Compute Gap Debate: Stop Moving the Goalposts
teortaxesTex · x · 2026-07-18
The post responds to claims questioning why the US-China compute gap hasn't shown up, arguing a core point: US labs already have massive numbers of chips online and have invested billions in GPU R&D. These costs shouldn't simply be dismissed as something easily erased by distillation.
The author further challenges the narrative of "waiting until the end of the next year to see." Every time a prediction fails to materialize immediately, the reasoning shifts to "booked compute isn't online yet," "the gap will emerge after the next expansion," or "let's wait for the next training cycle." This continuous moving of the goalposts without ever clearly validating the gap makes the argument increasingly untenable.
The text in the attached image reinforces the same message: explanations regarding chip inventory, signed-but-not-online compute, RL phases, and chip yields cannot be postponed indefinitely, lest they devolve into a perpetual "just wait another 18 months" narrative.
More from Infra
- SkyPilot exits stealth with $20M seed round and an AI compute platform for fragmented clouds — jfiance · 2026-07-22
- Nothing phone mockup turns a film joke into a modular design meme — ZeYanjie · 2026-07-22
- Actual Computer says its inference stack is tuned for Nvidia’s consumer Blackwell lineup — markjeffrey · 2026-07-22
- Ben Bajarin says CPU demand is still being badly underestimated — BenBajarin · 2026-07-22
- An energy model says the U.S. could run short of natural gas starting in 2028 — churchkey · 2026-07-22
- Devin adds e2b sandboxes for remote agent execution — badphilosopher · 2026-07-22