AI Already Made RAM Pricier; OpenAI’s Sam Altman Says Windows PCs Use It Just 5% of the Time

OpenAI CEO Sam Altman says it is “quite terrible” for everyone to keep the world’s limited DRAM in their own PCs, where it is used only 5% of the time, when AI data centers could put it to better use. After a jab from Elon Musk and a new X name—Ram Altman, naturally—he now says his remarks were taken out of context.

Meanwhile, MS, OpenAI’s longtime partner, has been promoting the opposite approach. Satya Nadella wants “unmetered intelligence” on every desk, with AI models running on PCs and accessing the cloud when needed. Both CEOs want AI everywhere, but they disagree about where it should run and where the world’s RAM should go.

Why Sam Altman Thinks Your PC’s RAM Is Better Used in AI Data Centers

Altman made the comment during the October 29 episode of Cheeky Pint, hosted by Stripe president John Collison. The two were discussing why the memory shortage was not resolving itself. Shortly after Collison joked that Intel could return to memory production “as is tradition,” Altman raised the topic of local AI.

“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. 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 being used 5% of the time. You want maximum utilization in the cloud if you’re trying to deal with this very supply-constrained environment.”

Sam Altman’s point is comparable to publishing. Imagine writing articles on paper, printing them, and sending them directly to readers’ homes like newspapers. It would be enjoyable, but not efficient, because we already have websites, the internet, social media, WordPress, and a highly efficient online publishing ecosystem.

However, the days when local models were not good enough are over, and powerful systems can now handle such models.

Altman has a point if you are building AGI and can decide where every gigabyte goes. But no one gets to make that decision. The market does, and people buy RAM for gaming, video editing, or, in my case, keeping nearly 100 Edge tabs open for research. My 16GB laptop used to remain between 90% and 95% RAM usage, making Altman’s 5% figure almost the exact opposite of my experience.

“Ram Altman” Backtracks, but His Argument Has Not Changed

The clip spread quickly among PC gamers already paying more for memory. On November 1, Elon Musk posted a meme featuring Hans Landa from Inglourious Basterds asking, “You are withholding DRAM from the cloud, are you not?” Musk added, “The people deserve to have their RAM too!”

On November 4, Altman changed his X name to Ram Altman at a user’s request and wrote, “i actually do want the world to have tons of memory and people to put as much as they want in their computers, and it is f***ing annoying that this was taken out of context. i have been arguing for the world to ramp up memory production for a long time.” In another reply, he added, “i want you to have so much ram” and “lets build more.”

To be fair, Altman also criticized memory manufacturers during the podcast for being too cautious about expanding production. However, I watched the entire podcast, and the quotation above provides the relevant context. Wanting more memory for everyone does not change his view that scarce DRAM delivers greater value in the cloud.

Satya Nadella Wants Unmetered AI on Every Desk and in Every Home

MS’s former mission was “a computer on every desk and in every home,” and Nadella has given it an AI-focused update. “Our goal is to deliver unmetered intelligence to every home and every desk with Windows,” he said in June.

In his 2026 annual report letter, Nadella writes, “Just imagine how much intelligence can effectively become unmetered as more and more capable models run locally,” with workloads moving between the cloud and PCs through “auto-routing.” Unmetered intelligence means the PC performs the work, eliminating the need to pay for every token generated by a local model. Users still have to pay for the PC itself, however, and PCs have become painfully expensive.

MS calls this approach Hybrid Intelligence. In an October 7 blog post, Windows chief Pavan Davuluri describes Windows as “a platform where agents can run locally when it makes sense, reach the cloud when they need to.” With user permission, Copilot can read relevant content on a PC, take actions across Windows, and use local models for appropriate tasks, which Nadella says helps “your tokens go further.”

MS is clearly not anti-cloud. Copilot’s Cowork and Autopilot operate through usage-based cloud billing, and MS has supported OpenAI since making a $1 billion investment in 2019. Redmond simply does not want Windows to become a dumb terminal for someone else’s data center.

MS’s 128GB Surface Laptop Ultra Shows How Much RAM Local AI Needs

For context, a local model runs on a PC’s CPU, GPU, or NPU. MS’s Windows ML documentation says this approach keeps “data private, eliminating cloud costs, and working without an internet connection,” but the PC must have enough memory to hold the model.

MS’s strongest example is the Surface Laptop Ultra, whose NVIDIA RTX Spark chip shares up to 128GB of unified memory between the CPU and GPU—enough for AI models with up to 120 billion parameters. The 128GB model costs $5,899.99.

On November 4, while Altman was renaming himself on X, MS engineers explained how they run MAI-Code-1.1-Flash on RTX Spark. The 137.9-billion-parameter coding model, with 6.8 billion active parameters per token, requires 276GB at 16-bit precision. Quantization, which stores each weight using fewer bits, reduces it to a 53GiB checkpoint. The model still requires additional memory for the KV cache, which retains the context of the code.

In MS’s setup, the model generated a median of 51.12 tokens per second with a 120K-token input. “The goal is more useful intelligence within the available resources, not fewer bits at the expense of accuracy or responsiveness,” the engineers wrote.

AI Data Centers Are Why Your Next PC’s RAM Will Cost More

Altman’s comments came at a difficult time for PC buyers. TrendForce expects conventional DRAM contract prices to rise by 10% to 15% quarter over quarter in Q4 2026. It also says PC DRAM supply could decline in 2027 “as more capacity shifts to server applications.” The HBM used in AI servers is not the same as the DDR5 found in laptops, but both compete for factory capacity among a small number of memory manufacturers.

It is also worth noting that PC shipments fell by approximately 20% last quarter.

Altman is correct that a data center extracts more value from every gigabyte than my laptop does, but I do not buy a PC to maximize global DRAM utilization. I buy it to work and play on my own terms. In my view, MS’s hybrid approach is the better idea, provided that “unmetered intelligence on every desk” does not ultimately mean a $5,899 desk.

Do you want AI running on your PC or in the cloud?

Windows Latest depends on readers like you. Make us your preferred source on Google Discover and Google Search, and help our independent reporting reach more people.