Stress-testing AI inference profitability
I built a small simulator to stress-test the unit economics of AI inference. The question I wanted to isolate is simple: under what assumptions does frontier AI inference become profitable enough to justify the current capex cycle? My current read …