Sam Altman, CEO of OpenAI, has labeled the widespread practice of housing large amounts of DRAM memory in personal computers a “terrible” allocation of global resources. In a recent interview on the Cheeky Pint podcast, Altman argued that because local RAM remains idle roughly 95% of the time, these limited components would be more efficiently utilized in cloud-based data centers to support the growing demands of artificial intelligence. Speaking around the 28-minute mark of the program, the co-founder of the ChatGPT creator also discussed mathematical advancements and the application of AI in cybersecurity.
The Global Shortage of DRAM Components
The semiconductor market currently has a severe imbalance in supply and demand. The rapid expansion of the artificial intelligence sector is consuming a significant portion of the world’s silicon production. Because manufacturers are prioritizing high-demand enterprise and data center hardware, consumer electronics have faced increasing pressure.
This shift in production has led to tangible market consequences. The scarcity of these components has contributed to a global decline in PC and smartphone shipments by approximately 20%. The industry is seeing price hikes across the consumer spectrum; the PlayStation 5 Pro saw a price increase in March specifically tied to RAM shortages, while the secondary market for consoles has seen prices reach as high as $1,399. Within the PC gaming sector, high-end components like Nvidia’s RTX 5080 have disappeared from official retail channels due to supply constraints, following the near-discontinuation of the top-tier RTX 5090. The Xbox Series X|S and Nintendo Switch 2 have also been affected, making the current generation the only one to become more expensive during its lifespan.
The Shift Toward Cloud-Based AI Infrastructure
Altman’s critique centers on the economic logic of resource distribution. He contends that rather than dispersing limited DRAM supply across millions of individual PCs—where the hardware sits underutilized—the industry should prioritize maximizing cloud capacity. For Altman, it makes no sense to keep 16 GB, 32 GB, or 64 GB modules sitting idle on a desk. He noted that while many people’s intuition suggests the opposite, the corporate logic is that OpenAI can train and run large-scale models faster when more massive data centers operate at full capacity.
While Altman frames this as an efficiency issue, he clarified that he is not proposing a policy to seize or restrict personal hardware. Instead, his comments highlight a fundamental tension regarding who receives priority in the semiconductor supply chain. Manufacturers such as Samsung have confirmed that the diversion of DRAM production toward enterprise AI solutions is expected to keep the market under pressure for several years, with some projections suggesting the current supply tightness could persist until 2028 or 2030. Micron has further warned that the worst phase has not yet arrived.
Sam Altman Says It's "Quite Terrible" That Everyone's PC Holds Scarce DRAM.The OpenAI CEO told John Collison on the Cheeky Pint podcast that memory in personal computers gets used "5% of the time," and "you want maximum utilization in the cloud" amid tight supply. pic.twitter.com/iJge6TR36U
The Value of Local Computing Independence
Despite the arguments for cloud-centric resource allocation, the reliance on local hardware remains a point of contention for users. For gamers, enthusiasts, and those interested in running AI models locally, internal memory remains a vital tool. Higher memory capacity increases the possibility of running more demanding models directly on the hardware, depending on the video card and processor. Possessing hardware capable of executing complex tasks without relying on third-party cloud services provides users with distinct advantages in privacy, data sovereignty, and functional independence.
Altman on the Future of AI Intelligence
During the conversation, Altman shared his vision for the evolution of the technology. He stated that he would prefer a society with “a trillion AIs with slightly superhuman intelligence” rather than one that depends on “a single AI with a trillion IQ.”
Does Sam Altman Mean Consumers Must Give up RAM?
Does Sam Altman’s statement mean consumers will be forced to give up their RAM?
No. Altman’s remarks were an opinion on global resource allocation and market efficiency during a period of scarcity, not a proposal for any mandatory collection or restriction of consumer hardware.
Why is there a shortage of RAM for consumer devices?
The surge in demand for AI-specific infrastructure has caused semiconductor manufacturers to prioritize the production of components for large-scale data centers. This has led to reduced availability and increased costs for memory used in PCs, smartphones, and gaming consoles.
How long is the current memory shortage expected to last?
Industry forecasts differ on the timeline for relief. Samsung has suggested market pressure could continue until 2028, while SK Hynix has indicated that the supply constraints could persist as long as 2030.
Related reading