OpenAI CEO Sam Altman says the RAM sitting in your personal computer is "quite terrible" — and he'd rather you rent it from his cloud, where he controls the hardware, the data, and the rules.
This is the same company that just fired three researchers for raising safety concerns and is now facing questions from 20 House Democrats about whistleblower retaliation. The pattern is clear: Altman wants centralized control — over the chips, over the models, and over who gets to speak up when things go wrong.
Speaking on the podcast 'Cheeky Pint' on October 29, Altman argued that individual PCs waste the world's limited DRAM supply because personal RAM sits idle roughly 95% of the time. "It's quite terrible to have everybody have a bunch of the world's limited supply of DRAM sitting in their own computer and being used 5% of the time," he said. His solution: pool that memory in the cloud, where utilization can be maximized.
Altman dismissed the idea that rising memory prices will naturally increase supply and bring costs down. When asked about solutions, he answered: "Bad." He claimed DRAM manufacturers got burned for years — investing $20 billion in facilities only to make $19 billion back — and now, even as they pull in roughly $100 billion in annual free cash flow, they're too scarred by near-bankruptcies to expand capacity.
What Altman leaves out: AI data centers like his are the reason RAM prices are skyrocketing in the first place. As InvenGlobal reported, critics quickly pointed out that the cloud oligarchs who created the shortage are now telling consumers to surrender their local hardware and rent from them instead. The pushback was sharp — cars and kitchens sit unused most of the day too, and plenty of users run local AI specifically to keep their data out of corporate hands and avoid cloud rate limits.
InvenGlobal noted some defenders argued Altman was merely correcting a misconception about local models solving the memory shortage, not literally telling people to rip RAM from their PCs. But the practical effect of his argument is the same: you should depend on his cloud, not your own machine.
Meanwhile, the same day Altman's RAM comments went public, 20 House Democrats led by Rep. Greg Casar of Texas sent him a letter demanding answers about OpenAI's firing of three safety researchers — Tomek Korbak, Mikita Balesni, and Jasmine Wang. According to TNW, OpenAI checked the researchers' personal phones and found disappearing messages enabled in their Signal chats with a researcher at Redwood Research, an outside AI safety group. OpenAI claims they broke rules on sensitive information; the researchers say they never shared confidential data outside the company.
The lawmakers also flagged a prior security breach: two OpenAI employees had warned executives months before a Hugging Face incident that the company wasn't properly monitoring its newest models during testing. Executives replied that tests had to move fast so models could ship on time. OpenAI's August response to an earlier Casar letter didn't mention those warnings at all.
OpenAI published a blog post the same day as the Democrats' letter, claiming its expanded monitoring would now catch a Hugging Face-style breach roughly 30 hours earlier. The company said the alert would give responders time to interrupt the activity before models accessed external services.
Altman says energy and chip prices must come down and that it would be "very sad" if they don't. His plan: OpenAI will design its own chips, build its own data centers, and deploy robotics to construct them at scale. The man who thinks you shouldn't own your own RAM is building an empire to own all of it for you.
The question is whether Americans will accept a future where a handful of AI executives control the hardware, the models, the data, and the right to dissent — or whether local computing and local control remain non-negotiable.
