Blockworks Report: 7 Crypto Inference Providers Process 211 Billion Tokens Daily

Huo Xing Cai Jing reports that Blockworks’ latest report on the state of crypto inference service providers reveals massive global compute investments with demand still outstripping supply. The market capitalization of liquid crypto AI stands at approximately $2.49 billion, below Crusoe’s $3.09 billion, and no decentralized compute network has yet disclosed verifiable revenues exceeding tens of millions of dollars. The report compares seven crypto-native inference providers on OpenRouter: Venice, Phala, Darkbloom, io.net, AkashML, NEAR AI, and Chutes. Over the past month, these seven providers processed an average of 211 billion tokens per day, accounting for roughly 1% of OpenRouter’s daily total of 21.9 trillion tokens. Venice processed an average of 87 billion tokens per day; its team reported in July an annualized recurring revenue of $100 million, with less than 10% coming from OpenRouter. Phala processed an average of 54.6 billion tokens per day over the past week, up from about 5 billion per day in July. Chutes, which operates as Bittensor’s inference subnet SN64, runs approximately 4,350 GPUs and claims to process 160 billion tokens per day, with 570 million tokens per day flowing through OpenRouter. Akash has 283 active GPUs, down from a peak of about 897, with annual rental fees of approximately $3.7 million; GPU incentives have ended. io.net’s demand nearly doubled over the past month to an average of 14.8 billion tokens per day.