Micron says it's meeting about 60% of customer memory demand

Micron executives said this week that demand for AI memory is still outrunning what the company and the broader industry can supply. At a J.P. Morgan technology conference on May 20, Micron executive vice president Manish Bhatia said demand continued to outpace both Micron’s ability and the industry’s ability to supply, and said tightness in HBM, DRAM and NAND should continue well beyond calendar 2026. A May 22 post on X by the account LeopoldATracker said Micron CEO Sanjay Mehrotra had described the company as meeting about 60% of customer demand tied to inference, robotics, agents and autonomous systems. Reuters could not independently verify that exact figure from a full official transcript or webcast excerpt available on the open web, but Micron’s public conference remarks and recent earnings materials support the broader point that AI-led memory demand remains ahead of supply. (fiscal.ai) ### Where did the “60% of demand” claim come from? The May 22 X post attributed the figure to Sanjay Mehrotra and paired it with charts on AI memory demand. The social post framed the shortfall around HBM and DRAM, and linked the gap to inference workloads, robotics, software agents and autonomous systems. (fiscal.ai) Micron’s own investor site shows the company appeared at the J.P. Morgan Global Technology, Media and Communications Conference on May 20. The event page confirms the date and webcast, but the accessible event page does not include a full searchable transcript. ### What has Micron said publicly about supply constraints? (seekingalpha.com) Manish Bhatia said at the May 20 conference that “demand continues to outpace our ability and the industry’s ability to supply” and that tightness in HBM, DRAM and NAND would continue well beyond calendar 2026. In the same appearance, he said AI demand was shifting from human interactions toward agentic and machine-to-machine interactions, with inference becoming a bigger share of customer workloads. (investors.micron.com) Micron said on March 18 that fiscal second-quarter revenue reached $23.86 billion, up from $13.64 billion in the prior quarter and $8.05 billion a year earlier. Sanjay Mehrotra said then that the results reflected “a strong demand environment, tight industry supply, and our strong execution,” and the company said it expected significant records again in fiscal Q3. (fiscal.ai) ### Why do inference and robotics matter for memory demand? Micron executives have been drawing a distinction between training demand and the next wave of AI workloads. At the May 20 conference, Bhatia said agentic workloads were driving inference and making memory “an even more strategic asset” for customers. Earlier, coverage of Micron’s March earnings call cited company estimates that autonomous cars could need 300 gigabytes of DRAM or more, with robots requiring similar amounts. (investors.micron.com) That gives a concrete sense of why Micron is tying future demand not just to data-center training clusters but also to physical AI systems. (fiscal.ai) ### Which parts of the memory market are under the most pressure? HBM has been the most visible pinch point because it is used alongside AI accelerators, but Micron has said the pressure is broader. Bhatia said on May 20 that HBM, DRAM and NAND all remained tight, while Micron’s March earnings release said memory had become “a strategic asset” for customers in the AI era. (theregister.com) The company has also been expanding product lines aimed at AI infrastructure. Alphastreet’s Micron coverage lists May 5 shipments of its 245TB 6600 ION data-center SSD and a May 12 announcement on 256GB DDR5 server modules, both positioned for higher-capacity AI systems. ### What should investors watch next? Micron’s next scheduled catalyst is its fiscal third-quarter earnings report, typically released in late June. (fiscal.ai) Investors will be watching whether Sanjay Mehrotra and CFO Mark Murphy provide updated figures on HBM output, DRAM tightness and how much customer demand Micron says it can actually serve across AI inference and adjacent markets. (finance.yahoo.com) (news.alphastreet.com)