2.3 Why does the emergence of @0G_labs make people rethink the relationship between AI and blockchain? In the traditional blockchain world, the chain's tasks are to execute smart contracts and settle transactions, while AI training and inference still run on centralized servers. @0G_labs aims to break this divide, enabling the chain not only to handle transactions but also to support data storage, training, inference, and service markets for AI models. It proposes a vision of a decentralized AI operating system, hoping to seamlessly integrate these functions into a modular architecture, allowing developers to deploy AI applications on the chain just like deploying regular dApps.

Practically, such a system demands extremely high data availability and network performance. @0G addresses broadcast bottlenecks by separating data publishing channels from storage channels, enabling the data availability layer to efficiently handle large-scale AI data at low cost. This approach not only significantly increases data throughput on the chain compared to traditional L1s but also offers almost unlimited scalability potential for the on-chain AI ecosystem.

However, the most critical question is: when a chain incorporates all the underlying resources needed for AI into a decentralized ecosystem, can it truly outperform traditional cloud service providers' centralized solutions in the future? If decentralized networks cannot match centralized platforms in stability and cost over the long term, can they genuinely support industry-level AI applications? This is a fundamental challenge to the entire concept of decentralized AI.