OpenAI Chief: RAM Better Used in Data Centres Than Home PCs
Sam Altman argues that scarce memory chips should power cloud AI, not sit idle in personal computers, amid soaring demand.

Illustration: AI-generated image
Sam Altman, CEO of OpenAI, has argued that scarce RAM chips would be better used in data centres than in home computers for running AI models, according to an interview with the YouTube channel Cheeky Pint reported by Index.hr.
He made the remarks amid a global shortage of memory chips driven by AI demand. Altman said personal computers often use only a fraction of their RAM for AI tasks, while data centre servers can run continuously for many users.
"From a global resource allocation standpoint, it's actually pretty terrible that everyone has a chunk of the world's limited DRAM in their computer and uses it maybe five percent of the time," Altman said, as reported by Index.hr. He added that when resources are constrained, the goal should be maximum utilisation in the cloud.
DRAM is used in PCs, servers, and other devices. Manufacturers like Samsung and SK Hynix have signed long-term deals with cloud providers, further tightening supply for consumer electronics.
Local AI running allows users to keep data private and work offline, but requires powerful hardware. Cloud-based AI eliminates the need for expensive equipment but ties users to subscription fees and provider policies.
Altman is not calling for a ban on local AI or everyday RAM use; his criticism focuses on efficiency for AI workloads. The discussion appears around the 29-minute mark of the interview titled "Why Sam Altman is against a single superintelligence".





