In remarks reported by The Verge, OpenAI CEO Sam Altman said at a Semafor forum on November 16, 2026, that people should be able to buy plenty of RAM for their computers. He also praised experimenting with PCs at home, while maintaining a conditional argument that shared memory in data centers could be used more continuously during a shortage.
The distinction is between access to RAM for personal computers and how scarce memory might be allocated across shared systems.
Why Altman made a conditional case for shared memory
Altman’s argument was about resource allocation when memory is in short supply: pooling it in data centers could, in theory, keep it in use more continuously. He also acknowledged a tradeoff: running AI locally on a person’s own hardware remains a reason to have memory in a personal computer.
The Verge reported that Altman described OpenAI’s infrastructure constraints as changing over time. Along with memory, he named rack assembly, data-center capacity and other components.
What OpenAI’s DRAM capacity target means
In October 2025, OpenAI announced strategic partnerships with Samsung Electronics and SK hynix. The companies planned an accelerated capacity rollout targeting up to 900,000 DRAM wafer starts per month. That figure is a planned target for wafer starts, not a report of production already achieved. OpenAI’s announcement