Shouldn’t a 32% increase in prices only make a modest difference to training FLOP? In particular, see the compute forecast. Between Dec 2026 and Dec 2027, compute increases by roughly an OOM and generally it looks like compute increases by a bit less than 1 OOM per year in the scenario. This implies that a 32% reduction only puts you behind by like 1-2 months.
The real point is where capital investment into AI declines because the economy tips over into a mild recession, and I’d like to see whether the tariffs make it likely that future AI investments decrease over time, meaning the timeline to superintelligent AI gets longer.
Shouldn’t a 32% increase in prices only make a modest difference to training FLOP? In particular, see the compute forecast. Between Dec 2026 and Dec 2027, compute increases by roughly an OOM and generally it looks like compute increases by a bit less than 1 OOM per year in the scenario. This implies that a 32% reduction only puts you behind by like 1-2 months.
Of course, tariffs could have more complex effects than just reducing GPUs purchased by 32%, but this seems like a good first guess.
The real point is where capital investment into AI declines because the economy tips over into a mild recession, and I’d like to see whether the tariffs make it likely that future AI investments decrease over time, meaning the timeline to superintelligent AI gets longer.
Sure, but note that the story “tariffs → recession → less AI investment” doesn’t particularly depend on GPU tariffs!