OpenAI chief warns idle PC RAM deepens global memory crunch

OpenAI chief Sam Altman warned on the Cheeky Pint podcast that keeping unused RAM in personal computers worsens the global memory shortage. Altman spoke with Stripe co‑founder John Collison, saying many assume local inference will solve AI needs, but in reality it ties up a share of the world’s limited DRAM for only five percent of the time. He argued that maximum utilisation should stay in cloud data centres to stretch the scarce supply. RAM is a core component of smartphones, laptops, tablets, car infotainment systems and game consoles. The market is dominated by a handful of makers, chiefly Samsung and SK Hynix, making the supply tight. AI firms have secured early access to new chips through long‑term contracts with cloud providers and projects such as OpenAI’s Stargate, which have earmarked a large portion of upcoming RAM output. The Wall Street Journal estimates OpenAI will spend about $750 billion on computing power through 2030. Similar expansion plans by Google, Meta and Amazon are driving up DRAM prices for consumer devices as well. Altman’s comments also touched on a split in Silicon Valley between open‑weight, self‑hosted models and cloud‑based proprietary large language models; Nvidia’s Jensen Huang and Meta’s Mark Zuckerberg have previously supported the open‑weight approach. During the same episode Altman admitted the pace of AI development is hard to explain and expressed mixed feelings about its impact. He highlighted OpenAI’s recent AI‑generated solution to the Navier‑Stokes problem and the release of results for nearly 400 long‑standing math problems, noting an internal model dubbed “Model Name‑Hold My Beer” was used. ## Context The DRAM crunch stems from a concentrated supplier base and soaring demand from AI data centres, pushing up costs for everyday gadgets. Keeping RAM idle in PCs adds pressure to an already strained supply chain.