1. Preface

— A Next‑Generation DePIN‑Based Web3 Resource‑Stack Network —

With the explosive growth of AI technologies and the continuous maturation of Web3 infrastructure, the global supply‑and‑demand landscape for computing resources is undergoing profound change. A new wave of AI services—represented by AIGC and autonomous Agents—is penetrating digital application scenarios at unprecedented speed. The underlying inference and training processes now demand heterogeneous compute, elastic scheduling, and localized processing capabilities. Meanwhile, the persistent operation of agents requires vast amounts of edge data, meaning that resource needs extend beyond computation to include storage, bandwidth, and access to users’ native data.

Traditional centralized resource platforms—such as large cloud providers or AI‑API vendors—can support initial model deployment in the short term, yet their high costs, poor composability, and opaque services have become bottlenecks that hinder the widespread adoption of intelligent services. In architectures centered on multiparty collaboration, real‑time feedback, and dynamic self‑learning, centralized platforms struggle to meet requirements for low latency, high privacy, and flexible composability.

As Web3 evolves, our understanding of resources and networks is being reshaped. In the traditional paradigm, resources are merely hidden supports for system operation. Under a Web3 architecture, resources can be “put on‑chain” and assigned clear ownership, becoming core assets that are participatory, tradable, and governable within on‑chain collaboration and economic systems. From decentralized storage and computation, to on‑chain data ownership and exchange, to modular packaging of agent execution logic and service pricing, Web3 technologies make “resources‑as‑a‑service” not only technically feasible but also institutionally sustainable. This shift is driving application architectures to evolve from a platform‑centric “API economy” to a node‑centric “Agent network,” where resources are no longer background elements but active participants in value flow.

Electric Capital and Messari’s 2024 annual report shows that the number of on‑chain developers focused on AI has grown by more than 80 % year‑on‑year, indicating that an increasing number of developers see on‑chain technology as fertile ground for AI services. DePIN‑related projects have doubled within a year, spanning distributed computation, bandwidth networks, energy coordination, and hardware‑incentive networks. The fusion of AI and Web3 is moving beyond model tokenization and simple incentives, penetrating deep into resource supply, service deployment, and incentive mechanisms, giving rise to a new resource‑collaboration paradigm. In verticals such as healthcare, finance, smart home, gaming, and social, systems must dynamically access heterogeneous devices, call local data, and build on‑chain autonomous agent services to satisfy diverse, real‑time edge‑compute demands. Heterogeneous resource coordination, edge‑computing autonomy, and composable agent deployment will be foundational capabilities for large‑scale roll‑out.

Against this backdrop, Kairos adopts the core philosophy of “Resource‑as‑a‑Service” and, following a DePIN‑based architecture, builds a Web3 resource‑stack network with high accessibility, strong programmability, and native incentive mechanisms. The system aims to provide unified, flexible, and scalable infrastructure for diverse application scenarios—including AI, healthcare, gaming, smart home, and social—driving the sustainable evolution of next‑generation decentralized intelligent systems.