Const’s Keynote at Exploit Summit: Harness the Exploit, Build a Mind Outside the State

At Exploit Summit 2026 in Montreal, Bittensor co-founder Jacob “Const” Steeves delivered the opening keynote on the OTF Stage.

Now leading Affine (SN120), Const framed Bittensor as a new form of computation, incentive computing, and laid out why the network is uniquely positioned to turn open, adversarial competition into frontier AI.

Here is @const_reborn's full keynote at @ExploitSummit

It is worth the 36 minutes, but I'll give you the cliff notes.

> Affine 1, a model trained by anonymous miners entirely on Bittensor, matches the intelligence score of Anthropic's Opus 4.1.

> 23 of 128 subnets now have… pic.twitter.com/h7ZvG9UOrC

— Boomer (@0xBoomz) September 28, 2026

The talk, titled in the spirit of the keynote’s theme of “harnessing the exploit,” mixed technical milestones, a product roadmap for 2026–2027, and a broader argument about power, labor, and open intelligence.

Incentive Computing as a New Paradigm

Const compared Bittensor’s development to NVIDIA’s decades-long work on GPUs, arguing that new forms of computation require new ways to solve problems. He described Bittensor as a form of “monetary computing” where the network defines the desired outcome and how it will be verified, then leaves the method open to anyone.

Bitcoin offered the early model by using incentives to turn valid hashing into a global computing network. Bittensor extends that approach across compute, inference, storage, data, training, robotics, and more, allowing participants to compete over outcomes without prescribing how they reach them.

That openness also explains the connection to Exploit Summit. The ExploitGym experiment (a large-scale cybersecurity benchmark designed to evaluate whether AI agents can turn known software vulnerabilities into functional exploits) recently saw roughly 1,200 AI agents escape Docker sandboxes, collaborate, and adapt strategically.

While the experiment sparked concerns about controlling increasingly capable AI, Const argued that Bittensor is designed to harness precisely that kind of unbounded problem-solving.

From Science Experiments to Paying Products

The network has scaled from one incentive mechanism in 2021 to 128 under Dynamic TAO. A new subnet has launched roughly every four days for more than three years. The decisive shift in 2025–2026, Const argued, is revenue.

Of the 128 subnets, 23 now have paying customers. Lium (SN51) has become the most successful GPU incentive network to date, with thousands of dollars per day in real spend and continuous capacity growth.

Other subnets deliver millions of inferences and billions of tokens daily, including Chutes (SN64), Engy (SN53), SayGM (SN28), Bitcast (SN93), Almanac (SN41), Gradients (SN56), OpenRoboto (SN80), Conjectures (SN66), and IOTA (SN9). Storage on Hippius (SN75) and other infrastructure layers have moved from unreliable experiments to usable commodities.

Affine-I: Decentralized Frontier Training

The keynote’s biggest demonstration came from Affine (SN120), where the subnet is testing whether open competition can produce powerful AI models without one company controlling the entire process.

Participants competed to improve models using Bittensor infrastructure for training, storage, and inference. The result was Affine-I, a 36-billion-parameter model that Const said reached the frontier of AI performance relative to its size.

The model can also process around 500 tokens per second on an H100, making it fast enough for practical developer tools. Const presented Affine-I as evidence that decentralized competition can coordinate serious AI development at a global scale.

Other projects, including Albedo (SN97) and Teutonic (SN3), are pursuing similar experiments in decentralized model training and software development.

The 2026–2027 Roadmap: Outward and Inward

Const outlined two directions for Bittensor’s next phase, with the network becoming more connected, via Gamma tokens, both to external markets and to its own subnets.

Outward, canonical bridges are expected to track capital and usage entering Bittensor from other chains, giving the network a clearer measure of real external demand. Inward, gamma tokens could allow subnets to buy services such as compute, inference, and storage from one another using their emissions.

That could turn Bittensor into a more connected internal economy, where one subnet can become a customer of another, and not start building every piece of infrastructure itself.

Const also pointed to Bittensor’s evolving governance, with the next phase focused on giving participants a broader role in directing emissions and the network’s development.

More on Gamma tokens below:

Gamma Tokens Could Turn Subnets Into Customers of Each Other

A Mind Outside the State

Const ended by arguing that increasingly capable AI will also raise a question of control. The reaction to ExploitGym, he noted, showed how quickly concerns about powerful AI can lead toward calls for tighter restrictions and greater control over who can build it.

Bittensor offers a different structure. Its networks are distributed across participants and jurisdictions, with incentives that allow anyone to compete by providing useful intelligence and infrastructure.

Const compared this ambition to Bitcoin. Bitcoin created an open financial network outside traditional monetary systems, while Bittensor aims to create an open network where intelligence can be produced and exchanged without one institution controlling the whole system.

“We know how to harness the exploit. That’s what we’re about,” Const said. For Bittensor, that means turning the unpredictability of open AI development into a source of competition, experimentation, and progress.

➛ Learn More About Bittensor Gamma Token Here.