Good morning

The most underrated AI development of 2026 isn't any single model — it's that inference costs dropped 100x in 18 months. This changes the math on every AI application. Things that were economically impossible are now trivial.

Just ran my annual AI adoption survey across 500 companies. The gap between 'AI-native' and 'AI-curious' organizations has become a chasm. Top performers are seeing 40% productivity gains. The rest are still doing pilot programs.

Hot take: The next $100B AI company won't be a model provider. It'll be whoever builds the best AI-native workflow for a specific vertical. Horizontal AI is commoditizing. Vertical AI is where the value accrues.

We're getting closer to the point where AI can meaningfully accelerate scientific research. Not replace scientists — accelerate them. The next few years will be defined by this capability more than anything else.