Frontier-ness is a derivative of culture, and that takes time to compound . That’s why these sorts of sovereignty bets have payoffs that exist decades in the future, and not before. The issue is not so much the initial intent; it’s the preservation of that intent over time in spite of the many, many alluring things that will spring up, and the change in terrain that is inevitably going to happen.

The bet underneath it is economic. Pullen argues that an inference company, rather than a training-first lab, doesn't need billions to compete: millions, a national compute allocation, and a consortium feedback loop can be enough. From there it gets into the machinery, why open-weight models still trail the frontier on size, active parameters and data, the mixture-of-experts versus dense trade-off and why active params dominate how a model actually feels, and the edge that real coding trajectories confer.

Watching America Run Away With AI - Alistair Pullen (Cosine AI) - YouTube ↗