Net income surged 895 percent to $85 billion. The company forecast first-quarter FY27 revenue of $61.5 billion, plus or minus $1.5 billion, with an expected gross margin of 86.25 percent.
Mehrotra highlighted Micron’s roadmap for future HBM products, noting a partnership with NVIDIA on a custom HBM4E implementation for next-generation GPUs and NVLink platforms, and cited expanding opportunities for SSDs in AI-related storage.