‘On Bittensor, we’re speed-running humanity…
corporation, platform, federation, religion, nation state.’
- Jacob Steeves, co-founder | Bittensor

Last week, Jake spoke at two of the world’s best universities: on Friday, at Imperial College London, and on Sunday at Queens College, Cambridge.

Const showcased Targon, Quasar, Chutes, Lium, Nova, Swarm and Oro - despite not using the term ‘subnet’ until almost an hour in.

But these talks were deeper than a product or subnet demo. More than in any other speech, Jacob attempted a complete explanation of Bittensor: theory, practice, past, and future.  

At Imperial, Jake’s talk sparked an hour and a half of non-stop Q&A. At Cambridge, Jake rearranged the talk, challenging the audience to think through Bittensor’s conception from first principles. And it worked.  

Each was excellent - here are the five key points from across both talks (but unfortunately, for a summary of his third talk at Oxford you’ll have to look elsewhere).

Defining Bittensor is hard. Its complexity means that, at some point, we resort to simply calling it ‘the biggest DeAI project’ or ‘incentivizing AI on the blockchain.’ 

But one of Jake’s most striking points was that combining these two technologies - the defining technologies of our time - creates a third element, one which is entirely new. 

Blockchain is a technology of immutability, of facts set in stone: twenty-one million Bitcoin, never more, your transaction final. AI is the opposite: weights that mutate, loss that flows, a structure adapting itself through feedback. They are philosophically and technically opposed. 

Bittensor doesn’t glue them together. It unifies them at a deeper level: sophisticated AI-grade validation attached to blockchain-grade monetary feedback. And what emerges belongs in neither of its parent categories.

“I would say that we're a blockchain company that deeply understood what artificial intelligence truly was, and integrated the two,” Jake said, in the Cambridge talk. “It's neither AI or blockchain. It's a unification of the two ideas."

Bittensor isn't a ‘bring-AI-on-chain’ project, but an ambition to generalise Bitcoin’s breakthrough, beyond the encryption of decentralized money. Bitcoin demonstrates that permissionless, borderless, 24/7 monetary feedback can organise a globally distributed effort with extreme efficiency. 

This is the building block inherent to Bittensor’s core concept, and in Jake’s depiction, one of nature’s most powerful forces. From slime moulds to thermodynamics, natural selection to reinforcement learning, Jake demonstrated how this feedback loop generates unstoppable development of its own accord. 

Bitcoin was the first feedback loop to organise a decentralized, globally distributed effort at scale with monetary incentives. Borderless, 24/7, autonomous, permissionless - the qualities that make it a structurally different coordination mechanism from any company or state. But Bitcoin lacks sophistication: it organises hash production, a problem trivial to validate. 

What has changed in the last two years is that AI itself can now act as the validator. Karpathy's coding agents grade their own ML experiments; the validation layer becomes arbitrarily sophisticated, which extends the loop's reach from neural weights to agents to companies to economies.

The two-axis design space: sophistication of validator (loss function → AI judge) on one axis; generality of the state being adapted (neural weights → agents → code → companies → economies) on the other.

Two axes define the entire design space - sophistication of validator, generality of state - and Bittensor's ambition is to push toward the upper-right corner.

"The ultimate feedback loop would be something where the feedback is really, really, really, really sophisticated… And that the structure is general… What if we could organise all state and use it with an AI?"

Yes, the wisdom of the crowd is heavily skewed by liquidity asymmetries, founder networks, and social media trends.

But Jake’s point was that this isn't a flaw in the system - it is the system. The market is partly measuring narrative because narrative is partly a signal of execution capacity. It’s not infallible, but as a heuristic it’s a good general indicator.

Where it got interesting was in the Q&A. Markets tend toward Pareto distribution, but specific design choices - registration cost, dTAO emission curves, dividend mechanics - amplify or muffle that tendency.

Bitcoin's deepest contribution wasn’t wealth creation or price discovery; it was the demonstration that escape from fiat capture was possible. Bittensor is a wager that the same escape is possible for AI - that ordinary participants do not have to watch a handful of trillion-dollar labs absorb all the value of the most consequential technology of the century. 

The features that make this possible are the ones that make the protocol odd-looking from the inside: visible governance, public ownership, a participation surface that includes mining, validating, building, holding. 

The point is not that Bittensor will out-compete the closed labs on any single benchmark. The point is that it gives people a stake - one worth fighting for. But it’s worth leaving the last word to Jake, who puts it like this:

"Our goal is to give people that have no hope in that future, a sense of hope and participation."