SAN FRANCISCO -- Micron Technology’s march toward a US$1 trillion valuation is nothing if not dramatic: a year ago it was a little over $100 billion.
That surge, though, was not built on its famed frugality, but on a nearly too-late push from Nvidia that pulled the U.S. memory chipmaker into the center of the AI boom.
For decades, the Idaho-based company survived by building factories on a shoestring budget, adopting used equipment and avoiding cutting-edge bets. That discipline helped it endure brutal boom-bust cycles in memory chips and outlast rivals, leaving it one of three global suppliers alongside South Korea’s Samsung Electronics and SK Hynix.
But that approach of treating memory chips as a commodity clashed with Nvidia’s vision for AI.
Three years ago, Nvidia CEO Jensen Huang met Micron boss Sanjay Mehrotra and outlined how he expected the memory market to evolve, Huang said in a media interview last month. Huang had long bet early that memory, and not just processors, would become a critical bottleneck for AI, forcing suppliers like Micron to rethink both technology and spending.
“I was really grateful that Micron and Nvidia really lined up all of our road map,” Huang said in the interview.
As Nvidia and other AI leaders rewired data centers, memory shifted from a commodity component to specialized high bandwidth memory (HBM) chips tailored to specific processors. These chips are co-designed for customers, making Micron’s offerings for Nvidia distinct from those it sells to Advanced Micro Devices or others.
Micron’s chips are now tightly integrated into AI systems, including Nvidia’s upcoming Vera Rubin platform. That alignment reshaped Micron’s trajectory, pulling it into long-term, higher-margin deals and giving investors greater confidence in its earnings.
Micron’s stock has surged roughly ten-fold over the past year. The company crossed the $1 trillion market capitalization on Tuesday, joining an elite group of trillion-dollar firms including Samsung. A day later, Hynix hit that mark.
AI growth reshapes Micron’s trajectory
Micron expects the HBM market it serves to grow to about $100 billion by 2028. Its posted a $14 billion profit in the latest quarter, a striking turnaround from the $5.8 billion loss it posted as recently as 2023, when the memory cycle turned and demand collapsed.
That rebound follows a misstep.
For years, memory was a commodity business, with customers such as Apple and Dell able to switch suppliers easily and drive down prices. That volatility made Micron wary of betting early on high-bandwidth memory, even as South Korean rivals pushed ahead.
Nvidia’s AI build-out forced a rethink. Micron said in March it had signed its first five-year supply agreement, a landmark shift for an industry long driven by short-term pricing swings.
Analysts expect Nvidia to be central to those arrangements, though neither company has confirmed it. “They are seeing long-term customer demand, with real commitment,” said Ben Bajarin of Creative Strategies. “That is the key driver getting them to spend money.”
Remaining nimble in a cyclical market
Micron’s old habits have not entirely disappeared and may still be an advantage, even as the company adapts them for a faster-moving market.
The challenge now is speed. Under Mehrotra, Micron has focused on shortening development cycles and quickly fixing production issues, a critical capability in AI where missing technical specifications can cost lucrative supply deals.
Micron’s position as the only major U.S.-based memory supplier also adds an edge as customers diversify away from Korea and governments push for domestic supply chains. But the real test will come when the cycle turns again.
Analysts expect AI to make the memory market structurally larger, but not immune to slowdowns. When that happens, Micron’s longstanding discipline — the same frugality that once held it back — could again set it apart.
“In the early days, nobody gave Micron a chance,” said Dan Hutcheson, vice chair of technology consulting firm TechInsights. “They’ve always had that back-against-the-wall attitude. If they lose that, like Intel lost it, they’ll die.”
(Reporting by Stephen Nellis in San Francisco and Anhata Rooprai in Bengaluru; Editing by Sayantani Ghosh and Nick Zieminski)
