OpenAI CEO Sam Altman has argued that large amounts of RAM should be concentrated in data centers rather than distributed across users’ personal computers. He made the comments during an interview on the Cheeky Pint podcast while discussing supply-chain bottlenecks in the technology industry.
Altman specifically questioned the idea that open-source artificial intelligence models running locally could reduce the need for large-scale computing infrastructure. Although local models can reduce reliance on cloud services for some tasks, he said that distributing scarce memory across personal computers may not be the most efficient use of the available supply.
He argued that much of the RAM installed in consumer computers remains underused. In his view, memory in cloud infrastructure could achieve higher utilization, particularly amid tight supply conditions. Altman said that leaving limited global DRAM capacity in individual computers, where it may be used only a small fraction of the time, represents poor allocation from a global perspective.
The comments come as data centers compete for memory components needed by artificial intelligence infrastructure. Manufacturers including Samsung and SK Hynix have signed long-term agreements with cloud service operators and projects such as OpenAI’s Stargate Initiative, reserving a significant share of global memory production for the coming years.
That competition is also affecting consumers. Increased demand from large technology companies is putting pressure on the availability and prices of memory used in notebooks, smartphones and other electronic devices.
Altman’s position adds to the debate over local AI. Users and developers continue to explore running models directly on personal computers to reduce dependence on major cloud providers, while the OpenAI CEO favors concentrating computing resources in shared data-center infrastructure that can be used more intensively.
Sources: Olhar Digital.