Gamma Tokens Could Turn Subnets Into Customers of Each Other

Bittensor co-founder Const introduced a new token standard at Exploit Summit called Gamma tokens, designed to let subnets buy and use products built by other subnets.

The idea is meant to solve a growing problem inside Bittensor. The network now has subnets building compute, inference, agents, robotics, data products and other forms of AI infrastructure, but those systems still operate largely as separate economies.

Gamma tokens are designed to connect them.

What Gamma Tokens Are

In case you missed it, here’s everything @const_reborn said about Gamma tokens.

In this 8-minute clip, he explains why Gamma could consolidate Bittensor’s value both internally and externally.$TAO → Alpha tokens → Gamma tokens

Better to hear it from the man himself. 👇 pic.twitter.com/7WTc0i0dX8

— The TAO Daily (@taodaily_io) September 28, 2026

Gamma tokens would represent usage credits for products and services built on Bittensor.

A compute subnet could have Gamma tokens that give access to compute. An inference provider could have credits that can be spent on model inference. Other subnets could then buy those credits and use the service inside their own products.

Const said these credits are being designed to be purchasable, holdable, transferable and fungible.

The key point is that Gamma tokens would represent something that can be used, not simply another token tied to a subnet.

Subnets Could Start Buying From Each Other

This is where the idea gets more important.

A subnet does not need to build every part of its product itself. It may be good at robotics, for example, but still need compute and inference from somewhere else.

Const used OpenRoboto as an example. The subnet could direct part of its emissions toward buying usage credits from infrastructure providers inside Bittensor, while the rest of its emissions continue supporting the miners doing its core robotics work.

The same could apply to a subnet like SOMA. If it needs compute or other AI services to run its product, part of its emissions could be used to buy those services from another subnet.

That means emissions would no longer only reward miners and validators. They could also help pay for the real resources needed to build a working product.

Bittensor Starts Looking More Like an AI Economy

This creates a very different relationship between subnets.

Instead of every subnet operating on its own, one project could provide compute, another could provide inference, another could build models, and another could combine all of those services into a product.

Const pointed out that many AI companies from Silicon Valley already spend a large share of the money they raise on compute. His view is that Bittensor should be able to support that same spending internally.

A subnet could use resources from other subnets instead of going outside the ecosystem for everything it needs.

That would make Bittensor less like a collection of separate projects and more like an economy where different businesses depend on each other.

Gamma Tokens Could Also Work Outside Bittensor

Const also said these usage credits are being designed to work beyond the network itself.

Bittensor is working on bridging technology that could allow Gamma tokens to be used from other chains. This could make it easier for people, applications and AI agents to access Bittensor services directly.

An AI agent, for example, could buy the credits it needs to rent compute or run inference without creating an account, logging into a website or using a traditional payment system.

That gives Gamma tokens two jobs. They could help subnets trade with each other internally while also making Bittensor services easier to sell to users outside the ecosystem.

TAO, Alpha and Gamma Now Have Different Jobs

The structure becomes easier to understand when each token is given a clear role.

TAO is still the base layer of the whole Bittensor network. Alpha tokens represent the economies of individual subnets. Gamma tokens would represent the products and services those subnets sell.

Alpha helped Bittensor build markets around subnets. Gamma is meant to help those subnets start doing business with each other.

By this, Bittensor moves closer to becoming a full AI economy where projects do not only compete for emissions, but also become customers, suppliers, and infrastructure providers to one another.

More details on how Gamma tokens will work are expected in the coming months, with the system set to go live in early 2027. For now, Const’s announcement gives the opportunity for new community discourse on how the economy can be built around products, services, and demand.