Altman says memory chips idling in personal computers are better used in AI data centers
OpenAI's CEO argues scarce DRAM chips should go to cloud data centers, not sit mostly unused in PCs, a view that touches device costs and local AI.
The short version
- Altman spoke in an interview with Cheeky Pint about local AI models and PC memory.
- His claim: DRAM spread across personal computers is mostly idle, which he called very bad for global resources.
- He says data centers should maximize use of memory while supply is extremely tight.
- OpenAI mainly serves closed models from large data centers, so the view serves its own interests.
- Memory demand from AI is also raising cost pressure on PCs and phones.
Why it matters
Memory chips are a shared, limited input, and how they are divided affects what laptops and phones cost. The remarks also show how a cloud-first AI company thinks about the rival idea of running AI on your own device. Treat it as a position from an interested buyer, not a neutral verdict.
What Altman said
As Sina Finance reported, Sam Altman was asked on the Cheeky Pint interview show about running AI models locally and what that means for PC memory. DRAM is the fast working memory inside a computer.
His point was about efficiency. If everyone ran a local AI model on their own machine, a large share of the world's limited DRAM would sit in individual computers and go largely unused. He called that a poor allocation, and said it is especially hard to defend when memory supply is extremely tight.
The logic and its limits
The argument is that a data center serves many people at once, so each chip is busy more of the time. Altman noted that many people feel differently, since owning a powerful machine seems natural.
It is also worth noting who is speaking. OpenAI sells its closed models mainly through large data centers, so a push to put memory there lines up with its business. Observers read the comments as part of a wider tug-of-war between cloud AI and local AI.
Why PC buyers are in the picture
AI demand has been pushing up memory chip prices, and the report says PCs, phones and other gadgets face higher cost pressure. The chips in PCs, phones and AI accelerator cards are of different types. Even so, making them competes for some of the same upstream chip-manufacturing capacity.
The report does not say prices will rise by any amount, and Altman did not announce any policy. His comments describe how he thinks resources should be used, not a decision anyone has made.
Other remarks from the interview
The interview covered more ground. Altman said he would rather see many AIs with slightly superhuman ability join society than one far more capable system, because cooperation and checks among them make the whole more resilient.
On AI agents, he said no product yet is one people could not imagine living without. He also said OpenAI considered charging per task, but customers wanted to see exactly what they pay for and preferred pricing by token, the small text chunks models process.
What to watch
- Does a fuller transcript or second report confirm the wording and context of the remarks?
- Do memory shortages actually show up in PC and phone prices in coming months?
Try it
If you plan to buy a laptop or phone, compare current prices at a few retailers instead of assuming a trend from one headline.
Sources and links
- ReportingSina Finance: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by Sina Finance · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.