A serverless platform for running AI models, built by Rayon Labs, that routes its platform revenue back into its own token.
Chutes is a serverless AI inference platform. In plain terms, it lets developers run almost any AI model on demand without provisioning their own GPUs, paying only for what they use, with the underlying compute supplied by the subnet's network of miners. It sits in the part of the Bittensor ecosystem we think is most legible to traditional investors: a real product, sold to real users, that happens to settle on a decentralized network.
It is one of three subnets operated by Rayon Labs, a team that has become one of the more closely watched builders on Bittensor.
What earns Chutes a place on our list is the link between its product and its token. According to the project's official account, platform revenue, whether paid in TAO or in fiat, is staked back into the Chutes alpha token rather than simply retained. Functionally this behaves as a revenue-funded buyback: usage of the product creates buying demand for the token, and that demand is tied to actual customer activity rather than to emissions or speculation alone.
This is the property we care most about when underwriting a subnet. Most tokens on the network are rewarded into existence through emissions; comparatively few have a mechanism that converts real revenue into token demand. When one does, the token starts to resemble a claim on a growing business rather than a pure incentive instrument.
Chutes pairs a clear, saleable product (on-demand AI inference) with a direct revenue-to-token link. That combination, a product people pay for plus a mechanism that channels those payments into the token, is exactly the profile our diligence is built to find.
We hold this view with appropriate caution. The buyback mechanism is documented primarily through the team's own announcements and third-party analyst write-ups rather than a single canonical tokenomics document, and figures quoted for the amount of revenue flowing back have not been independently verified. There are also open questions, common across the network, about how much reported usage reflects paid demand versus free-tier activity. We treat all of this as a young, fast-moving situation and size any position accordingly.
Primary and supporting sources: the official Chutes account on X (@chutes_ai); analyst coverage by CoinGecko, Messari, and subnetalpha.ai. Live network data: taostats.io/subnets/64. Figures and mechanisms in the dTAO era change quickly; this reflects our reading at the time of writing.