14 hours ago
3
The world runs on memory chips. Every AI model, every data center, every smartphone relies on DRAM and NAND flash memory to function. And according to Micron Technology’s CEO Sanjay Mehrotra, there simply aren’t enough of them to go around, a problem he doesn’t see resolving anytime soon.
Mehrotra indicated during Micron’s fiscal Q1 2026 earnings call that tight supply conditions for both DRAM and NAND will persist through and beyond 2026, with projections suggesting the crunch could stretch into 2027 or later. The culprit is straightforward: AI infrastructure is consuming memory at a pace that manufacturers can’t match.
A shortage that spans the entire industry
This isn’t just a Micron problem. SK Hynix, one of the three companies that dominate global memory production alongside Micron and Samsung, reported being sold out of memory products through 2026. When two of the three major suppliers are effectively tapped out, that’s not a bottleneck. That’s a structural deficit.
Micron executives have said that memory products tailored specifically for AI applications are fully booked.
AI workloads require significantly more advanced memory configurations than traditional computing. Large language models, image generation systems, and the training infrastructure behind them all demand high-bandwidth memory (HBM) and other specialized DRAM products that are harder to produce and yield less per wafer than standard memory chips. That production complexity means you can’t just flip a switch and double output. Even with aggressive investment, new fabrication capacity takes years to come online and ramp to full production.
Micron’s $100 billion bet on New York
Micron is investing $100 billion in a new megafab in Clay, New York, with groundbreaking set for January 2026. Additional facilities in Idaho are also part of the expansion plan.
Semiconductor fabs don’t produce chips the day you break ground. The typical timeline from construction start to volume production is three to five years. That means even this massive investment won’t meaningfully ease supply pressure until the late 2020s at the earliest.
What Wall Street is seeing
Analysts have responded to the supply outlook with notable optimism for memory stocks. Firms including J.P. Morgan and Morningstar have projected that supply challenges could extend into 2027, potentially kicking off a sustained period of pricing power for memory manufacturers.
Micron, Samsung, and SK Hynix collectively control the vast majority of global DRAM and NAND production. With all three facing similar capacity constraints, there’s limited risk of one player flooding the market and undercutting the others.
There’s also a geopolitical dimension that adds another layer of complexity. Memory manufacturing is concentrated in a small number of countries, primarily South Korea, Japan, and the US. Micron’s New York megafab is partly a response to government incentives designed to reshore semiconductor production.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) · 