Sam Altman Says Having RAM Sitting Inside PCs Is 'Quite Terrible' Amid Global Memory Shortage
Last Updated: 2026-11-02 16:47:47
OpenAI CEO Sam Altman has sparked debate among technology enthusiasts after suggesting that keeping large amounts of computer memory inside personal computers is an inefficient use of resources when the world is facing a shortage of DRAM.
His comments come at a time when demand for memory chips is increasing as technology companies expand their artificial intelligence infrastructure. The growing requirement for high-performance computing hardware has put additional pressure on memory supply, raising concerns about the cost of PCs, smartphones and other consumer electronics.
Latest and Popular Mobiles
Altman made the remarks during an appearance on Cheeky Pint, a podcast hosted by Stripe co-founder John Collison. The discussion covered AI computing, local inference and the resources required to operate increasingly powerful artificial intelligence models.
During the conversation, Altman questioned the assumption that open-source AI models running on personal computers would significantly reduce the need for large data centres.
Related Articles
He argued that although consumers increasingly want to run AI models locally, doing so could result in substantial amounts of the world’s limited DRAM supply remaining underutilised in individual machines.
“A lot of people want local inference,” Altman said, discussing the possibility of running AI models directly on personal computers.
He subsequently argued that, from a global resource allocation perspective, it was “quite terrible” to have people keeping large amounts of the world’s limited DRAM supply in their computers when that memory might be used only a small percentage of the time.
Altman suggested that concentrating memory resources in cloud infrastructure could allow them to be utilised more efficiently.
The comments have attracted attention because they come from the head of a company that depends heavily on large-scale data centres to develop and operate AI services. OpenAI’s products, including ChatGPT, require substantial computing infrastructure to handle requests from millions of users.
However, the argument also raises questions about the future of local AI computing. Running AI models directly on personal computers allows users to process information without depending entirely on cloud services. It can also offer benefits involving privacy, offline access and control over computing resources.
For PC enthusiasts, developers and gamers, memory is an essential component that directly affects system performance. Modern computers increasingly require larger RAM capacities to handle demanding applications, multitasking, gaming and AI workloads.
A growing shortage of memory chips could make these systems more expensive, particularly if manufacturers prioritise supplying data-centre customers.
This is also relevant to the Indian consumer electronics market, where rising component costs could eventually influence laptop and smartphone prices.
Altman’s comments do not mean that consumers should stop buying computers with large memory capacities. Nor do they establish that personal computers are responsible for the current memory shortage.
Instead, they highlight a debate about how limited computing resources should be distributed between individual devices and centralised infrastructure as AI adoption expands.
The disagreement between local AI computing and cloud-based processing is likely to become more significant as AI models become more capable and memory demand continues to rise.