#45
2026-11-12
Duo
28:46
NO KINGS 🤴 Diversification and Discrimination in the Token Rush
Key Takeaways
- Concentration is a cluster risk. Here, power and competence are spread across more heads. The picture of a Europe left behind misses how robust that is.
- The token rush carries debt. The data centres are credit-financed and only pay off with rising token consumption. Subsidised subscriptions are the way in; through the API it gets expensive.
- Smart force instead of brute force. Everyday tasks rarely need a frontier model. Small local models in a good architecture handle them without a token bill and without silent updates from the provider.
- Personal productivity is not yet a company result. Without coordination everyone builds for themselves and the same solution is created a hundred times over. Agent registries and tracing create an overview, yet hardly any company uses them.
- Diversification: never everything on one provider. Whoever ties themselves to one carries that provider's bet. Several providers and your own model on your own infrastructure reduce the dependency.
- Discrimination: allocate tokens deliberately. Tokens are finite. Budgets therefore follow demonstrated benefit, measured in the team and against the company goal. Data protection and codetermination limit the view of individuals. Token efficiency is likely to weigh on careers nonetheless.
Summary
The American AI bet rests on a handful of companies and their bosses. The price swings after the US midterms and the turbulence around the Nvidia chief show how quickly that much concentrated capital can start to slide. The motto of this DEEPTECH DEEPTALK episode: No Kings.