Home » Sam Altman Calls Local AI a Waste of Memory Amid Global DRAM Shortage

Sam Altman Calls Local AI a Waste of Memory Amid Global DRAM Shortage

By

–

Sam Altman Calls Local AI a Waste of Memory Amid Global DRAM Shortage

Sam Altman has a view on your laptop’s RAM, and as ever it is a convenient one. Speaking on the Cheeky Pint podcast, the OpenAI chief said the world’s limited supply of DRAM is poorly spent sitting in personal computers that run AI models locally. Data centers, he argued, can keep the same chips busy all day.

“You want maximum utilization in the cloud if you’re trying to deal with this very supply-constrained environment,” he said. “And somehow, many people’s intuition is the opposite of that.” Local inference, in his telling, leaves memory idle. It is “quite terrible” from a global allocation perspective, he said, to have DRAM “used 5% of the time” in everybody’s machine.

Whose interests

The detail that matters is who benefits. Nvidia and Microsoft, both OpenAI partners, have been promoting new PCs with large memory pools for running models on the device. Local AI has a real privacy advantage. It also lets consumers and enterprises drop paid cloud AI, which is a direct threat to the business models of OpenAI and its peers. PCMag described Altman’s argument as obviously self-serving.

Altman says many people assume open-source models settle the matter, since they can be run at home and need less in the data center. His answer is about hardware allocation, not software.

The invisible hand, or not

The shortage is expected to last through 2030. Asked how the search for solutions was going, Altman answered, “Bad.” In theory, high demand should send companies rushing to build memory, the sort of incentive Adam Smith called the invisible hand. “I think the invisible hand is failing right now,” Altman said. “It has a checkered track record at best. But, I think it’s not quite working here.”

His explanation is that memory makers have been burned before. They must stay cautious about new fabrication plants, because overbuilding creates a glut. “I think they all remember their history,” he said. “How many times they almost went bankrupt by making too many fabs and being promised this time it was different.”

Meanwhile the squeeze is visible in shops. New PCs tuned for local AI, such as Nvidia’s RTX Spark laptops, are expensive enough that mainstream adoption looks unlikely until prices fall.

Sources

Photo: TechCrunch (CC BY 2.0)

Get breaking stories first on our Telegram channel.

More stories by

Most read

Consumer Tech & Goods

Pets

Gaming & Esports

Gaming