Sam Altman addresses RAM comments: people should ‘put as much as they want in their computers’
Nov 4, 2026, 6:46 AM · The Verge
Sam Altman says his RAM remark was taken out of context, but the joke on X doesn't change the fact that AI data centers are soaking up the world's memory supply.
Why it matters
On a podcast, OpenAI's chief executive said it was "quite terrible" for so much of the world's limited DRAM to sit in personal computers used a small fraction of the time. That landed badly, because ordinary buyers are already paying more for hardware while AI data centers buy up memory.
On election night he tried to walk it back on X, changing his display name to a RAM pun and saying he wants the world to have tons of memory and has long argued for ramping up production. The shortage itself isn't going anywhere fast: The Verge notes Intel expects it to last through at least next year, and SK Hynix has suggested it could run until 2030.
From the desk
We'll take Altman at his word that he wants more memory built. The trouble is that the original comment revealed a worldview worth naming. Framing the RAM in people's own machines as wasteful idle capacity treats personal computing as an inefficiency to be optimized away in favor of centralized compute. That's the logic of a company whose business depends on renting out cycles in its data centers.
The walk-back doesn't really engage with that. Calling for more to be built is fine as far as it goes, but it skips the near-term reality. Memory makers can't conjure new capacity overnight, and while they catch up, the biggest buyers set the terms. Right now those buyers are AI data centers, and the people who feel it are ordinary buyers facing tech price hikes.
There is a fair argument that utilization matters and that shared compute can be more efficient. We don't dismiss that. But local hardware is also about ownership, privacy and resilience, and a lot of useful AI is moving onto devices precisely because people want it to run on machines they control. Undercutting that, even rhetorically, cuts against a healthy future for the technology.
I'm watching whether the industry's appetite gets matched by actual supply investment, and whether anyone at the top of the AI business treats consumer pain from the shortage as their problem rather than a side effect.
Context
The original remark came on the Cheeky Pint podcast. The Verge's report ties the shortage to AI data center demand and cites forecasts from Intel and SK Hynix on how long it could last.
Who feels it
- PC buyers and gamers
- Higher prices and tighter supply for memory look set to persist, regardless of what executives say on social media.
- Memory makers
- Pressure to expand production grows, but so does the leverage of the largest AI customers.
- OpenAI
- A self-inflicted reputational hit that sharpens the perception that AI growth comes at consumers' expense.
What to watch
- Announcements of new DRAM production capacity and when it actually comes online.
- Retail price trends for RAM and PCs over the coming quarters.
- Whether AI companies make concrete commitments on memory supply rather than posts.
- How on-device AI plans hold up if memory stays scarce and expensive.
Companies: OpenAI