AI inference demand is driving a structural shift in memory markets. Micron, SK Hynix, and Samsung sit at the center of it, with supply constraints tightening across HBM and DRAM as data centers scale inference workloads faster than fabs can respond. Shortages are no longer contained to enterprise hardware: they are beginning to hit consumer devices.
The video breaks down why inference, not training, is now the dominant demand driver for memory, and why that distinction changes the investment thesis entirely. The supply constraint analysis starting at 6:52 is the most technically grounded segment, covering how lead times and capacity allocations are creating pricing leverage for the big three players.
The forecasting section at 20:04 is where the argument either holds or falls apart. Worth watching to stress-test the timeline, not just the conclusion.
[WATCH ON YOUTUBE →]