Sam Altman wants your RAM. It’s there in your computer doing nothing much, and it would be much better powering ChatGPT’s enormous RAM demands. After all, his company is a core part of the reason there is a RAM shortage to begin with, so it stands to reason that he’s dreaming of the RAM in your home, too.
So, it’s time to lock up your computers, because Altman is coming to town.
RAM Altman is coming to a PC near you
Memory is better in data centers
ChatGPT CEO Sam Altman initially made the comments while talking to Stripe CEO John Collison on his Cheeky Pint YouTube channel, where they were discussing issues with the memory supply chain, local AI inference, and the allocation of memory to regular folks.
From a global allocation perspective, actually, 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. Like, you want maximum utilization in the cloud if you’re trying to deal with this very supply constrained environment. And somehow many people’s intuition is the opposite of that.
It doesn’t read well; there are no two ways about it. Altman’s place as a leader of frontier AI design, presiding over one of the world’s largest artificial intelligence companies, gives special credence to his word. And his word is that you shouldn’t have RAM at home.
Understandably, Altman was slammed immediately after the video went live. AI companies like ChatGPT are directly responsible for the enormous rise in consumer memory prices — OpenMarkets says ChatGPT has bought over 40 percent of all capacity in 2026 — along with other PC hardware, so it’s beyond tone-deaf for Altman to even talk about memory, let alone suggest that the general public should help with better AI inference.
It’s all a big, jolly misunderstanding
At least, that’s what Altman is now claiming
As always, that’s not all there is to the story. Sam Altman changed his name to RAM Altman on X, pushing back against the claims that owning RAM privately is terrible in a series of posts after even Elon Musk piled on. Please, Sam, don’t make me agree with Elon.
Altman’s counter to his own statement is that it was really meant to be illustrative that the world should be building more RAM in general, claiming that “I actually do want the world to have tons of memory and people to put as much as they want in their computers, and it is f****** annoying that this was taken out of context.”
More RAM, for everyone, then? Not entirely. At least, not yet. When asked about general compute demand, Altman also made a broader point that stands up to closer reasoning than his previous comments about memory. Resources are scarce currently and compute costs in general are high, but as with most other big technological advances, it will change eventually as more comes online.
Like memory’s gotten really expensive, and so new memory makers are going to flood the market any moment now.
That’s the broader point Altman was trying to make, as much as he put his whole foot and almost leg in his mouth.
More RAM is great for consumers
But also pretty good for massive AI companies
It’s impossible not to look at Altman’s apparent 180 on his “using RAM at home” statement with a fairly decent pinch of skepticism, seasoned with a good twist of cynicism.
Altman isn’t wrong that new memory makers flooding the market would be really good. PC Part Picker’s memory charts painfully illustrate how bleak consumer memory is, with each DDR5 category rising by over 400 percent in 18 months. Read that again: 400+ percent in 18 months.
Prices dropping would be great news for everyone finding it painful to upgrade, replace hardware, or so on. But a drop in prices with new manufacturers entering the market would be sweet music to Altman’s ears, too. There isn’t an exact figure on how much RAM ChatGPT bought in 2026, but some figures quote “900,000 memory wafers per month”, translating to something like millions of terabytes per month.
And that’s just for ChatGPT.
Even if new-memory-maker on the block, CMXT (ChangXin Memory Technologies) could upgrade its fabrication process in time (and grab a huge exclusion from its current 50 percent US import tariffs), there is a very strong chance Altman’s ChatGPT and all of the other AI companies and hyperscale AI data centers already have some agreements in place for that capacity. Given the state of memory manufacturing, it would be insanity if they didn’t, even with the tariff premium accounted for.
He didn’t mean what he said if you ignore the things that he said to begin with
At the crux of what I think Altman really meant was that if all memory went to AI data centers and development companies and they leased it back to us as compute, it would be more efficient and cost-effective. They could reduce costs, add more capacity, and so on.
But that simply doesn’t escape the implications his statement creates, nor the giant consequence-colored elephant in the room with his RAMaggedon retractions, because RAM prices aren’t coming down any time soon, no matter what Altman thinks.
