Why AI Could Keep Memory Prices High Until 2030 - Hardware, PCs and semiconductors

Rising demand from AI data centers, robotics and autonomous vehicles could prolong the memory supply shortage, with SK Hynix warning that constraints may persist through 2030.

Consumers hoping for cheaper DRAM may have to wait longer as demand from artificial intelligence and other emerging technologies continues to pressure global memory supplies.

OpenAI chief Sam Altman has argued that DRAM should be prioritized for data centers, claiming that consumer memory is used only a small portion of the time. The position has drawn criticism because the expanding AI infrastructure market is itself contributing to the shortage.

OpenAI’s Stargate data-center project, announced in early 2025, was expected to require around 40% of the world’s memory supply. At the same time, major memory manufacturers had previously reduced production to help stabilize prices, making it more difficult for supply to respond quickly to new demand.

The immediate risk is not that consumers will physically lose the RAM installed in their computers. However, limited production capacity could increasingly be directed toward the high-bandwidth memory used for AI training and inference, where manufacturers can target a more lucrative market than consumer memory.

SK Hynix, one of the world’s leading memory producers, has warned that the supply shortage could continue through 2030. The outlook could become even more uncertain if OpenAI and Anthropic enter public markets during that period, potentially increasing pressure on the companies that control roughly 90% of global memory production.

Demand from AI may also be joined by growth in robotics and autonomous vehicles. Humanoid robots, warehouse automation, autonomous taxis and self-driving cargo vehicles all require memory to run their onboard AI systems. If these technologies expand across manufacturing, hospitality and logistics, they could compete with consumer electronics for the same constrained resources.

The result could be a prolonged imbalance in which memory production favors data centers, robots and autonomous vehicles over personal computers and other consumer devices. For buyers, that would make a return to consistently affordable memory less likely in the near term.

Sources: TechRadar.