# Bitcoin.com News: "Can Bittensor Beat OpenAI? Inside the Race to Decentralize AI"

- Title: Can Bittensor Beat OpenAI? Inside the Race to Decentralize AI (podcast title: "Inside Bittensor: The Race to Decentralize AI")
- Channel: Bitcoin.com News (host Alex Richardson)
- Date: 4 August 2026
- Length: 53:12 (podcast audio)
- URL: https://www.youtube.com/watch?v=AH-pzoRxEtc
- Jacob's start offset: none given; a cold-open montage of Jacob's lines runs 0:00–1:20, the interview proper starts at 1:23
- Method: YouTube blocked from this VM. The "Bitcoin.com News Interviews" podcast RSS feed (Anchor) carries the same episode; audio downloaded and transcribed locally with faster-whisper `medium`. Timestamps from the podcast audio, accurate to a few seconds.
- Format: research transcript. Host paraphrased; Jacob's key lines quoted verbatim.
- Speakers: HOST = Alex Richardson; JACOB = Jacob Steeves.

## Outline with timestamps

0:00 Cold open: four of Jacob's lines cut together (Bitcoin as the largest computer; "a form of totalitarianism that we've never seen before"; "mining for intelligence directly").

1:23 HOST intro ("a true gigabrain"). Asks about the DARPA contractor who was "an OG Bitcoin whale."

1:50 JACOB: the boss "gave his sister a thousand Bitcoin to get her into Bitcoin." Backstory of neuromorphic chips ("that field is called neuromorphic computing because it's computers that are like the brain"), the DARPA SyNAPSE program funding IBM's TrueNorth, and the boss leaving to start his own company, named as "Gnome" in the transcript (Knowm). Explains memristors and the tree analogy: trees "branch and then they branch again... trying to find where they can get the most amount of light... a physical analogy to a searching algorithm in nature."

5:29 First Bitcoin for a blog article, 2015: "I was like, what the hell is that?" He thought mining was "crunching prime numbers." The boss "taught me about Bitcoin at a very deep level, a philosophical level... It's a freedom technology."

6:50 Why compute mattered: "Anthropic raised $60 billion in the last quarter... a good 70, 80% of that is just going into compute." Bitcoin's two sides: "freedom technology, something that we can use to pull ourselves out of fiat debt slavery," and "as a consequence of its design, the largest supercomputer in existence." "How is it possible that this technology has accidentally created the largest supercomputer on earth."

8:18 The numbers: "a thousand times larger than every supercomputer on earth combined... for about $50 billion if you were to sell every single Bitcoin at the price they were mined... about 9,000 times more efficient." Why: "a permissionless 24/7 liquid market for something very well defined... a metric based and permissionless market." Paraguay miners on rivers.

9:42 "I like to call it a monetary computer." Miners "need to beat the previous guy in terms of hashing power or they fall from the network." "If you were to create a company and be like, let's produce Bitcoin hashing, you'll never be better than the mechanism that selects companies that does the thing."

11:02 The founding question again: "The most expensive computational problem should have the most powerful computational primitive." The abstraction: "take Bitcoin mining and pull out everything to do with hashing and transactions and just think about what this monetary computer would look like." Examples: data scraping, inference (Chutes, Engy, Cortex) paid "in proportion to their speed or their inference bandwidth."

13:44 Why it took time: "it is an art to build these mechanisms." In a company you can see and sue people; in crypto "you actually have to get down to the very precise measurement of the thing. And not everyone has the brain for that." "we turned ourselves into basically the world experts in all of these problems."

15:07 The pinnacle: "imagine Bitcoin mining, but what you're mining is a machine. You're mining for intelligence directly." "an artificial intelligence that is collectively built by people, Bitcoin miners effectively." Against the lab release cycle: "a market where at any point people are mining the next block of intelligence. That's a release for the latest model every 12 seconds."

16:29 The why: "It doesn't matter about me. It doesn't matter about you because they're permissionless at their core." On ownership: "You know Anthropic and you know OpenAI, but you don't really know who runs those companies and you don't really know who owns them." "that occlusion, that occult nature of ownership and control is a very big problem for corruption." "we're building an AI company, but we're doing it with the guts out." "this cabal of geopolitical insider trading to own this technology."

19:18 Ad. HOST asks for an idiot-friendly architecture explanation.

19:50 JACOB: a subnet is "an encapsulation of Bitcoin," defined by "the function that translates the outputs from the miners into a reward." GPU example (Lium, Targon): "instead of measuring how many SHA-256s... we measure how many GPUs you have available for our customers." Permissionless on both sides: no KYC to rent; "You need to know somebody to get some of these GPUs." Validators are "computers that are continuously pen testing all of the compute." Two levels: subnets, then the chain that pays subnets by price and sequestered TAO. "You can think of it as just kind of a tree with two levels."

23:25 HOST: is mining harder than Bitcoin? JACOB: "Not anymore. So yes and no." Mechanisms change, "which is annoying. It's also kind of our superpower." Some tasks are "come up with an algorithm that solves the traveling salesman problem faster than the world state of the art." Bitcoin mining may be harder now because it is converged. "there's like a really beautiful science or like art to the creation of these mining networks." Now: "if you are somebody that knows how to use Claude Code or Cursor... you can more or less just tell your agent to go and try to mine these subnets on Bittensor and make pretty good money."

26:11 HOST: why start a subnet rather than a company? JACOB tells the Cohere story: he wanted them to build Bittensor with him; they built a classic company. "there was basically no way that we'd be able to beat nation-state fiat money from Canada... So what exactly is our edge?" "if you don't have at least something, you're just going to lose to the people with more money." With a mechanism at the core: "you have a workforce at your disposal, which is guaranteed to be the best in the world and is constantly cycling out the best of the world. You have a machine that actually can move faster than any corporation on earth." Bittensor as "a hyper efficient incubator program and also permissionless incubator program." Double edge: "we also get the worst of the worst hackers."

30:18 HOST: which subnets win? JACOB: product-first subnets where the mechanism is an afterthought "has never worked very well." Winners: big commoditised markets (compute, inference, storage) and "those that understand that you can use Bittensor to do real machine learning work... you don't even need to be any good at that yourself."

31:43 HOST: the Templar/Covenant exit after the Jensen moment. JACOB: "they rugged. Of course, they rugged." "a combination of insanely bad timing and really malicious behavior on behalf of somebody that had earned a lot of respect in the community over a long period of time through my help." Takes blame: "a lot of fault actually lies at my own feet for enabling the engineer Sam Dare, who I promoted as basically a general... What I should never have done was also give him the keys." "how do you explain insanity?" On why rug pulls exist: "we built these things to be permissionless." "because we're not inside the jurisdictions that have investor protection rules, people get hurt... this is the price we pay for choosing to be so courageous as to go outside of the standard rules." "all that the engineer did was just sell his own shares." Recovery: three weeks from regaining control of the subnet; a model already trained permissionlessly that outperforms the one that got attention. "nothing fundamentally changed, no damage was really done."

37:16 Conviction upgrade: teams can lock funds to signal commitment; "We didn't enforce that." Also gives alpha holders "voting rights, let's say, over the subnet itself."

38:38 HOST: how does TAO accrue value? JACOB: subnets pay "platform fees to the chain... as a dividend to holders of TAO," "about $100 million per year of dividends to about $300 million worth of inflation. So we're in an inflationary period." You must buy TAO to buy alpha. The dollar analogy again. Next: tie the commodities directly into TAO "in the next six to eight months."

41:24 HOST on Venice's token-plus-equity. JACOB: "it creates a disalignment between the executive team and their token. It's why there's no equity for TAO. Like my equity is TAO. I am TAO and Ala Shaabana, my co-founder, is also TAO." Two tokens create "two privileged classes." "the first commandment is like, you shall have no other God but me... It's very bad to split people's psyche around where the value is."

43:10 HOST on Algod's criticism. JACOB: "I believe in epiphenomenal facts of markets. And he's extremely cynical about them." "it's a good sign to have outspoken people that care in your ecosystem."

44:50 HOST: the Elysium warning; a breakaway technocratic civilisation. JACOB's fullest statement of the fear on tape: "we risk, maybe for the first time ever in history, a form of totalitarianism that we've never seen before." Why now: "forever, humans have held a lot of weight against the emperor... they have more feet and hands and they can wield a sword." "war is becoming incredibly mechanized... you can be a tyrant that loses 90% of your workforce potentially, and it doesn't even affect your economy." "this balance of power between the mass of individuals and the power holders, that balance is drastically changing with artificial intelligence." He names it: "intelligence fascism... because there never has been anything like this before." "an unholy union between nation-states and their tech giants," with the current US administration "playing them against each other." "there will be more or less nothing that the average individual can do." The answer: "reorganizing ourselves into systems that have more transparent and distributed ownership and ground up production." And the condition: "it doesn't matter if it's decentralized and it's permissionless and it's transparent if it doesn't do anything." "the beginnings of a top down intelligence fascism in both China and the United States." "We want to build it by organizing humans and stitching us together." "if Bittensor can't do it, I think that there's a lot of people in this ecosystem that will do it."

49:40 HOST: how does Bittensor get users from ChatGPT and Claude? JACOB: users already exist on Chutes and Engy. "Where we have not been able to compete yet is just at the SOTA frontier." "what we've built in 2026 is machines that go extremely fast in the directions we point them. And all we truly need is to align these machines perfectly." "we've shown that we can beat them on more narrow band problems, mining compute, mining inference, mining storage. We can beat these giants. So it's only a matter of time."

52:19 Close: "everyone should mine Bittensor. Like if you know how to use an AI, you should get involved."

## Jacob's stage-worthy lines (verbatim, Whisper medium)

- "How is it possible that this technology has accidentally created the largest supercomputer on earth." (8:10)
- "The most expensive computational problem should have the most powerful computational primitive." (11:02)
- "you'll never be better than the mechanism that selects companies that does the thing." (10:30)
- "imagine Bitcoin mining, but what you're mining is a machine. You're mining for intelligence directly." (15:20)
- "a market where at any point people are mining the next block of intelligence." (16:29)
- "You know Anthropic and you know OpenAI, but you don't really know who runs those companies and you don't really know who owns them." (17:20)
- "we're building an AI company, but we're doing it with the guts out." (18:20)
- "You can think of it as just kind of a tree with two levels." (23:20)
- "what exactly is our edge?" (28:00)
- "You have a machine that actually can move faster than any corporation on earth." (29:00)
- "this is the price we pay for choosing to be so courageous as to go outside of the standard rules." (35:30)
- "my equity is TAO. I am TAO." (41:45)
- "we risk, maybe for the first time ever in history, a form of totalitarianism that we've never seen before." (45:50)
- "humans have held a lot of weight against the emperor." (46:00)
- "you can be a tyrant that loses 90% of your workforce potentially, and it doesn't even affect your economy." (46:30)
- "an unholy union between nation-states and their tech giants." (47:05)
- "it doesn't matter if it's decentralized and it's permissionless and it's transparent if it doesn't do anything." (48:40)
- "We want to build it by organizing humans and stitching us together." (49:15)
- "machines that go extremely fast in the directions we point them." (51:10)