Sam Altman Suggests Running AI Locally Is a Waste of Memory - PCMag UK

(Photo by Heather Diehl/Getty Images) It’s not exactly a surprise, but OpenAI CEO Sam Altman isn’t a fan of people running AI models locally because it can pull precious memory supplies away from data centers. “You want maximum utilization in the cloud if you’re trying to deal with this very supply-constrained environment. And somehow, many people’s intuition is the opposite of that,” Altman said in a new interview on the Cheeky Pint podcast. In contrast, OpenAI partners, including Nvidia and Microsoft, have been touting the benefits of running AI models on new PC hardware with high memory capacity. Local AI offers better privacy, but it also means consumers and enterprises can ditch cloud-based AI without paying, making the business model a direct threat to OpenAI and other top providers. "A lot of people want local inference . And somehow, they assume that the problem is solved now because we are going to have open-source models," Altman said. "We’re going to be able to run them on our computers. And so we’re not going to need all the stuff in the data center. But from a global allocation perspective, actually, it’s quite terrible to have everybody have a bunch of the world’s limited supply of DRAM sitting in their own computer. And be used 5% of the time.” Hence, it seems like Altman views the local AI trend as a waste of precious DRAM, although it's obviously a self-serving argument. He made the comment while discussing the worsening memory shortage, which is expected to persist through 2030. When asked, “How are we doing on the solutions front?” he responded, “Bad.” Recommended by Our Editors To Stop Bots, Reddit Will Require All Visitors to Log In Google's AI Mode Suddenly Can't Recall Who Won the 2020 Presidential Election With Gemini 4 Carbon, Google Touts Big Leap in Cybersecurity Capabilities The demand should cause a wave of enterprising companies to focus on building memory, an incentive that economist Adam Smith described as the “ invisible hand ,” which can improve market conditions for all. However, Altman said: “I think the invisible hand is failing right now," adding, "It has a checkered track record at best. But, I think it’s not quite working here." Altman then pointed out the memory industry needs to be conservative with building new manufacturing capacity, rather than go all out to meet demand; if they over-invest, they risk creating a glut, which could leave them with an oversupply of chips. “I think they all remember their history. How many times they almost went bankrupt by making too many fabs and being promised this time it was different,” he said. In the meantime, the memory shortage has been so severe that new PCs optimized for local AI, such as Nvidia's RTX Spark laptops , are very expensive and likely won't see mainstream adoption until prices come down. Disclosure: Ziff Davis, PCMag's parent company, filed a lawsuit against OpenAI in April 2025, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.