Part 5/11:
One of the most striking points from these conversations is the near-linear relationship between compute capacity and revenue when dealing with AI. Greg Brockman mentioned that if they had 10 times more compute, their revenue could approach a tenfold increase—an almost unprecedented event in business, where diminishing returns are typical. This indicates that AI demand is so high that every incremental increase in infrastructure directly translates into revenue growth, fueling a cycle of persistent shortages.
Satia Nadella further reinforced this idea, highlighting that the demand cycle in AI is cyclical and driven by the capacity to supply. The current challenge is aligning buildout speed with demand, which is growing faster than supply due to technological and logistical constraints.