Micron's stock is climbing fast. The memory chipmaker jumped nearly 8% Friday after CEO Sanjay Mehrotra told CNBC's Jim Cramer that AI-driven demand is hitting the company hard and fast. "We need more and more memory to address that demand," he said, confirming what investors have been betting on for months. The move signals that the worldwide memory shortage powering AI infrastructure buildouts isn't slowing down anytime soon.
Micron's Friday surge caps a remarkable month for the memory chipmaker. Shares have now climbed 56% over the last month, riding the wave of accelerating artificial intelligence infrastructure spending. The move came just hours after Taiwan Semiconductor Manufacturing Company reported robust earnings on Thursday, signaling continued enterprise AI buildout across cloud providers and tech giants.
What's driving the rally is a simple supply reality: memory is running out. Micron is one of the primary manufacturers of memory and storage for AI systems, and the shortage is real. "AI driven-demand is accelerating," CEO Sanjay Mehrotra told CNBC's Jim Cramer. "It is real. It is here, and we need more and more memory to address that demand."
The details behind that confidence are striking. When Micron entered 2025, the company expected server memory to grow around 10%. Instead, it finished the year in the "high teens," a meaningful miss on the upside. The company also saw stronger-than-expected growth in memory for PCs. That kind of divergence signals something has shifted in how aggressively companies are deploying AI infrastructure. Memory acts as the critical bridge between data storage and GPUs, letting large language models run without constant slowdowns. Without it, the entire AI stack bottlenecks.
That tightness is translating directly into pricing power. Memory chip prices are expected to rise an estimated 55% in the first quarter, according to CNBC's reporting. The shortage is being driven by an intense race to supply companies like Nvidia, AMD, and Google as they race to deploy more AI compute capacity to keep up with enterprise demand.
Mehrotra's messaging to investors went beyond short-term shortages. "We see that tightness continuing into 2027, so we see durable industry fundamentals over the foreseeable future, driven by AI demand," he said. That's the kind of multi-year visibility that investors live for, and it explains why the stock is trading at elevated valuations despite the semiconductor industry's traditional boom-bust cycles.
Backing that confidence is concrete capital deployment. Micron is spending $200 billion to build out new production capacity in the United States, including two fabrication plants in Idaho and a new 600,000-square-foot mega-fab in Clay, New York. The company broke ground on the New York facility Friday, with Commerce Secretary Howard Lutnick in attendance. That single fab will represent a $100 billion investment and will take a few years to complete as Micron builds out clean rooms and production equipment.
The timing matters. Building fabs isn't quick work, which is partly why memory remains constrained. By the time Micron's New York facility comes online, the company will have spent years trying to squeeze more production out of existing facilities. Mehrotra said Micron is already working to maximize output from current production lines, but there's only so much capacity can be wrung out of existing equipment. That's why the broader AI infrastructure buildout feels so durable right now. The supply constraints are structural, not cyclical. It'll take years of new fabrication capacity to catch up with where demand is heading.
What Micron's Friday move really signals is investor confidence that the AI infrastructure buildout isn't a short-term splurge. The memory shortage extending into 2027, combined with the company's massive capital commitments, suggests the industry has consensus on something larger happening: AI adoption at the enterprise level is real, durable, and hungry for infrastructure. When TSMC reports strong earnings and Micron's stock jumps on CEO commentary about multiyear demand visibility, that's when the market's treating AI infrastructure less like a speculative bet and more like a structural shift in how computing works.
