Sam Altman says personal computers should not hold DRAM ‘used 5% of the time’

OpenAI CEO Sam Altman has sparked another debate over the growing demand for artificial intelligence infrastructure after questioning why people keep large amounts of computer memory in their personal devices.

In an interview, Altman argued that limited supplies of Dynamic Random Access Memory (DRAM) should be used more efficiently. He said it was “quite terrible” from a global allocation perspective for people to own memory in their computers when it is being used only around 5 per cent of the time. His remarks have drawn attention because DRAM is an essential component of everyday computers, while demand from AI data centres is putting increasing pressure on memory supplies.

Why Does Sam Altman Want DRAM In Data Centres?

DRAM is a type of computer memory that temporarily stores data required by a processor. It helps computers run applications, manage multiple tasks and process information.

AI data centres require substantial amounts of high-performance memory to operate advanced models. Altman's argument is based on the idea that resources sitting idle in individual computers could be put to greater use if they were concentrated in shared computing infrastructure. He said that in a supply-constrained environment, “you want maximum utilisation”. However, personal computers serve a different purpose from data centres. Their memory is available to users whenever they need it, even if it is not being used continuously. The debate therefore raises a wider question: should efficiency across the computing industry take priority over individual ownership and access?

From AI As A Utility To DRAM: Altman's Earlier Controversies

This is not the first time Altman's comments about AI and technology have attracted criticism. He previously described a future in which intelligence could become a utility, similar to electricity or water, with people paying for it based on usage. Critics raised concerns about the concentration of control over an increasingly important technology. He also faced criticism over comparisons between the resources needed to train AI models and those involved in raising a child, as well as remarks suggesting society would have to accept some negative consequences to allow AI innovation to progress. In early 2026, a post expressing gratitude to software developers who once wrote code character by character also drew criticism amid concerns about AI-driven disruption to software jobs.

What Does This Mean For PC Users?

Altman's comments do not amount to a proposal to restrict personal computer ownership or take memory away from consumers. They reflect an argument about how limited computing resources should be allocated. For PC users, however, the issue is relevant. AI infrastructure is competing for resources that also support consumer electronics, servers and other computing products. The larger question is whether the rapid expansion of AI data centres will make memory and other hardware resources more expensive or harder to obtain for ordinary buyers.