Amazon and Google are winning the AI capex race — but what’s the prize?

Two hundred billion dollars and you still don't know if it's worth it. That's the actual story.

Amazon and Google are basically building fortresses—chip fabs, power infrastructure, data centers scaled for training. It's a capital arms race. But here's the thing: capex doesn't predict model quality or user adoption. Bigger infra doesn't mean better reasoning, cheaper inference, or products people want.

They're betting the scale alone will compound into moat. Maybe it does. But they're spending like the game's already won when the competition hasn't even solidified.