Micron forecasts tighter memory supply in 2027 and 2028 than in 2026

The forecast comes from a company that sells memory ^[24], and its order book backs it up. Micron has committed more than 75% of its 2027 output, and most of its customer discussions are now about 2028, Mehrotra told analysts ^[11]. That leaves less than a quarter of 2027 production for buyers who have not yet signed ^[1]. Much of the committed supply sits in multiyear take-or-pay contracts that Micron calls strategic customer agreements, or SCAs ^[12]. "Any new discussions on SCAs where pricing is involved are negotiated with higher pricing based on prevailing market conditions and outlook," Mehrotra said ^[13].

Neil Shah, vice president of research at Counterpoint Research, still tells enterprises to sign ^[15]. "Companies should secure multiyear pricing for the computing capacity they know they will need," he said ^[16]. Moving workloads to the cloud will not avoid rising hardware and energy costs "because providers will pass them on," he added ^[17].

New factories will not change the outlook soon. "We do not have line of sight to when supply and demand will return to balance," Mehrotra said, even with new cleanroom space planned across the industry ^[2]. "Production from new DRAM and NAND fabrication facilities takes time to ramp and gradually becomes more meaningful starting a few quarters after initial output," he said ^[4]. Chief financial officer Mark Murphy said Micron's "inventory levels and supply remain extremely tight" ^[5].

The thing this doesn't tell you is how high prices will go. Mehrotra's outlook, as reported, describes supply-demand conditions and does not include a price forecast for 2027 or 2028 ^[1]. TrendForce sees a market that is still undersupplied, with price increases slowing ^[8]. Those two readings do not conflict. A smaller quarterly rise still lands on top of the increases buyers paid through 2026, when the shortage pushed up the cost of servers, storage and PCs ^[24].

On refreshes, Shah splits the fleet by workload. "For general back-office PCs and routine file servers, stretching lifecycles from three years to five is harmless," he said ^[18]. On a five-year cycle, roughly a fifth of a fleet is due for replacement in a given year. On a three-year cycle it is about a third ^[2]. "But for core infrastructure and engineering seats, delaying refreshes can backfire," Shah said, adding that aging equipment can drag on productivity, lose software support and fail more often ^[19].

His cheapest suggestion needs no purchase order. "Enterprises often waste 30% to 50% of memory by provisioning for peaks that rarely occur," he said ^[21]. The report does not say how that range was measured. An operations team can test it against its own utilization data before ordering more modules. Right-sizing virtual machines, quantizing AI models and batching workloads more efficiently can cut memory use significantly, according to Shah ^[22]. "CIOs should think about optimizing on the silicon already in place," he said ^[23].