In a presentation beneath the glass pyramid of the Louvre in Paris, the co-founder of a crypto project called Bittensor made a bold claim: the enormous computing power currently used to mine Bitcoin could instead power the next generation of artificial intelligence. Here’s why that idea could reshape both industries.
By Tomas Novak | June 2, 2026
Here’s a number that might surprise you: Bitcoin mining uses more electricity than many small countries. All that power goes toward solving complex math problems that secure the network. But what if that same computing infrastructure could do something more productive — like training AI models?
That’s the pitch Ala Shaabana, co-founder of Bittensor, delivered at the Proof of Talk summit in Paris. Bittensor (ticker: TAO) is building a network where people contribute computing power to train AI models and earn tokens in return — similar to how Bitcoin miners contribute computing power to secure the network and earn BTC.
What Is Bittensor?
Think of Bittensor as “Bitcoin for AI.” Instead of mining a cryptocurrency, participants mine intelligence. The network is divided into 128 “subnets” — specialized groups that focus on different AI tasks. One subnet might specialize in image generation, another in language translation, another in code writing, and so on.
People who contribute computing power or useful AI models earn TAO tokens. The better your contribution, the more you earn. It’s a decentralized marketplace for AI, where no single company (like Google or OpenAI) controls all the computing power or the resulting models.
Why Does This Matter?
AI needs enormous computing power. Training a model like ChatGPT costs millions of dollars in computing resources. Right now, a handful of tech giants control most of that infrastructure. Bittensor’s vision is to democratize it — let anyone with a computer contribute and earn.
The crypto industry needs a new narrative. Bitcoin has proven itself as “digital gold” — a store of value. Ethereum powers decentralized finance. But the market is hungry for the next big use case. “Decentralized AI” is the leading candidate. If Bittensor’s 128-subnet architecture works at scale, it could be the infrastructure that connects crypto and AI in a meaningful way.
The Louvre setting matters. Presenting at one of the world’s most famous cultural institutions, at a conference attended by financial elites and tech leaders, signals that decentralized AI is being taken seriously at the highest levels. This isn’t a fringe crypto project presenting in a hotel basement — it’s a legitimate technology pitch in a prestigious venue.
The 128-Subnet Architecture Explained
Bittensor’s structure is what makes it unique. Instead of one giant AI model, the network has 128 specialized subnets, each competing to be the best at a specific task. Competition drives quality — if your subnet produces better image generation than another, you earn more tokens.
This is similar to how free markets work: instead of one government-run factory making everything, many companies compete, and consumers benefit from better products and lower prices. Bittensor applies this logic to AI development.
The Risks
- TAO is still a small, volatile token. It’s not Bitcoin or Ethereum. Its price can swing wildly based on hype and sentiment.
- Decentralized AI is unproven at scale. Can a network of distributed computers really compete with Google’s and OpenAI’s massive, centralized data centers? That’s still an open question.
- Regulatory uncertainty. Both AI and crypto face increasing regulation. A project that sits at the intersection of both could face challenges from multiple directions.
The bottom line: Bittensor represents one of the most interesting ideas in crypto right now — using blockchain incentives to build decentralized AI. Whether it succeeds depends entirely on execution. But the concept of “mining intelligence instead of tokens” is compelling enough that it deserves attention from anyone interested in where crypto and AI are heading.
The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.
128 subnets competing on intelligence output is a genuinely different approach. Most AI projects are racing to build one model to rule them all. Bittensor is basically betting that competition beats consolidation.
shaabana comparing bittensor compute to 600k supercomputers is wild if accurate. would love to see independent verification of that number tho
the 600k thing is about btc hashpower not tao lmao. whole point is applying that same incentive model to AI compute. im long TAO either way
The fair launch narrative is compelling but 21M cap and halvings dont automatically make something valuable. What matters is whether subnets actually produce useful intelligence or just burn tokens competing.