Why the AI boom could keep memory prices high until 2030

OpenAI CEO Sam Altman has sparked controversy by suggesting that consumers don't require dynamic random-access memory (DRAM) as it is only used "5% of the time." Altman advocates for prioritizing DRAM supplies for data centers to support cloud environments and fuel the growth of AI. While Altman's viewpoint may seem appealing to AI companies, it raises concerns for consumers seeking affordable memory prices.

The DRAM shortage stems from OpenAI's ambitious Stargate datacenter project in 2025, which required 40% of the global DRAM supply. This demand surge was exacerbated by memory producers entering a production downturn to stabilize prices. OpenAI cannot physically confiscate consumers' RAM, but the current memory scarcity poses several risks for the future. SK Hynix, a major memory manufacturer, warns that the supply shortage could persist until 2030, potentially preventing any return to affordable pricing in the near future.

As OpenAI and Anthropic prepare for public markets, uncertainty looms over how memory giants control 90% of the world's supply will adjust their production. Given the significant revenue difference between the consumer memory market and AI, a substantial shift in production could occur towards high-bandwidth memory needed for AI. This trend might be amplified by AI's impact on employment, with firms implementing hiring freezes as automation replaces repetitive tasks, potentially reducing demand for consumer memory.

Beyond memory, the robotics industry is growing rapidly, with applications in various sectors such as manufacturing, hospitality, and logistics. AI-powered robots, capable of working continuously without rest and without labor laws, could replace human workers. Although robots require memory for their AI systems, the demand for consumer memory could decrease further, as businesses prioritize AI development over expanding memory production for the consumer market.