A recent forecast indicates that memory demand driven by humanoid robots is expected to skyrocket 20-fold by 2030. This prospect means that, following AI servers and consumer electronics, humanoid robots are poised to become the next growth driver for Samsung Electronics and SK Hynix.
Market research firm Counterpoint Research released data on the 11th showing that DRAM memory demand from the humanoid robot industry will grow approximately 20-fold, rising from about 8,300 Tb (terabits) this year to 172,000 Tb by 2030. The forecast reflects two major trends: humanoid robots are beginning to be equipped with robot foundation models—high-performance "brains" built on physical AI—which increases the DRAM capacity required per device; at the same time, robot shipments themselves are officially entering a growth trajectory.
Counterpoint expects that the DRAM capacity required per humanoid robot will more than double, increasing from 19 GB this year to 39 GB by 2030. Data from Smart Analytics Global shows that global humanoid robot shipments will reach 60,000 units this year, three times last year's figure; by 2030, shipments will grow to approximately 500,000 units.
Industry insiders say this trend will significantly boost demand for low-power DRAM (LPDDR, the memory commonly used in smartphones), because energy efficiency levels largely determine robot performance. High-capacity NAND flash will also open up new markets in the robotics sector. Currently, HBM faces capacity bottlenecks, and NAND flash, as a complementary AI storage solution, is attracting market attention.
Memory manufacturers have already begun positioning themselves. Samsung Electronics said in July that the company is proactively developing a comprehensive next-generation semiconductor solution for robots, covering high-bandwidth low-power memory, image sensors, foundry services, and advanced packaging. Its Device eXperience division is leading efforts in this market and has specifically established a Robot Experience Business Unit to conduct humanoid robot-related research and development.
SK Hynix plans to jointly develop next-generation memory products for Nvidia's Jetson Thor robot chip platform. Micron mentioned in its latest earnings call that L4 and above autonomous driving vehicles typically carry more than 200 GB of memory and several terabytes of storage space, adding that humanoid robots will require memory and storage capacities at a similar level.