Hyperscalers are spending ~50% of CapEx on memory, with DRAM/NAND markets running ~30% short of demand.

Memory chip makers operate at 80-90% gross margins, higher than Nvidia's approximately 70% gross margin.

“Nvidia is up there at like 70%, the Mory guys are at 80 to 90% now.”

Nvidia is considering releasing lower-memory versions of Rubin Ultra partly because it may not secure enough high-bandwidth memory for the production run.

“Nvidia is considering releasing the lower memory versions partly due to concerns that they won't be able to secure enough high-bandwidth memory for the production run.”

The AI inference bottleneck is not compute but data movement, because running larger models requires moving weights and the KV cache into compute units.

“The inference problem is not really a compute problem because as the models get bigger, you now need to move the weights and of course what we call the KV cache into the compute units. And that is essentially a data movement problem.”

DRAM, NAND, and PCB markets are currently running approximately 30% short of demand.

“We're already like 30, we're 30% short all these things as we are now.”

Microsoft, Meta, and Google are each on track to spend over 50% of their revenue on capex in 2026.

“Microsoft, Meta and Google are all on in line to spend over 50% of revenue on capex this year.”

Micron is achieving 80-85% gross margins on HBM memory due to AI-driven demand, indicating severe supply constraint.

“Micron producing numbers where they have 80 85% gross margins.”

Either Intel or Samsung will break capital discipline and overbuild semiconductor capacity, ending the current supply constraint.

“One of Intel and Samsung they're not going to stay disciplined. They will break.”