OpenAI chief executive Sam Altman has waded into the fiercest debate in consumer technology, arguing that keeping large amounts of memory inside personal computers is a poor use of a scarce global resource while artificial intelligence data centers go hungry. His remarks, made on a podcast that went live on October 29, landed just as memory prices are pushing up the cost of phones, laptops and game consoles.
Speaking on Cheeky Pint, the podcast hosted by Stripe co-founder John Collison, Altman pushed back on the idea that running open-source AI models on personal machines would ease pressure on data centers. "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," he said. "You want maximum utilization in the cloud if you're trying to deal with this very supply constrained environment."
An argument about utilization, not just supply
Altman's case rests on utilization rather than ownership. Memory in a data center serves many users through the day, he argued, while the RAM in a laptop or phone sits idle for long stretches. He described the assumption that local open-source models remove the need for data centers as backwards in a shortage.
The comments drew a sharp reaction from gamers and PC builders, many of whom noted that OpenAI itself is one of the largest buyers of the memory in question. Reports have tied OpenAI to supply arrangements with Samsung and SK Hynix that could account for up to 900,000 high-bandwidth DRAM wafers per month, roughly 40 percent of global output, as the company races to expand its data-center capacity. OpenAI has also moved into custom silicon, announcing an inference chip in August 2026 that it says delivers 1.5 to 1.9 times better performance per watt than reference systems.
Consumers are already paying for the squeeze
Whatever the merits of Altman's allocation argument, the shortage is already visible in shop prices. Memory costs have climbed so steeply this year that brands across categories have passed them on to buyers, and industry watchers warn that mobile-grade memory now costs several times what it did a year ago. Intel, reporting alongside the podcast's release, forecast fourth-quarter revenue of 15 to 16 billion dollars, short of the roughly 17 billion analysts expected, and said it expects the PC market to shrink 12 percent this year and another 10 percent in 2027, with no easing in memory supply or pricing next year.
HP has said memory made up about 35 percent of a laptop's production cost in July, up from 15 to 18 percent a quarter earlier. On the other side of the trade, memory makers are posting record results: SK Hynix reported a record quarterly operating profit, said more than half of its 2027 and 2028 production is already booked, and warned the shortage could last through 2030.
A shortage with no quick fix
Altman conceded that the market has not corrected itself, saying "the invisible hand is failing right now," and said he wants far more supply in the market. He pointed to OpenAI's own chip work, its data-center building efforts and plans to use robotics to expand manufacturing capacity as longer-term answers.
For now, the tension is unlikely to ease: consumers want capable machines of their own, AI companies want every chip they can secure, and demand still outruns supply. Altman's intervention has made plain how the world's biggest AI company sees that trade-off — and why the price of RAM is now a mainstream economic story.



