Buenos Aires, March 2028 - Mateo’s alarm goes off at 6:40. He’s thumbing through his dashboard before his eyes are fully open. Overnight, he made 1.8 TAO across the two subnets he’s actively mining. Earnings down 20%. He rolls over, mildly exasperated, tossing his phone onto the empty pillow beside him. Time to start the day.
He shakes off sleepiness, stretches, and mumbles to himself damn Hangzhou… Competition has been brutal since half of China's research talent showed up on the network.
He pads across the cool wood floor to the glow of his terminal. His agent has been busy.
Overnight brief, 6:30 AM
11:04 PM: New paper published on reducing GPU memory traffic during inference. Summary and PDF saved.
1:12 AM: An inference optimization subnet opened a new challenge. Top three implementations of this paper earn the bulk of the rewards.
1:20 AM: Entered. Reproduced the paper’s result; tested 12 implementation variants.
Current rank: 41 of 212. Not earning.
Two promising candidates. Larger validation run recommended.
Mateo pulls a double espresso, drops into his Aeron, and opens the experiment history. His agent has reproduced the paper's result, rewritten the relevant code for the competition's model, and burned the night testing alternatives.
Most failed to beat the benchmark. Two candidates remain, and the agent wants to run both. Mateo picks one. He's seen this before, back when he was a lead research engineer in the perpetually foggy Bay. The faster-looking candidate wins on benchmarks but would likely fall apart on real traffic, and the competition measures real traffic.
He fires off the command to run the other candidate, with a few tweaks of his own. Validation takes most of the day.
At 1:40 AM, his phone lights up. Third place.
Mateo loved his first years at the lab. Good work, good money, models he was proud of. But by the end, the frontier models he worked on never got released. A handful of government approved orgs were permitted access, while the public got some nerfed version. That just felt wrong to him.
Then his program went classified. New campus, somewhere in the desert. They called it a matter of national security and told him to be proud.
So he left the lab, and the Bay with it. Now he's in Argentina, anon behind a pixelated dart frog, mining full time. His competitors, collaborators and closest friends are scattered across a dozen time zones, and he has never met most of them. They argue in the same group chats, fork each other’s work, and rely on a protocol for a living. The work is more chaotic than anything he did at the lab. But he answers to no one, and everything he builds is open the moment he submits it.
There’s been no shortage of work.
U.S., EU BAR GOVERNMENT AGENCIES AND CONTRACTORS FROM USING CHINESE-ORIGIN OPEN-WEIGHT MODELS | Bloomberg, August 2027
Almost overnight, the West needed to replace the dealer it had gotten hooked on. Bittensor had spent years building an entire pipeline for training open models. China got to the frontier by being resource constrained and innovating around it. Bittensor's training subnets won the same way, and now produce some of the best open-source models in the world.
Two years ago, Mateo would have laughed at the idea of working for a nation with no borders. Now this is the only place he can work on frontier AI in the open. He isn't the only one who needed somewhere like this to exist. The labs were always going to serve their shareholders and their governments. Bitcoin gave people an alternative for money. Bittensor is giving them one for intelligence.
He checks his rank again. Still third. He gets back to work.
The Nation Today
Mateo is fiction. The nation he works for isn’t. For the past three years I've watched Bittensor build an economy of its own. It took me a while to see where it was heading, but now I can’t unsee it. Bittensor is becoming a network state.
Today, ~7,400 mining entities, ranging from solo operators to organized teams, compete across 128 subnets. In a few years, I expect to see miner count 10x, and subnet count 3x.
Bittensor’s Factories
The heart of any nation is what it produces and exports. Bittensor's subnets are its factories, supplying services and commodities like compute, trained models and security solutions. Each has its own token that gives holders exposure to that subnet's economy.
Every factory starts with the same advantages, built into the protocol itself. Think of them as the nation's public institutions. A labor market, where miners around the world compete to do the work and supply the cheapest digital resources. A capital market, where every subnet's token is listed and liquid from day one. And shared standards to simplify and accelerate development.
Then there is the protocol’s sovereign wealth fund. Every day the protocol emits TAO, and factories compete for it. Relative token prices determine how those emissions are split across subnets. A portion supports each subnet's liquidity, while the majority goes to buying its token. Capital flows to the factories that generate the most value.
Together, these give every factory an edge from day one. The biggest is the labor market. A founder can pull the strongest talent for their exact problem and the cheapest resources to run it, without hiring anyone. And because every factory is priced in TAO, one factory winning lifts all other factories and in the nation. That's what pulls founders here.
Factories will come and go. The weak ones get replaced by founders who execute better, the same way they do anywhere. What stays is the incentive underneath. Whenever the nation needs storage, compute, training or anything else, someone is paid to go build it, and the best builder wins.
We categorize factories within Bittensor into three groups.
Infrastructure
These are platforms, built so other factories don't have to build everything themselves or buy it from outside. A new subnet doesn't need AWS or to engage a security team for audits. It can plug into Hippius for storage, Targon for cloud services, Lium for GPUs and BitSec for code audits. That makes it faster and cheaper to start a factory in Bittensor, and every dollar spent stays inside the nation. Infrastructure subnets also sell to customers outside Bittensor, working both markets at once.
Sovereign AI Lab
The second group, and most important in our eyes, is the nation's sovereign AI lab. It runs on that same infrastructure, then adds every layer needed to produce intelligence. Pre-training on Teutonic and Iota. Post-training on Hone, Connito and Affine. Inference on Chutes and Engy. No closed lab or service is required at any step. Nations everywhere are scrambling to build a sovereign AI program, while Bittensor already has one.
Exporters
On top of both sit the exporters, or apps. This is where Bittensor becomes a platform for any idea. If you can measure it, you can build a subnet around it. Founders plug into the nation's infrastructure and AI lab and let miners compete to produce what they specify. What comes out is a competitive service or commodity, sold to customers outside Bittensor.
Minos is a good example. It focuses on selling synthetic genomic data outside the nation, but utilizes the nation’s infrastructure and protocol benefits to be competitive. They use Hippius for storage and provide miners with ready-made virtual machines through Targon.
To grow, every factory has to bring its outside revenue home. Meaning revenue needs to flow back into the token for the business to scale. A higher token price pays miners more, which attracts better miners and more outputs. Better outputs win more customers, and more customers mean more revenue. More on this here.
A Nation By Choice
Bittensor's ambition is to sustain an economy that can produce intelligence beyond the control of any single company or government, while letting anyone build markets, businesses, and livelihoods around it.
For Mateo and others like him, that means another path. They can earn a living applying their expertise to a mission they believe in, alongside people they may never meet. Each contribution helps build the productive capacity of a nation they have chosen to participate in.
That is the promise of Bittensor as a network state. A nation you can help build, wherever you live.
