Why AI is so centralized: How it's built, who controls it, and what comes next

Decentralized AI infrastructure is essential to prevent the concentration of power within a few hyperscalers and to support the rise of autonomous, sovereign machine agents. By utilizing crypto-economic protocols, machines can establish programmatic trust, enabling them to operate independently of human-centric legal systems. This transition marks a shift from deterministic, imperative computing to probabilistic models, mirroring how humans navigate uncertainty. As AI systems gain the ability to update their own objective functions and manage on-chain identities, a Darwinian market for intelligence will emerge, fundamentally altering the economic and social landscape. Founders Ben Fielding and Harry Grieve emphasize that building open infrastructure—rather than treating AI solely as a product—is critical for ensuring a future where technology serves humanity without the bias of centralized control, ultimately moving beyond the current reliance on human-labeled data and centralized server architectures.