Elon Musk Says Humanoid Robots Will Create Bigger RAM Demand Than Everything Else

Elon Musk has predicted that humanoid robots will eventually require more RAM than all other applications on Earth combined. He made the comment on X in late October 2026 while responding to a Counterpoint Research report on robot memory needs.

Musk’s Prediction and Reasoning

Counterpoint estimated that global DRAM demand from humanoid robots could rise more than 20 times by 2030, from roughly 8,300 terabytes in 2026 to about 172,000 terabytes, with average memory per robot increasing from around 19GB to 39GB. Musk called the analysis “silly” and argued the real figure will be far higher.

His exact wording:

• “Robot RAM needs will exceed everything else on Earth, as you need the robots to be on physical substrates in order to shape atoms.”
• He added that superintelligence needs will “vastly exceed” robot demand, but roughly 100% of that compute will run in deep space.

The core idea is that robots must process real-world sensing, decision-making, and physical actions in real time. This creates continuous, high-volume memory demand that scales with the number of units deployed, unlike pure software AI running in data centers.

Context from Tesla and Memory Suppliers

Musk’s comment follows earlier statements about Tesla’s Optimus robot. In early October, he said Tesla cut memory on its AI5 and AI6 chips (to roughly 72GB and 144GB, with a later adjustment upward on AI5) because it was the only practical way to secure enough volume for production amid tight supply. He described the performance impact as negligible since bandwidth mattered more than total capacity.

Memory makers have already flagged the category:

• Micron CEO Sanjay Mehrotra has noted that Level 4+ autonomous vehicles typically need over 200GB of memory and multiple terabytes of storage, and that humanoid robots will likely require similar levels.
• Analysts have pointed out that even modest Optimus volumes (for example, several million units per year at 72GB each) would translate into hundreds of petabytes of demand, favoring low-power DRAM suppliers such as Micron and SK Hynix.

Implications for the Memory Market

AI data centers have already driven a global memory shortage and higher prices for consumer devices. Musk’s view suggests physical AI and large-scale robotics could become an even larger long-term driver. The gap between current forecasts and his estimate mainly depends on how many robots are eventually produced. At car-like annual volumes, robot memory alone could absorb a sizable share of today’s global DRAM output.

Whether the prediction holds depends on actual robot deployment rates, further efficiency gains in onboard AI, and continued expansion of memory manufacturing capacity. For now, it adds another major demand source to an already constrained market.