Humanoids Are Becoming the Memory Industry’s Next Growth Engine

The memory industry has found its next growth story, and it walks on two legs. As humanoid robots move from laboratory demonstrations toward commercial deployment, the chips that give them their memories are becoming one of the semiconductor industry’s most closely watched new markets.

According to market research firm Counterpoint Research, demand for the DRAM used in the “brains” of humanoid robots will rise from roughly 8,300 terabits in 2026 to about 172,000 terabits by 2030 — an increase of more than twenty times in four years. The average DRAM capacity fitted to each robot is expected to more than double over the same period, from 19 gigabytes this year to 39 gigabytes by 2030.

The pattern is familiar. Just as AI data centers sparked a boom in high-bandwidth memory, humanoid robots are emerging as the memory industry’s next major growth engine for Samsung Electronics and SK Hynix — the two companies that dominate advanced memory manufacturing.

Why robots are so hungry for memory

Humanoid robots process information collected in real time from cameras, LiDAR, and tactile sensors to decide how to move. The spread of end-to-end AI models that fuse multiple data types — plus “world models” that learn and predict the surrounding environment — is pushing memory requirements steadily upward.

Power is the binding constraint. Humanoids run on batteries, so they must process large amounts of data while minimizing consumption — which is why low-power DRAM and high-capacity NAND flash are expected to be the key demand sources, with three-dimensional DRAM a longer-term variable.

A forecast, not a guarantee

Counterpoint’s analysts attach a caveat worth repeating: actual market growth will depend on how fast humanoids commercialize, how quickly production scales, and how fast per-product memory specifications improve. A forecast is a model of the future, not the future.

The compute side of the robot is expanding in parallel. Counterpoint expects demand to grow for processing chips such as NVIDIA’s graphics processors, alongside custom semiconductors being developed by automakers including Tesla and Xiaopeng.

For Samsung and SK Hynix, the prize is diversification. Counterpoint estimates that in the second quarter of 2026, SK Hynix held 50 percent of the global high-bandwidth-memory market and Samsung 33 percent — dominance built on the AI data-center boom that both companies now hope to extend into robotics.