Own the bottleneck, not the brand: the chokepoint every AI chip depends on, at a discount to all of them.
Parameter scaling hit diminishing returns — GPT-4 to GPT-5 gap smaller than GPT-3 to GPT-4
Four independent memory stack improvements compound: HBM4 (2x) × KV compression (4-6x) × optical (70% energy) × architecture (2-10x)