The $10 Trillion Token Economy — Alex Atallah, OpenRouter & Anjney Midha, AMP
Source
youtube.com
Author
Latent Space
Date
Context
OpenRouter addresses the distribution problem between model developers and application builders. Its documentation says it gives access to hundreds of models through a single API endpoint, with the requested model specified in the call. That lets an application change which model it uses while retaining the same basic integration. In the interview, the company's cofounder presents this shared access layer as a bet that the future of AI will be multi-model. Its value proposition depends on making that variety usable.
Key takeaways · AI-distilled
OpenRouter serves more than 10 million developers and over 10 trillion tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → per day, and the episode covers its move into Stripe.
Its early Mixture of Models experiment, MOM, failed in 2024; Atallah says model fusion works much better now, and the team brought the idea back years later.
The Mistral price war is described as the first real proof that a competitive inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → marketplace could work.
Midha argues inference gateways are becoming fraud targets and that Stripe's fraud infrastructure matters strategically as autonomous agents begin attacking token flows.
Terms in this piece · Glossary
model routing — Sending each request to a model chosen by the difficulty of the task, rather than using one model for everything.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
