OpenAI CEO Sam Altman has sparked controversy after suggesting that installing large amounts of RAM in personal computers is an inefficient use of the world's increasingly limited DRAM supply. Speaking during an October 29, 2026 episode of the Cheeky Pint podcast with Stripe cofounder John Collison, Altman argued that centralized cloud infrastructure could utilize memory resources more efficiently than individual PCs. His comments arrive as AI data centers consume an increasing share of global memory production, putting pressure on component availability and affordability for gamers, developers, and everyday computer users.
The discussion focused on local AI inference, where users run language models directly on personal hardware instead of relying on cloud services. Altman questioned whether distributing scarce memory across millions of individual devices was an efficient approach, suggesting that much of this capacity would remain unused for AI workloads.
"It's quite terrible to have everybody have a bunch of the world's limited supply of DRAM sitting in their own computer."
— Quote by: Sam Altman, OpenAI CEO
Altman estimated that memory installed in personal computers might be actively used for AI workloads only around 5% of the time, arguing that centralized infrastructure could achieve higher utilization by sharing computing resources between users. However, the estimate was not accompanied by supporting data, and it overlooks the broader purpose of system memory, which supports gaming, content creation, software development, professional applications, and everyday computing. RAM that is not being used for AI is not necessarily wasted, nor should its value be measured exclusively through the requirements of cloud computing.
The comments are particularly controversial given the ongoing memory shortage, which has been heavily influenced by the rapid expansion of AI infrastructure. As Samsung recently warned about worsening DRAM shortages through 2028, major technology companies are securing substantial manufacturing allocations through long term agreements, potentially reducing the flexibility of suppliers serving consumer and smaller business markets. Rising DDR5 prices and increasing demand for enterprise memory are creating additional challenges for PC builders and hardware manufacturers.
Altman also acknowledged that memory manufacturers have historically been reluctant to expand production aggressively because of the financial risks associated with oversupply. While these concerns are legitimate, the current situation raises a broader question about whether a small number of major AI customers should have disproportionate influence over access to critical computing components.
Cloud infrastructure can provide efficiency advantages for certain workloads, but it cannot replace the flexibility, privacy, independence, and long term value of owning a personal computer. Local AI, gaming, creative applications, and professional computing all benefit from accessible hardware that users can control without depending on subscriptions, internet connectivity, or centralized providers.
As AI infrastructure continues expanding, the industry may also face increasing pressure for greater transparency in memory supply agreements and appropriate market oversight to protect competition and prevent excessive concentration of scarce resources. These measures would need to balance legitimate enterprise demand with fair access for consumers, independent developers, and smaller businesses.
We strongly disagree with the idea that memory installed in personal computers represents a terrible use of limited resources. Personal computing and gaming helped build the hardware industry long before the current AI infrastructure boom, and consumers should not be treated as inefficient users of technology simply because their computers are not continuously processing AI workloads.
Owning a powerful computer is about more than utilization percentages. It represents independence, privacy, creativity, entertainment, and the freedom to choose how technology is used. A gamer purchasing 32GB or 64GB of DDR5, a content creator building a workstation, or an enthusiast running AI models locally should not have to justify that investment against the commercial priorities of trillion dollar technology companies.
The responsibility for solving memory shortages should not be shifted onto consumers while major AI corporations secure enormous manufacturing allocations. If limited production capacity and concentrated purchasing power begin undermining fair market access, governments and industry regulators should examine whether stronger competition safeguards, supply transparency, and consumer protections are necessary.
We support technological innovation and the advancement of AI, but not at the expense of affordable personal computing. The future of technology should allow everyone to own, upgrade, and control their hardware, not force consumers into a cloud dependent ecosystem because local computing has become prohibitively expensive.
Should AI companies be allowed to secure most of the available memory supply, or should regulators ensure that gamers, PC builders, and everyday consumers continue having fair access to affordable hardware?
