Abstract
TRD Network is decentralized infrastructure for AI. It unifies GPU compute, virtual machines, sovereign storage and autonomous agents into one two-sided network — one side rents capacity, the other supplies it — coordinated by a dispatch layer that matches workloads on benchmark tier, reputation and sovereignty constraints.
What sets it apart is verifiability: every unit of work emits a cryptographically signed R+2 receipt (Ed25519 over RFC 8785 canonical JSON), with arguments hashed rather than stored raw. This paper covers the architecture, the five products, credit economics, the dispatch reverse-auction, the full receipt specification, the platform API, and the security model. Figures are illustrative; live pricing renders in the dashboard.