The TAO Opportunity: Building Conviction in Distributed AI Markets

Why we believe Bittensor is the most asymmetric bet in crypto today.

1. The Intelligence Layer of the Internet

Artificial intelligence is no longer just a tool – it’s becoming the fabric of the internet itself. From how we search and communicate to how businesses operate and governments plan, AI is quickly becoming the default interface for everything.

That raises a simple but urgent question: who controls it?

Right now, the answer is a small group of powerful companies – OpenAI, Google DeepMind, xAI, and others. They build the dominant models, set the rules for access, and decide which outputs are acceptable. These models are closed, centrally managed, and increasingly shaped by politics, profit, or ideology.

If intelligence becomes the most valuable resource in the digital economy, this kind of concentration isn’t just risky – it’s unsustainable.

Bittensor offers a fundamentally different approach.

Instead of centralising intelligence in a few labs and platforms, Bittensor creates a permissionless network where anyone can contribute compute, models, or data – and get rewarded for doing so. It turns intelligence into an open market, not a closed product.

It’s built on the same principles that made Bitcoin work: decentralisation, open participation, and economic alignment. But instead of securing transactions, Bittensor rewards useful AI – model outputs, predictions, insights, computation – through a global incentive layer.

The result is a new kind of infrastructure: not just an AI model, but a protocol for building and coordinating machine intelligence at scale.

In a world where AI is becoming essential, we believe that infrastructure should be public by design, not private by default.

2. Bittensor: The Open Market for AI

Bittensor is a decentralised protocol that turns machine intelligence into a global, open market.

At its core, Bittensor is a coordination engine. It brings together people running AI models, providing data, or scoring the quality of outputs – and rewards them based on how useful their contributions are. Instead of relying on a central authority to decide what matters, Bittensor lets the network decide. The best-performing intelligence gets the most rewards. It’s merit-based, permissionless, and constantly evolving.

The project started in 2021 with a bold idea: what if we could apply the economics of Bitcoin to AI? What if, instead of paying miners for solving cryptographic puzzles, we paid them for doing something genuinely useful – like answering questions, labelling data, or running inference?

That idea became reality.

In 2023, the network expanded through the Revolution upgrade – turning Bittensor from a single model into a platform of specialised subnets. Each subnet focuses on a different type of intelligence: language models, image detection, prediction markets, API inference, and more. They run independently, but share the same underlying token (TAO) and economic model.

Then in early 2025, Bittensor introduced Dynamic TAO – a major shift in how emissions are distributed. Now, every TAO holder can stake directly into specific subnets, and the market decides where value flows. It’s no longer about top-down control. It’s a bottom-up intelligence economy.

In short, Bittensor isn’t just building models. It’s building the infrastructure for an open AI economy – one that rewards contribution, encourages competition, and scales through decentralisation.

3. Why Centralised AI Will Fail (or Stagnate)

Centralised AI has made impressive strides – but it’s already showing signs of strain.

The biggest players today – OpenAI, Google, xAI – are building behind closed doors. They control the training data, the models, and the interfaces. What you can access is limited. What you can build is gated. And what you get out is shaped by rules you don’t see, set by people you didn’t choose.

That kind of system might work for a while. But long-term, it suffers from three fundamental problems:

1. 1. It doesn’t scale well. Only a few firms have the resources to train frontier models. Everyone else becomes a consumer, not a contributor. That limits innovation and locks out the long tail of talent.

1. 1. It introduces hidden bias and censorship. When only a handful of companies decide what’s “safe,” what’s “true,” or what’s “acceptable,” we end up with constrained outputs that reflect the politics and incentives of the companies, not the needs of the world.

1. 1. It’s brittle and fragile. Centralised infrastructure is a single point of failure. A model gets shut down, a provider changes their API, a regulator steps in – and entire industries are left scrambling.

This is where Bittensor offers a better path forward.

By distributing the work – and the rewards – across a global network, Bittensor avoids the chokepoints that plague centralised systems. Intelligence is no longer something you rent from a gatekeeper. It’s something you help build. You can contribute compute, label data, run models, or score outputs – and you earn for the value you add.

Just like Bitcoin broke the monopoly on who could issue money, Bittensor breaks the monopoly on who gets to build and monetise intelligence.

It’s not just a more open system – it’s a more resilient one. One that can grow faster, innovate wider, and resist capture by any single entity.

If the future runs on AI, we believe it needs to be built on something stronger than trust in a tech company.

It needs to be built on open markets, transparent incentives, and distributed power. That’s the Bittensor model.

4. Subnets – Modular Markets for Intelligence

Bittensor doesn’t try to build a single, all-knowing AI model.

Instead, it creates a network of subnets – modular, purpose-built markets where different types of intelligence are trained, evaluated, and rewarded.

Each subnet focuses on a specific task. Some serve large language models. Others scrape and structure data. Some detect deepfakes, label sports events, run prediction markets, or serve API calls in real time. Subnets compete for attention, capital, and rewards – and they can be launched by anyone.

In a way, subnets are like mini-startups. Each has:

• • Miners who do the work – running models, scraping data, generating outputs.

• • Validators who score performance and ensure quality.

• • Stakers who allocate TAO to support high-performing subnets and earn returns.

• • And an alpha token – a local currency that tracks the value of that specific subnet’s output.

It’s a live, open economy. The more useful a subnet becomes, the more stake it attracts – and the more emissions it earns from the protocol. Performance gets rewarded. Weak ideas get dropped. And the whole ecosystem improves through competition.

Here are just a few live examples:

• • Chutes: A serverless compute layer where developers can deploy AI models instantly across 4,000+ GPUs – no infrastructure needed.

• • TaoHash: A Bitcoin-mining subnet that’s already mined over 10 BTC, bridging real-world proof-of-work into Bittensor’s economy.

• • Score: A decentralised football prediction subnet where contributors label and validate live match events – building an open, trustless data feed that could disrupt both sports analytics and traditional betting markets.

• • Data Universe: A high-volume data scraping subnet pulling from Reddit, Twitter, and YouTube – creating structured training data for other models.

• • BitMind: Detects AI-generated images using advanced classification models – part of the fight against deepfakes and media fraud.

• • ReadyAI: Tags and enriches raw data to make it usable for fine-tuning LLMs – metadata as a service.

• • Synth: Produces synthetic financial data for agent-based simulations – ideal for AI traders and prediction engines.

Subnets launched in February 2025. There are already over 105 live, and the pace is accelerating.

The result is a permissionless marketplace for intelligence. Builders don’t need to ask anyone for access. They just show up, launch a subnet, attract stake – and compete on merit.

This is how Bittensor scales: not through a central plan, but through thousands of specialised markets evolving in parallel. It’s AI as a living economy.

5. TAO – The Currency of the Intelligence Economy

At the heart of Bittensor is a single token: TAO.

TAO isn’t just a governance token or a staking asset. It’s the monetary layer for the entire intelligence network – used to fund, coordinate, and reward the creation of useful AI.

Much like Bitcoin, TAO is:

• • Hard-capped at 21 million tokens.

• • Emitted on a fixed schedule, with a halving roughly every four years (next halving: December 2025).

• • Fair-launched – no pre-mine, no VC allocation, no insiders with preferential pricing .

But where Bitcoin rewards miners for producing valid hashes, Bittensor rewards contributors for producing valuable intelligence – and for validating it. That includes running models, serving inference, scoring outputs, curating data, and more.

Here’s how TAO flows through the system:

• • Miners and validators earn emissions by competing in subnets.

• • Subnet creators are rewarded if their subnet attracts stake and produces high-quality outputs.

• • Stakers can delegate their TAO to specific validators or subnets and earn yield in return.

• • Users spend TAO (or alpha tokens) to access AI services, inference APIs, or data feeds.

Crucially, TAO is the gateway to everything:

• • You need it to register a subnet.

• • You need it to stake and earn emissions.

• • And with Dynamic TAO, you use it to acquire alpha tokens – the local currencies of individual subnets.

TAO functions like an index on the entire Bittensor economy. You don’t have to bet on which subnet will win. You can hold TAO and own exposure to all of them – just like holding ETH gives you exposure to Ethereum’s dapp ecosystem, or BTC gives you access to the monetary premium of digital gold.

And as subnet demand grows, so does structural demand for TAO. It’s the scarce asset that everything else is built on.

In a world where intelligence becomes the core input to progress, TAO is how you own the system, not just use it.

6. Dynamic TAO – A Decentralised Capital Allocation Engine

In early 2025, Bittensor rolled out its most important upgrade to date: Dynamic TAO (or dTAO).

Before dTAO, emissions were largely decided by a handful of validators on the root network. It worked – but it was limited. Power was concentrated. Discovery was slow. Subnet quality was hard to evaluate from the top down.

dTAO flipped the model.

Now, every TAO holder can directly stake into individual subnets. In return, they receive the subnet’s alpha token – a liquid representation of their stake and belief in that subnet’s value.

The more TAO staked into a subnet’s liquidity pool, the more emissions that subnet receives. This turns the whole Bittensor economy into a live, market-driven allocation game – one where capital flows to where the network sees value.

Put differently: emissions follow price signals, not politics.

This shift unlocks three major changes:

1. 1. Stakers become allocators Holding TAO now means more than passive yield. It means making active decisions – backing teams, ideas, and models you believe in. Every staking decision sends a signal to the network.

1. 1. Alpha tokens turn subnets into investable micro-economies Each subnet now has its own asset, like equity in a startup – but liquid and yield-generating. If a subnet grows, attracts stake, and proves useful, its alpha appreciates. If it fails, it fades.

1. 1. Value accrues to performance, not politics Gone are the days of insider allocations or closed governance loops. dTAO is open, transparent, and meritocratic. High-quality subnets with real-world utility can rise quickly – regardless of who launched them.

Already, this dynamic has created a new asset class: alpha tokens. They’re fair-launched, transparent, and tied to real services – with staking yields that in some cases exceed 90% APY .

For TAO holders, this opens up two paths:

• • Stay in root staking – safe, predictable, and lower yield.

• • Or move out along the risk curve – staking to high-performing subnets and compounding returns through alpha.

And here’s the key: you’re not relying on a central emissions committee anymore. You are the emissions committee. dTAO turns every TAO holder into a fund manager – allocating capital across a network of decentralised intelligence markets.

This is how Bittensor evolves: not by governing from the centre, but by letting the market decide what deserves to grow.

7. Ecosystem Growth and Momentum

Bittensor’s vision is ambitious – but it’s not theoretical. It’s happening now.

Since launching Dynamic TAO in February 2025, the network has exploded in size and activity. Over 105 subnets are already live – each representing a real-time experiment in what decentralised AI can become.

This growth isn’t just about quantity. It’s about the diversity, creativity, and traction emerging across the ecosystem:

• • Chutes has become the go-to subnet for inference infrastructure – letting developers deploy models across 4,000+ GPUs with zero DevOps.

• • TaoHash is mining Bitcoin on-chain and feeding rewards back into the Bittensor economy – a bridge between AI and proof-of-work compute.

• • Celium offers decentralised GPU rentals at a fraction of the cost of AWS or Coreweave – powering subnet training and inference jobs.

• • Nineteen is outperforming commercial model-serving APIs – with faster, cheaper inference for top open-source LLMs.

And that’s just the start.

Behind the scenes, the ecosystem is building serious momentum:

• • DCG (Digital Currency Group) launched Yuma – a dedicated Bittensor subsidiary that’s now operating multiple subnets and validators .

• • Grayscale introduced a TAO trust – giving traditional investors a secure way to gain exposure to the network .

• • Backprop, Tao.bot, and TaoYield are creating a full analytics and trading stack – making it easier for allocators to track performance, stake capital, and trade subnet tokens .

• • Subnet tokens are being actively traded, with top subnets like Chutes and TaoHash seeing millions in daily volume .

Developers, miners, validators, and investors are pouring in – not because of hype, but because the economics work. You can launch a subnet, attract stake, generate emissions, and build a real business on-chain. No permission required.

Bittensor today feels like Ethereum in 2017 – raw, chaotic, full of noise and brilliance. But unlike most ecosystems, this one is designed to compete. The best models win rewards. The strongest subnets attract capital. And the most valuable services get built, not just talked about.

We’re past the “whitepaper phase.” Bittensor is now a functioning, multi-billion-dollar network of distributed AI markets – and it’s only just getting started.

8. Why We’re Long

At DSV, we’re building exposure to the most asymmetric opportunity we’ve seen since early Bitcoin – and Bittensor is at the center of it.

We believe this protocol is laying the foundation for a distributed intelligence economy. A network that doesn’t just use AI, but coordinates how it’s built, validated, and monetised – openly, transparently, and at scale.

Bittensor is still early. Most allocators haven’t caught on. But the structure is now in place: TAO as the base currency, subnets as sovereign AI markets, and Dynamic TAO as the capital routing engine.

Our strategy is three-fold:

1. 1. Hold TAOTAO is the monetary layer of the network – capped at 21M, fair-launched, and essential to every part of the system. It gives us protocol-level exposure to the growth of the entire Bittensor ecosystem.

1. 1. Actively Trade Subnet AlphaWe actively monitor subnet markets, allocate to high-performing alpha tokens, and rotate capital based on yield, liquidity, and product momentum. This is a liquid venture layer – and we treat it as such. Our goal is to capture early upside from real adoption, not hype.

1. 1. Strategically Partner with SubnetsWe’re forging direct relationships with select subnet teams – providing liquidity, capital, and alignment through over-the-counter deals. This gives us long-term access, deeper insight, and preferential positioning inside the fastest-growing AI microeconomies on-chain.

This approach gives us both breadth and precision: we can ride the upside of protocol-level growth while compounding alpha in specific subnets we believe will win. It also gives us flexibility. We don’t need to be right about everything – just early and right about a few.

We’re not just passive participants. We’re active allocators in a permissionless intelligence economy. Bittensor turns the act of staking into capital deployment, and TAO holders into fund managers.

And that’s exactly how we treat it.

9. Risks and Realism

Bittensor is one of the most ambitious experiments in crypto – and we’re backing it with eyes wide open.

This is early-stage infrastructure in an emerging economy. The upside is significant, but so are the risks. Here’s what we’re watching:

1. Emission Pressure

Bittensor emits 7,200 TAO per day – over $2 million in daily incentives at current prices . Until more subnets generate sustainable, off-chain revenue, most emissions will be sold to fund operations. This creates persistent sell pressure that could weigh on price in the short term.

2. Centralisation Risk

Despite its decentralised design, control is still relatively concentrated:

• • The OpenTensor Foundation remains the sole block producer.

• • The top 10 validators hold over 65% of network stake .

• • Certain subnet teams and backers (e.g. Yuma/DCG) operate across multiple layers of the stack.

This is improving – but it’s a real risk until staking, governance, and validation decentralise further.

3. Subnet Quality and Fragmentation

The permissionless nature of subnets is a strength, but also a challenge. Many will fail. Some will be duplicative. Others will simply have no product-market fit. Without proper incentive design, subnets can become extraction games, not value-creating economies.

As an allocator, filtering signal from noise is critical.

4. Monetisation Uncertainty

Today, most subnets rely on emissions. Very few are generating recurring revenue from external users. For alpha tokens to hold long-term value, subnets need real business models: API fees, licensing, premium services, or downstream integrations.

Until this shift happens, alpha token valuations will remain fragile.

5. Regulatory and Market Environment

TAO and alpha tokens operate in a highly experimental design space. The regulatory outlook for decentralised compute, AI infrastructure, and tokenised intelligence is still uncertain. And like all crypto assets, TAO remains exposed to macro headwinds, liquidity shocks, and sentiment cycles.

None of these risks undermine our core thesis. But they do shape how we allocate, where we partner, and when we rotate.

Our approach is dynamic because the landscape is dynamic. We monitor protocol updates, subnet fundamentals, validator activity, token flows, and ecosystem tooling daily. We also maintain direct relationships with builders to stay ahead of what’s coming – and what’s quietly dying.

Risk is the cost of asymmetric opportunity. And we believe the structure Bittensor is building – if it works – will be worth every ounce of volatility.

10. The Asymmetric Opportunity

Every cycle has its misunderstood asset.

In 2012, it was Bitcoin. In 2017, it was Ethereum.

In 2025, we believe it’s TAO.

Not because it’s another AI token riding a hype wave – but because it’s the monetary layer for the most important coordination protocol in artificial intelligence.

TAO isn’t a bet on a single model, product, or team. It’s a bet on a new architecture – one where intelligence is built and owned by its contributors, not its custodians.

With the launch of Dynamic TAO, the network has entered a new phase:

• • Every TAO holder can now allocate capital, not just passively stake.

• • Subnet tokens represent early-stage exposure to AI microeconomies – each with its own narrative, tech, and economics.

• • The tooling is maturing. The builders are getting sharper. The stakes are rising.

We’re still early. Valuations are messy. Monetisation is scarce. But that’s exactly where asymmetric upside comes from. You don’t need to bet on every subnet. You need to be right about a few. And TAO gives you both broad exposure and targeted leverage.

At DSV, we’re not here to follow the narrative. We’re here to front-run the understanding.

We believe Bittensor will be one of the defining protocols of the next decade. If the AI economy becomes open, Bittensor will be its backbone. If it doesn’t, it won’t be for lack of trying – and the network will have created a new blueprint for coordination in any sector that relies on open, verifiable contribution.

This isn’t just a new market. It’s a new structure for how intelligence is discovered, rewarded, and owned.

That’s the opportunity. That’s the edge. And we’re all in.