Recently in an internal meeting, the Delphi Research team pondered how many AI applications truly require crypto. Meanwhile, researchers and investors have questioned whether the intersection of crypto and AI is a forced marriage or a necessary evolution. While AI has advanced rapidly under the control of centralized tech giants like OpenAI, Google DeepMind, and Anthropic, the question remains: Is crypto fundamentally necessary for AI’s future?

For now, the leading AI applications function well enough without blockchain-based systems. Most of today’s AI models run efficiently on cloud infrastructure, remain within corporate-controlled ecosystems, and require enormous amounts of capital for compute-intensive training—funded primarily by venture capital and large corporations. AI development remains highly centralized, with a small number of companies controlling access to compute resources, model deployment, and infrastructure. However, as AI expands beyond software and into the physical world—powering robotics, IoT, smart grids, and autonomous infrastructure—the limitations of this centralized model will become apparent.

Centralized coordination models, like those used by Tesla and Boston Dynamics, work well in closed ecosystems but face limitations as Physical AI scales into multi-stakeholder environments. AI-powered robotics, IoT devices, and autonomous infrastructure must coordinate across independent entities with conflicting incentives, requiring a framework that reduces reliance on trust. Centralized