Micron’s CEO Sanjay Mehrotra projected that global demand for memory solutions will continue to outpace supply through at least 2027, driven by accelerating adoption of AI, cloud computing and 5G infrastructure. During the company’s recent investor presentation, he cited hyperscaler capital expenditure plans totaling more than $500 billion over the next three years, which he says underpin durable industry tightness. Mehrotra highlighted that enterprise and data-center customers have locked in multi-year contracts that guarantee volume growth of over 20 percent annually, signaling a multi-year upcycle in DRAM and NAND pricing.

To alleviate chronic shortages and secure its market position, Micron has committed approximately $200 billion for new fabrication facilities in the U.S., alongside a $24 billion expansion in Singapore and a $1.8 billion DRAM fab partnership in Taiwan. The U.S. build-out includes three leading-edge sites targeting HBM and LPDRAM production, with first silicon expected in late 2025 and ramp to full capacity by 2028. The Singapore project will add 200 000 wafer starts per month by 2027, while the Taiwan deal secures access to advanced process nodes for next-generation memory.

Micron highlighted that high-bandwidth memory (HBM) demand is compounding at over 25 percent annually and is on track to become a roughly $60 billion addressable market by 2028. Management noted that HBM now represents more than 15 percent of total revenue, up from 5 percent two years ago, and enjoys gross margins exceeding 60 percent. They expect HBM unit shipments to triple over the next three years as AI model training and inference workloads proliferate, solidifying memory as the key constraint in advanced computing deployments.