1. Introduction

Since 2018, when the global crypto industry entered an infrastructure-centric development cycle, the cross-innovation of blockchain and artificial intelligence has accelerated continuously. From verifiable computation, computing power networks, and ZK cryptography to the explosive growth of generative AI, the global demand for training computing power has surged exponentially. This has made "who can provide provable, measurable, and allocable computing power" the core competitiveness of the next-generation digital economy. However, in traditional centralized computing power systems, structural problems such as opaque resource allocation, high node participation thresholds, underutilized devices, and limited value return remain unresolved. A large number of GPUs, CPUs, and edge nodes possess real computing power but cannot be integrated into the AI training value chain, resulting in significant resource waste.

1. Introduction

Since 2018, when the global crypto industry entered an infrastructure-centric development cycle, the cross-innovation of blockchain and artificial intelligence has accelerated continuously. From verifiable computation, computing power networks, and ZK cryptography to the explosive growth of generative AI, the global demand for training computing power has surged exponentially. This has made "who can provide provable, measurable, and allocable computing power" the core competitiveness of the next-generation digital economy. However, in traditional centralized computing power systems, structural problems such as opaque resource allocation, high node participation thresholds, underutilized devices, and limited value return remain unresolved. A large number of GPUs, CPUs, and edge nodes possess real computing power but cannot be integrated into the AI training value chain, resulting in significant resource waste.

Blockchain's verifiable computation mechanism offers an opportunity to redefine the computing power market: computing power can be proven, tokenized, and fairly allocated, while training tasks can be recorded, tracked, and incentivized, forming an open and transparent global computing value network. At the intersection of this technological trend and market demand, cAIToken (Compute AI Token) emerges as a "decentralized AI training computing power token."

The cAIToken token structure is unique:

Only 80,000,000 cAI tokens are minted network-wide, all written once into the unique mining pool contract; both the tokens and the pool have renounced all permissions, with no reserves, no team allocation, and no private sale share.

All token output (100%) is released by the mining pool according to on-chain rules, strictly limited to two types:

PoW mining from real computing contributions;

Nodes staking cAI to gain verification qualifications, earning ecological rewards based on verification behavior.

This design makes cAIToken a genuine "computing output asset." Any token acquisition relies on real computation or ecological contribution, with no background issuance or manual control, ensuring high transparency and credibility.

From a technical architecture perspective, cAIToken centers on Compute-PoW (Computational Proof-of-Work) and Proof-of-Training, converting gradients, inference tasks, and model fragment processing generated during training into verifiable on-chain contribution certificates. Meanwhile, the Stake-Boost mechanism enables nodes with long-term participation intent to secure stable task and verification rights, increasing reward distribution efficiency and network activity without altering total supply.

Through the design logic of "single pool, zero permissions, dual proof models," cAIToken transforms computing power from a closed, centralized resource into a freely participatory, verifiable on-chain asset; every GPU, CPU, and edge device receives a measurable, on-chain identity; and real training tasks become the core driver of value circulation.

This white paper systematically details cAIToken's technical architecture, computing consensus mechanism, mining pool release logic, token economic model, node ecosystem, governance system, risk control, and future roadmap.

The mission of cAIToken is to become the "decentralized computing infrastructure for the AI training era," establishing a novel computing economy centered on real computing power and task verification.