KM4BI

Gavin Baker on Anthropic’s S-1, GPUs and SpaceX

Gavin Baker discusses Anthropic’s filing, tight compute markets, open-model inference providers and the operational demands of scaling AI infrastructure.

Published

Gavin Baker sees Anthropic’s S-1 as an upside surprise, with meaningful enterprise adoption and a customer base he considers more diversified than some market narratives suggest. He also points to fast-growing inference providers built around open models.

Compute remains constrained

Despite plans for new capacity, Baker says GPU pricing and forward commitments remain tight into 2027. Building a cluster is not simply a matter of buying hardware: power, deployment speed and operating expertise determine whether capacity can earn revenue. Fixed-price long-term deals can become a disadvantage when spot prices rise.

Agents and new funding channels

Baker describes AI agents as practical tools for investment research and information routing. He also discusses possible AI applications in biology and argues that public listings could broaden investor access and help finance future infrastructure.

The central question is execution: which companies can bring compute online reliably, and whether productivity gains and demand will justify the scale of investment.

Source