Micron CEO Sanjay Mehrotra sees humanoid robots as next AI memory driver

Micron has been one of the best-performing stocks in the S&P 500 this year, rising more than 265%, largely because it is the only major U.S.-based manufacturer of high-bandwidth memory needed for AI. That data-center demand already shows up in its fiscal Q4 2026 results: revenue hit $54.23 billion, up 379% from a year ago, with net income of $37.7 billion. All four business units grew sharply, led by Core Data Center revenue at $18.0 billion, up 1,041% from $1.58 billion in Q4 2025.

Now CEO Sanjay Mehrotra is pointing to a second potential growth leg: physical AI. He cited autonomous vehicles as the benchmark, noting they require more than 200 gigabytes of memory and multiple terabytes of storage, and said humanoid robots should have comparable needs. Analyst Shay Boloor of Futurum Equities described adding 200 gigabytes of DRAM to humanoid robots as "effectively putting server-level memory into millions of machines outside the data center." Tesla's plans for its Optimus robots, including shutting down Model S and Model X lines to ramp up production, suggest this market may not be far off.

The stakes for Micron hinge on whether humanoid robot production actually scales and whether memory content per unit rises as Mehrotra anticipates. If it does, Micron's growth window could extend well beyond the current AI infrastructure build-out. But that outcome is uncertain, and the timeline — by the end of this decade, according to Mehrotra — leaves room for both technical and market delays. For now, the immediate driver remains data centers, as reflected in the company's strong Q1 fiscal 2027 guidance of $60 billion to $63 billion.