Elon Musk Just Uttered 3 Massively Bullish Words for Micron, Sandisk, and SK Hynix

Elon Musk stated that memory—not compute—is the rate limiter for agentic AI systems, spotlighting demand for DRAM and NAND flash producers MU, SNDK, and SKHY. Agentic AI requires vast specialized memory and storage for planning, task execution, and data retrieval, driving an explosive demand cycle. GS estimates agentic AI will consume 120 quadrillion tokens monthly by 2030, representing 24 times current usage levels. The projection suggests the memory up-cycle could extend well beyond current expectations despite recent pullbacks in memory stocks.

The token forecast underscores a structural shift in AI infrastructure spending. As AI agents proliferate across enterprise and consumer applications, memory capacity becomes the critical bottleneck rather than processing power. This dynamic positions memory manufacturers at the center of the next phase of AI buildout, with demand visibility extending through the decade.