Sam Altman defends RAM comments, rebrands as ‘Ram Altman’ on X

Sam Altman spent election night fighting a different campaign: convincing the internet he is not coming for your computer’s memory.

The OpenAI CEO changed his display name on X to “Ram Altman” on November 4. He then insisted that comments he made about PC memory last week had been stripped of their context.

What Altman actually said

The trouble started on the Cheeky Pint podcast on October 29, 2026. Around the 29-minute mark, Altman described the use of DRAM in personal computers as “quite terrible” when viewed through a global allocation lens.

His reasoning was about utilization. Altman noted that only about 5% of DRAM capacity goes into personal computers. He argued that the memory would be better used in AI data centers, where hardware runs at much higher utilization rates.

The remarks came during a broader conversation about local AI inference. That means running AI models directly on your own machine instead of sending requests to a remote server. The discussion took place against a backdrop of tight memory supply driven by the AI infrastructure boom.

The internet did not hear a nuanced point about allocation efficiency. It heard the head of the world’s most prominent AI company suggesting your laptop doesn’t deserve its RAM. Elon Musk was among those who responded.

The ‘Ram Altman’ defense

On X, Altman pushed back hard. He said he genuinely wants people worldwide to have abundant memory and to load their machines with as much as they like. He added, with some profanity, that it was annoying to see his words taken out of context.

Altman also claimed he has spent a long time arguing for the world to ramp up memory production. In his telling, the point was never that consumers should get less. It was that everyone should get more.

In a reply to another user, he put it more simply:

“i want you to have so much ram lets build more.”

Why the comments landed so badly

Consumers are already feeling the memory crunch in their wallets. DRAM prices have reportedly jumped fivefold in some segments compared with the previous year.

Those costs have flowed downstream. PCs, laptops, gaming consoles and other consumer electronics have all gotten pricier as memory costs climb.

OpenAI’s own purchasing power adds another layer. Reports estimate the company has commitments with Samsung and SK hynix for up to 900,000 DRAM wafers per month. That figure could potentially represent 40% of global output.

The bigger shift in the memory market

Memory makers including Samsung and SK hynix have been prioritizing production for server memory and high-bandwidth memory, known as HBM.

HBM is a specialized, stacked form of memory designed to feed data to AI chips at very high speeds. It is essential for training and running large AI models. It is also far more lucrative than the commodity DRAM found in a typical desktop.

Some forecasts point to tight memory supply lasting through 2028-2030. Building new chip fabrication capacity takes years and enormous capital, so supply cannot simply be switched on to meet demand.