Currently, AI large models are very hot around the world and the competition is fierce, but many of them are only in the training stage and have not been widely used. A large model is like a public chain, and inference applications are the ecosystem on the public chain, and the development of the ecosystem takes time.

At the same time, compared with training, inference will occupy more than 90% of the life cycle of large models in the future, and the cost will be higher. We believe that training the network is currently unachievable. There are many limiting factors, such as transmission speed, privacy, etc., so we believe that the inference network is more realistic, especially in the field of web3, and It’s more in line with future development.

Compared with Bitcoin's meaningless PoW calculation, the PoIW (Proof-of-Infer-Work) algorithm we launched is compatible with the underlying AI inference algorithm and is a truly meaningful PoW calculation.