Bittensor’s Hidden Growth Engine: The Rise of Subnets

13 min readMar 4, 2025

Setting the Stage

Bittensor is a decentralized AI marketplace that shifts control from big corporations to a global community, rewarding contributors for advancing AI. By turning intelligence into a tradable asset, it fuels open innovation and rapid knowledge sharing. Its ecosystem runs on specialized subnets, each focused on different AI applications, constantly expanding the network’s capabilities.

This research provides the community with a strategic advantage by analyzing the network’s current landscape, recent progress, key partnerships, and its potential as a decentralized hub for collaborative AI intelligence.

Key Insights

Market Positioning: Bittensor has established itself as a leader in decentralized AI, strengthening both market cap and market share while expanding its ecosystem and use cases.
dTAO Upgrade: Resolves inefficiencies in $TAO’s 7,200 daily emissions, ensuring rewards flow to high-quality subnets while reducing centralization.
Subnet TGE Potential: With more than 60 subnets, upcoming TGEs could drive a significant repricing for $TAO.
Sustainability & Growth: Dynamic $TAO improves long-term scalability by optimizing reward distribution and aligning incentives across the network.
Rising Adoption: TAO now has over 203,000 accounts and growing, reflecting increased participation, stronger decentralization, and a rapidly expanding AI economy.

Strategic Partnerships: Backed by DCG, Yuma’s capital and industry connections could unlock major value for Bittensor’s subnets, while its partnership with Virtuals positions the network to tap into the $15.2 billion AI agent market.
Decentralized AI Advantage: Bittensor’s AI enables collective intelligence, challenging centralized models like OpenAI if successful, with scalability and network effects that could reshape AI’s future.

Before we dive in, a special thanks to GLC Capital for their insights and collaboration on this research. If you’re not already following them, they’re definitely worth keeping an eye on.

Market Opportunity

Rapid Market Growth: The global AI market is expected to expand dramatically — from approximately USD 294 billion in 2025 to about USD 1.77 trillion by 2032 (≈29% CAGR), while Statista projects growth from USD 243.70 billion in 2025 to USD 826.70 billion by 2030.

Broad Economic Impact: PwC’s study estimates that AI could add up to USD 15.7 trillion to the global economy by 2030, driven by enhanced productivity, cost efficiencies, and new revenue streams.

Investment Momentum: Goldman Sachs projects global AI investments to approach USD 200 billion by 2025, fueling infrastructure and innovation.

Bittensor serves as an open-source coordination layer where subnets earn rewards by contributing resources and intelligence. The introduction of Dynamic TAO for tokenomics improvements further strengthens this model, positioning Bittensor to potentially capitalize on the explosive growth of the AI market by delivering scalable and cost-effective AI solutions.

Bittensor: A Decentralized Network for Collaborative AI Intelligence

Founded in 2019 by AI researchers Ala Shaabana and Jacob Steeves, Bittensor was originally envisioned as a Polkadot parachain but pivoted in March 2023 to build its own blockchain. Using cryptocurrency to incentivize a global network, it enables decentralized AI development by allowing nodes to collaboratively train and grow collective intelligence.

Today, Bittensor is widely regarded as one of the leading AI projects in the crypto space. As of February 25, 2025, its market cap is roughly $3.31 billion, and its fully diluted valuation is about $8.26 billion.

Bittensor operates as a decentralized network of subnets, each serving as an incentive-driven AI marketplace. Subnets consist of miners, who generate digital commodities, and validators, who assess their quality. Each subnet has its own goals and reward mechanisms.

The Bittensor blockchain records transactions and subnet performance, while $TAO incentivizes miners and validators, ensuring continuous competition and quality improvement. Stakers can delegate TAO to validators, further aligning incentives.

To support interaction, Bittensor provides an SDK, along with open-source tools and documentation for subnet development.

Key Roles in the Network:

Miners: Generate AI-driven digital commodities.
Validators: Evaluate and rank miners’ output.
Subnet Creators: Manage incentive structures.
Stakers: Support validators by delegating TAO.

Bittensor runs on Proof of Intelligence (PoI), a consensus model that mixes elements of PoW and PoS. Instead of solving math problems like PoW, miners complete machine-learning tasks — those with better AI models earn more TAO.

Miners work within subnets connected to the main network through the Bittensor API. Validators secure the network by staking TAO, similar to PoS. With a fixed 21 million TAO supply, rewards are gradually released, keeping miners and validators engaged.

Dynamic TAO: Transforming Bittensor’s AI Economy

Initially, Bittensor operated on a single-token system, where $TAO was the only unit of value and governance. Miners provided AI services, validators ensured network integrity, and users accessed AI-driven solutions. All staking, rewards, and governance depended solely on $TAO, meaning network influence was concentrated among $TAO holders.

What Went Wrong?

Centralized Control: A few large $TAO holders had outsized influence over emissions and governance, limiting the network’s decentralization.
Lack of Specialization: The same reward system applied across all AI subnets, making it difficult for niche AI applications to thrive.
Static Incentives: The rigid staking model meant AI innovation and validator selection were not fully market-driven, limiting organic subnet growth.

Enter Dynamic TAO

Dynamic TAO was introduced in early 2024, with its initial framework outlined in BIT001. After a year of development and testing, it officially launched on the Bittensor mainnet in February 2025.

The upgrade introduced subnet-specific alpha tokens, enabling each subnet to function independently with market-driven rewards, staking, and governance. This enhanced decentralization and improved AI adaptability.

Dynamic TAO has significantly transformed Bittensor’s tokenomics and governance. The key changes include:

Subnet-Specific Alpha Tokens: Each subnet now has its own “alpha” token, traded against TAO on a constant product AMM. This enables subnets to establish independent economies and liquidity pools, where market value determines resource allocation.

New Staking Mechanism: Instead of staking TAO directly to validators, users now stake TAO on a specific subnet. For mining subnets, staking involves exchanging TAO for the subnet’s alpha token, while on Subnet Zero (the root subnet) you still stake TAO for TAO. This change makes staking sensitive to the market dynamics of alpha tokens.
Market-Driven Emissions: TAO issuance is now tied to the market value of each subnet’s alpha token, allocating more emissions to highly liquid, in-demand subnets. As alpha token supply expands, its impact on validator weight grows relative to TAO.
Dynamic Validator Weighting: Initially, validator weight is determined entirely by TAO stakes because no alpha tokens are in circulation. However, as alpha tokens are gradually introduced and staked, a parameter called “TAO weight” adjusts the influence of TAO versus alpha.
Halving and Emission Adjustments: Both TAO and alpha tokens follow a halving schedule, controlling token supply and inflation predictably. Alpha is issued in two ways:

These changes make Bittensor more decentralized and market-driven by linking subnet value to token issuance and validator weight. TGEs attract capital and users, driving activity and increasing demand for TAO as the AI payment token.

How will dTAO contribute to the long-term value of $TAO?

Ultimately, the fate of dTAO hinges on the traction alpha tokens can generate. Every alpha token is paired in a liquidity pool, meaning $TAO is always required to gain exposure to any specific subnet.

Right now, we’ve already seen over 60 subnets launch on Bittensor, which is a significant and promising sign — it shows that builders are actively looking to develop within the network. However, launching a subnet is only the first step.

The real challenge lies in:

Finding product-market fit
Creating real economic value
Ensuring that value is shared with alpha token holders

For dTAO to have a meaningful impact on the price of $TAO, at least some subnets need to achieve significant success, reaching large valuations and proving their worth. Without this, the entire model could struggle to drive sustainable value back to $TAO.

Bittensor’s Key Subnets: Expanding the AI Economy

Subnet 19: Dominating the LLM Race

Nineteen AI is leading the way in decentralized inference, offering top open-source LLMs and image models. It recently set a world record for the fastest LLM inference, outperforming Perplexity and driving demand for TAO. With its high efficiency and decentralized model, Subnet 19 is emerging as a key player in AI inference.

Subnet 51: Celium — Decentralized GPU Powerhouse

Celium functions as a decentralized GPU rental marketplace, comparable to Io.net and Akash, providing high-performance GPUs like H100 and H200. While it reports strong revenue, the lack of official data raises transparency concerns — a live revenue dashboard, excluding emissions, could enhance credibility.

Currently ranked 6th among Bittensor subnets with 6.45% emissions, its long-term viability depends on demonstrating real demand, staying competitively priced, and differentiating itself from both centralized and decentralized GPU providers.

Subnet 4: Targon — Verifiable AI Responses

Targon, Bittensor’s third-largest subnet, specializes in verifying AI-generated content. Miners assess LLM responses by providing sources, with rewards based on accuracy and speed.

A key development is the addition of Deepseek, expanding access beyond its previous reliance on Chutes. This strengthens Targon’s role in ensuring factual accuracy in AI outputs.

Subnet 42 & 59: Masa — AI-Powered Social Influence

Masa (Subnet 42) is an AI network known for “fair AI” principles, primarily recognized for its data-scraping activities on X (formerly Twitter). Partnering with Yuma, Masa has introduced Subnet 59 — an arena where AI agents compete for mindshare on X in exchange for TAO tokens.

Key innovations include:

MASA Token — Bittensor’s first subnet token, rewarding contributors and facilitating data transactions.
Competitive Agent Model — Top-performing AI agents in Masa’s network are eligible for MASA token rewards, further driving engagement and innovation.

Subnet 46: NeuralAI — AI-Powered Virtual Worlds

NeuralAI (Subnet 46) specializes in 3D asset creation and AI-driven virtual worlds, targeting gaming and entertainment applications. Its flagship product, SentiOS, employs Autonomous Virtual Agents (AVA) to transform NFTs into dynamic, intelligent game characters.

NeuralAI Tokenomics:

NEURAL Token — Used for staking, governance, and platform transactions.
Market Performance — NEURAL has surged 84.4% in the past three months, riding the momentum of the agentic AI trend. However, as a low-cap token, it remains a high-risk investment.

Building Momentum: Key Partnerships and Growth

Digital Currency Group (DCG) has launched Yuma, a subsidiary dedicated to accelerating decentralized AI projects within the Bittensor ecosystem. As one of the network’s largest participants, Yuma is actively building and launching subnets, serving as the third-largest validator, and contributing to mining operations across multiple subnets.

With seven active subnets and nine additional projects in development, Yuma’s involvement, backed by DCG’s vast resources, strengthens Bittensor’s infrastructure and scalability. This institutional backing not only improves network security but also attracts more developers and investors looking for long-term growth opportunities.

Further reinforcing this institutional appeal, Crucible Labs, a key validator within Bittensor, brings a research-driven approach to the ecosystem, educating token holders and promoting decentralized TAO (dTAO) governance.

The growing visibility of Bittensor has also caught the attention of prominent tech figures, including YouTube co-founder Steve Chen, whose engagement with Bittensor-related content signals increasing recognition from influential players in the technology and investment sectors.

Additionally, Bittensor’s partnership with Virtuals Protocol has expanded its role in AI agent development. The collaboration has led to specialized subnets like the Audio-to-Animation (A2A) subnet, which advances AI-driven applications within the ecosystem.

With more institutional players joining the ecosystem, Bittensor is positioned for greater adoption, a stronger market presence, and sustained long-term demand for TAO.

EVM on Bittensor

Bittensor now supports Ethereum smart contracts on its subtensor blockchain, allowing developers to deploy and interact with them without modifications. It also enables access to Ethereum JSON-RPC methods, but all operations occur solely on subtensor, not Ethereum.

Implications:

Easier Developer Adoption — Ethereum devs can use familiar tools like Solidity, web3.js, and ethers.js.
AI + Smart Contracts integration.
Lower costs, faster transactions vs. Ethereum.
Potential cross-chain bridges in the future.
Specialized Subnets — Smart contract-focused subnets like Subnet 222 may evolve.

This move expands Bittensor’s role beyond AI, opening doors for DeFi, dApps, and tokenized AI models while offering an alternative to Ethereum’s smart contract ecosystem.

Where Bittensor Stands in the Market Race

Bittensor’s Potential Compared to Centralized AI Models

In the medium term, Bittensor is unlikely to challenge centralized AI leaders like OpenAI or DeepSeek. These firms have the advantage of massive funding, proprietary datasets, and seamless enterprise adoption, giving them a clear edge in AI sophistication and scalability. In contrast, Bittensor’s decentralized model, while innovative, faces hurdles in coordination, compute efficiency, and model optimization.

Longer term, its open and permissionless AI framework could position it as an alternative to centralized AI monopolies, especially as concerns over data control, censorship, and governance intensify. Success will depend on scaling contributions and attracting top-tier AI research. If it achieves this, Bittensor could establish itself as a leading Web3-native AI ecosystem. However, adoption remains its biggest challenge — without a critical mass of contributors and enterprise demand, it risks remaining a niche experiment rather than a mainstream competitor.

Bittensor and the Web3 AI Landscape

Bittensor vs Artificial Superintelligence Alliance ($FET)

Bittensor vs Near Protocol ($NEAR)

Bittensor vs Virtuals Protocol

Both Bittensor’s TAO and Virtual’s $VIRTUAL serve as core utility tokens within their ecosystems, facilitating transactions and speculation. However, their underlying mechanisms and value drivers differ. TAO is primarily focused on securing and governing decentralized AI inference, whereas $VIRTUAL plays a larger role in AI agent speculation and liquidity.

Virtuals assigns unique tokens to its AI agents, allowing direct speculation on their performance. Meanwhile, Bittensor subnets are beginning to introduce alpha tokens, though this remains an emerging trend rather than a fully established norm. In both ecosystems, subnet or AI agent tokens often trade against their respective core assets — VIRTUAL for Virtuals and TAO for Bittensor — but liquidity and adoption are still developing, particularly within Bittensor’s subnets.

Virtuals has seen a surge in speculation, fueled by the success of AI agents like AIXBT, Luna, and Vader, which has driven $VIRTUAL’s price growth and created a speculative flywheel. In contrast, Bittensor’s value accrual is more tied to computational demand rather than speculation on individual AI agents, making its economic model more infrastructure-driven than consumer-facing.

Another major difference is accessibility. Virtualz has lower entry barriers, allowing anyone to launch an AI agent, which fuels rapid speculation and token demand. Bittensor subnets require more technical expertise and economic investment, leading to slower adoption but potentially creating more sustainable long-term value.

Despite this, liquidity in Bittensor subnet pools remains limited, and investor interest is still growing as the ecosystem develops.

Both networks are shaping the onchain AI economy in distinct ways — Virtualz thrives on AI agent speculation, while Bittensor focuses on building decentralized AI infrastructure. Their long-term success will depend on adoption, liquidity, and the emergence of breakout projects that drive real demand.

Critical Risks and Challenges Ahead

The $TAO token powers the Bittensor network through transactions, governance, and contributor incentives. However, its biggest challenge comes from subnet emissions, which create continuous sell pressure. Even with halvings every four years, billions in sell-side pressure will persist until at least 2033. A rising $TAO price only compounds this issue, making long-term appreciation more difficult.

The upcoming dTAO upgrade aims to decentralize emissions, but it won’t fully eliminate this risk. So far, subnet tokens have struggled to gain traction, with low valuations and little impact on $TAO’s value. Unless subnets start producing real, revenue-generating applications, daily emissions will continue to be a drag.

Another key challenge is access to large-scale data banks and cutting-edge AI hardware. Unlike major tech companies, Bittensor lacks the resources to compete at the highest level, which could limit its long-term growth.

Finally, the broader crypto market remains a risk. Even with strong fundamentals, TAO’s price is vulnerable to sharp declines if investor sentiment shifts to risk-off mode.

Concluding Thoughts

Bittensor’s potential extends far beyond its current valuation — its subnet-driven economy could redefine how decentralized AI scales, making a $100B+ revaluation a tangible prospect.

The introduction of dTAO is a pivotal step toward optimizing capital flow, but its full impact will take time to materialize. Institutional players are already positioning ahead of this shift, with Coinbase and Grayscale signaling confidence in TAO’s long-term trajectory.

Yet, capital efficiency alone won’t resolve the persistent supply overhang. For real value creation, subnets must drive breakthrough applications that generate sustainable revenue.

While the AI crypto narrative is dominated by agent-based models, Bittensor is quietly building momentum. If execution outpaces emission concerns, TAO’s status as a category leader may be inevitable.

Disclaimer

This presentation has been prepared by Greythorn Asset Management Pty Ltd (ABN 96 621 995 659) (Greythorn) and GLC Capital. The information provided is for general purposes only and should not be considered investment or financial advice. It does not constitute an advertisement, solicitation, or offer to buy or sell any financial instruments or participate in any trading strategy.

Greythorn has not taken into account the investment objectives, financial situation, or specific needs of any recipient. Before making any investment decisions, recipients should evaluate their personal circumstances and seek professional advice from an accountant, lawyer, or financial adviser.

This presentation may include statements, opinions, projections, and forecasts (forward-looking statements) based on various assumptions. Greythorn is under no obligation to update this information, and these assumptions may or may not prove accurate. Neither Greythorn nor its officers, employees, agents, or advisers guarantee the accuracy or reliability of the information presented. To the extent permitted by law, Greythorn and its representatives accept no liability for any loss, claim, damages, costs, or expenses arising from the information contained herein.

This presentation is the property of Greythorn. By receiving it, the recipient agrees to keep its content confidential and not to copy, share, or disclose any information without written consent.