Key Insights
- Bittensor is aiming to create a market for artificial intelligence (AI) that allows anyone to train their AI model and anyone to use it, given a small fee paid in TAO
- TAO has seen many peaks and valleys, starting at $0.12, having an all-time high of $759.01, and now sitting around $290, seemingly on a continuous upward trend, especially recently
- The fully diluted value of TAO sits at $6 billion, with its current market capitalization being half of that at $3.09 billion, as only 10 of the total 21 million supply of TAO is in circulation
1. Thesis & Trade Recommendation
| Direction | Entry Zone | Stop Loss | Price Target |
| Short | $270 – 290 | $320 (hard) | $180-200 |
TAO is a narrative-driven token without a direct fee or revenue capture method. Despite this, it has a $3.1B market capitalization, it faces a global macro risk-off environment driven by the Iran-Israel War and oil shock, a BTC technical weakness with no confirmed reversal above $83K, a relative underperformance when the crypto market is in a bull market, and a structural ceiling with 48% still locked. Thus, we recommend a short position.
2. Protocol / Project Overview
Bittensor is a Layer 1 DeAI blockchain built on the Substrate framework. It is an open-source, decentralized blockchain network designed to create a peer-to-peer marketplace for AI. Rather than hosting generic smart contracts, it runs 128 subnets that act as competitive arenas. Here, AI miners produce digital intelligence, such as inference, data, and compute, that respond to a prompt (paid for using TAO by an external user). Validators then rank their quality using a Yuma Consensus mechanism, and votes are weighted according to stake amount. TAO rewards the best-performing miners, and the worst are pruned using the Darwinian replacement algorithm.
Total Addressable Market
Artificial intelligence is the defining technological transformation of this decade. 10 years ago, it may have been considered an industry vertical. Nowadays, every company is utilizing and promoting AI; AI has now become horizontal across industries. However, the infrastructure powering these forward leaps remains concentrated in a handful of hyperscale corporations. Amazon Web Services, Microsoft Azure, and Google Cloud collectively control ~65% of global cloud infrastructure. Decentralized AI (DeAI) seeks to redistribute this infrastructure through blockchain-based incentive systems, aiming to create an open, permissionless alternative to centralized AI services.
Given Bittensor’s nature, the total addressable market involves all areas of AI. Currently, global AI infrastructure is about $101B (2026E), growing at a CAGR of 27.7%. DeAI’s infrastructure is ~$14B, AI Inference Service is ~$8B, and decentralized compute is ~$2B. At a $3.1B market cap, TAO already commands roughly 22% of the DeAI market despite generating zero revenue.
Current Market Analysis
Before examining the opportunity, the current macro environment demands acknowledgement. BTC has dropped approximately 47% since its October 2025 all-time high above $125,000 . Bitcoin fell below $65,000 in February as tariff announcements triggered broad risk-off selling across crypto and equities. Altcoins and AI-focused tokens have been hit harder given their higher beta to overall market sentiment. TAO is currently trading just under $300, down significantly from its previous high of $370 after a shout-out from Jensen Huang.
This divergence highlights a key issue. While the narrative around decentralized AI remains compelling, Bittensor’s current price and positioning are still heavily narrative-driven with limited real usage, revenue generation, or defensible demand to support its current market capitalization. Price action has largely tracked shifts in sentiment rather than fundamental adoptions, which leaves the token exposed in risk-off environments. As speculative capital exists in the market, assets with limited real-world traction are likely to see disproportionate downside.
Business Model
Bittensor has no trading fees, protocol-owned yield, or subscription revenues. Instead, value is generated via
- Incentive emissions: miners and validators earn TAO for contributing quality AI computation. 41% goes to miners, and 59% goes to validators and stakers.
- Registration fees: subnet creations and miner/validator registration require TAO payments that are cycled (not burned) into unissued supply to delay future halving.
- Stacking lock-ups: 70% circulating supply is staked to suppress liquidity and create a structural scarcity.
TAO is a narrative and staking demand token. It does not accrue cash flows, but rather it resembles commodities like BTC or gold.
Traction & On-Chain Metrics
3. Devil’s Advocate: Why Long TAO?
Poaching Market Share and Long-Term Goal
Amazon set a precedent that gaining market share is more important than short-term profits. Given the proven strategy, the market has a similar outlook to Bittensor and believes the fundamental purpose is an achievable, highly valuable long-term goal. Fundamentally, Bittensor’s revenue is almost negligible, and its main source of funding is the appreciation of TAO. Yet the potential of a competing network of open-source AIs accessible to everyone continues to drive TAO’s price.
Market Sentiment Caused by NVIDIA and AI
NVIDIA’s GTC 2026 keynote reinforced AI’s role in the future, lifting its AI hardware revenue outlook to at least $1 trillion through 2027. This event caused many AI-linked cryptos, including TAO, to move upwards, with TAO up ~14.1% on March 16th.
Recent AI developments also propelled TAO’s prices. Covenant-72B’s progress in Bittensor Subnet 3 illustrated clearly that AI can now be trained on a decentralized platform. It surpassed LLaMA-2-70B (Meta’s open-source AI) and LLM360 K2 (LLM360’s open-source AI).
Especially as investments and developments of AI continue to expand exponentially, TAO’s relevance and differentiated take on AI will likely rally TAO even further.
Long Term Holdings & Other Investments
~73% of available TAO supply is staked, illustrating long-term conviction. Large holders include name brands such as DCG and Polybrand. Institutional staking is reinforced by the dTAO’s enhanced decentralization via market-based reward redistribution and inherent mimicry of BTC via scarcity, deflationary mechanics, and halving schedule.
Summary
We see these catalysts as more short-term and narrative-driven, thus not being representative of TAO’s long- term performance. As such, we believe more factors will negate and be detrimental to TAO’s performance instead
4. Investment Thesis: Why Short TAO?
TAO’s Decentralized Computation
Decentralized computation becomes attractive as centralized computing becomes more expensive. But this does not benefit the DeAI sector. Instead, it benefits compute-specific protocols first.
- According to Messari, the Akash network posted a 428% YoY growth in usage with utilization above 80% heading into 2026. It also has a fee-capture with its Burn Mechanism Enhancement, where every dollar spent on Akash burns $0.85 of AKT.
- Decentralized platforms like Fluence are gaining traction with their cost being 85% lower than that of centralized cloud.
TAO subnets can theoretically serve this purpose, but TAO is not at the forefront as a replacement when AWS gets expensive. Instead, Akash, Fluence, and projects like io.net are.
According to Messari’s State of Akash Report, the network generated approximately 529% increase in new lease revenue alongside a 54% increase in GPU usage and a 55% increase in deployed GPU capacity. There is a sustained free generation with quarterly revenue reaching 860,000 and continuing to grow with deployment activity.
Practicality and Pricing of Decentralized AI against Centralized AI
Moreover, considering the target audience of Bittensor, the practicality of this project is called into question. Given that the general population only needs products and services that are ‘good’ and adequate rather than the ‘best’, most individuals will gravitate towards free options such as Anthropic’s Claude, OpenAI’s ChatGPT, Google’s Gemini, and other models that are easy to access without payment.
Under this assumption, the main users of Bittensor would be individuals who require highly trained, realistic AIs and have the means to pay for them. Not only does this significantly reduce the customer base, but it would also require decentralized AI models to be more effective than private, capital-backed, and well-established centralized models. Additionally, TAO’s value would likely also be compressed as each request would likely require one or more tokens as compensation, creating selling pressure to use the trained models. In order to even compete with existing centralized models, which are free on basic plans or ~$0.01-0.06 per 1,000 API tokens, TAO would see massive price compressions, or each transaction would cost minimal amounts of TAO, both undermining the utility and price of the token.
The previous assumption that rising AI demand will drive users toward decentralized alternatives is increasingly flawed. Instead, advancements in centralized AI infrastructure continue to reinforce their competitive dominance. Recent optimizations, such as the Goggle’s KV cache update reduces inference costs and latency by reusing prior computation rather than recomputing tokens from scratch up to 6 times (from 18KV to 3KV). This not only lowers marginal cost per query but also enables more efficient deployment across constrained hardware environments, including edge devices and SIM-based architectures. As centralized models become cheaper, faster, and more portable, they further entrench their advantage across both cloud and distributed systems. In contrast, Bittensor’s decentralized model lacks comparable optimization mechanisms and economies of scale. This disadvantage would make it structurally less competitive on both pricing and performance.
Iran War and Oil Shock
The US-Israel strikes on Iran on Feb 28, 2026, triggered a 8.5% surge in a single day. Qatar LNG halted its operations, and the fear-and-greed index has been at 26 for weeks. Not only will this cause surges in energy costs, negatively affecting TAO given its relationship with AI and crypto, but, as noted by Morgan Stanley, an active geopolitical shock will also push capital away from risky assets, such as crypto, tech, and AI, into more risk-averse assets, namely gold, silver, and energy. This macro concern exists even if Bittensor was not a decentralized AI, but rather the larger AI training scene will be largely impacted by increased energy prices.
Bitcoin (BTC)
BTC dominance is at 57%, which, based on historical data, suggests capital is not rotating into altcoins. In this environment, even a good second-order argument would be ignored. Spot BTC ETFs have outflow of $1.7B in the past week, outperforming almost all prior inflows.
Furthermore, BTC’s depression to $67k from its peak at $126k has negative implications for TAO: if BTC’s narrative as a crypto ever deflates or fundamentally erodes, TAO’s mimicry of BTC will also shatter, paving the way for failure.
AI Narrative Reprices & Declining Credit Confidence
According to Bain&Company, AI infrastructure expansion would require ~$2 trillion in annual revenue by 2030. Even under generous assumptions, revenues fall $800B short. Thus, only profitable protocols will survive as people are made aware of whitepaper scams. Morgan Stanley notes that markets want evidence that the AI CapEx will pay off. Reinforced by the increased vigilance of consumers, there is a shift towards capital-driven instead of narrative-driven protocols, and TAO does not fall into a capital-driven protocol.
To reinforce this, the tech industry (especially AI) is highly leveraged, notably via private credit, due to the amount of venture capital and private investments. If private credit does falter, tech loses funding and confidence, which causes a cycle of inadequate capital and reduced confidence.
Jamie Dimon made comments that liken private credit to cockroaches. More recently, BlackRock limited withdrawal amounts from its $26 billion HPS Corporate Lending Fund to just 5%, and other private equity and credit firms, such as Blackstone and Blue Owl, saw a massive increase in redemptions from clients. Blue Owl investors asked to withdraw 40.8% of shares in its technology fund, while Blackstone saw a record 7.9% redemption request of its credit fund. Such activity is a potential symptom of the instability and faltering confidence that the market has in private credit.
Declining User Sentiment
Artemis data shows that where the core team is growing (+380% active devs, +29.5% commits), the sub-ecosystem is declining (−80% external devs, −98.9% commits ). Even as internal upgrades accelerate, Bittensor is collapsing, and users are leaving, paralleling the ’empty app store’ problem.
Additionally, a dilution-adjusted supply analysis reveals a downward trend in TAO’s price. Only 51% of the total token supply (10.77 million of 21 million) is in circulation, while 3,600 TAO is minted daily. The resulting $1,008,000 causes new selling pressure, which needs to be sustained by buy demand. Artemis on-chain data, however, identifies no meaningful external demand driver, causing supply to overwhelm demand and driving the price downwards.
We believe that this is a major point of evidence for our loss of confidence in TAO’s success. Given a declining user base, the company will not have a strong market share to implement an effective revenue capture strategy, nor will it see strong confidence from the market.
5. Risk and Mitigants
Conflicting Technical Analysis
TAO’s rally toward $200 on March 7 was rejected immediately. Spot CVD declined, funding rates turned negative, OBV broke below local support, and Open Interest rose 6% during the failed rally. TAO has been range-bound $165–$200 since mid-February, with momentum pointing toward the lower bound. Key resistance is from $185–$195, with key support at $165.
While this technical analysis is sound, TAO skyrocketed to $299 (16th March 2026). As such, it can be clearly seen that external events and successes (mentioned above) can and will influence the movement of the coin regardless of technical analysis. However, after the rise, TAO has been steadily decreasing to its current price of $268 (-10% in 2 days). As such, it can be stated that while news can influence TAO upwards, the price will fundamentally fall back down.
Iran War Ceasefire
The probability of a ceasefire is low, and the market does consider that. Polymarket shows confidence is below 50% that a US-Iran ceasefire will be reached before May 31st (48% by March 31st and 71% by December 31st). Given how long the markets have, the results of a ceasefire impacting energy and AI will likely be priced into TAO, minimising the positive effect.
This has been reinforced by Trump’s constant reassurance in ending the war yet constant delays with his most recent statement being criticised as a speech that begins a war rather than ending one. Additionally, he announced there would be devastating damage in 48 hours (namely, April 6th) and mentioned earlier this week that the next two weeks would be devastating to Iran. Alongside the fact that Iran declined the peace proposal, it is unlikely that a ceasefire will happen soon.
Clarity Act & Grayscale Project
The Clarity Act is proposed U.S. legislation that aims to clearly define when crypto is a security (regulated by SEC) or a commodity (regulated by CFTC). The status quo’s ambiguity has prevented a lot of capital from being invested within the industry.
Given TAO’s tokenomic similarities with Bitcoin, it is likely that TAO would be classified as a commodity if the Clarity Act is passed. However, this classification is not a clear catalyst for upside. Market expectations around regulatory clarity are already partially priced in and limit the incremental impact of such a decision. While a commodity designation would subject TAO to relatively lighter regulation compared to securities, the speculative nature of this outcome reduces its ability to drive sustained new capital inflows.
Furthermore, the passage of the Clarity Act itself remains uncertain. Even in a favorable scenario, the likely timeframe will be between April 13 and October 5. This extended window provides the market with adequate time to realise losses previously accrued, and also intersects with our 12-week holding period. Additionally, Grayscale’s potential TAO ETF remains contingent on regulatory clarity and introduces further delays on top of the 15-week prediction beyond the Act’s passage. As a result, both regulatory and classification and ETF speculation are unlikely to serve as near-term catalysts, and their effects may already be reflected in current pricing.
6. Comparable Projects
Artificial Superintelligence Alliance (ASI) (Ticker: FET)
Merger Between Fetch.ai, SingularityNET, and CUDOS
Fetch.ai is a blockchain-based platform with the goal of combining AI with autonomous agents to automate real-world actions and complex processes beyond communication. Singularity is an open blockchain-based marketplace, founded by Dr. Ben Goertzel, where developers publish, share, and monetize finished AI tools, and users purchase access using AGIX tokens. CUDOS is a decentralized network that provides infrastructure for Web3 and enterprises that require GPU virtual machines.
After the merger, ASI has used Fetch.ai’s ticker FET. ASI is a simpler and less technically ambitious version of Bittensor that has a fee-capture mechanism (service payments using AGIX). Thus, TAO is narrative-driven and has the largest DeAI market cap, yet it has zero revenue, contrary to its competitors.
Kaggle
Kaggle is a machine learning (ML) competition platform founded in 2010 and acquired by Google in 2017. Fundamentally, Kaggle provides datasets and competitions where models are trained, ranked, and rewarded based on performance. This platform allows Google to outsource and identify talent while also helping train its own AI models.
This system appears to be very similar to Bittensor’s incentive system. However, the main difference between the two is that Kaggle is episodic, meaning the competitions have a deadline and end, whereas Bittensor is continuous and on-chain. Essentially, a centralised versus a decentralized AI training system.
Hugging Face
Hugging Face is an open-source platform hosting over 2 million AI models and serving 18 million monthly users, also known as the “GitHub of AI”. It is backed by Google, Amazon, and Nvidia at a $4.5B valuation with $130M in revenue.
Ironically, CEO Clem Delangue warns the industry faces an “LLM bubble”. He predicts that smaller, specialized models that are cheaper, faster, and enterprise-deployable would replace general-purpose models. This aligns with Bittensor’s decentralised model but challenges the LLM-centric thesis.
Render
Render is a decentralized GPU compute network that connects idle GPUs from individuals and data centers to users who need to compute for tasks such as 3D rendering, AI training, and inference. These jobs are split across nodes and paid for through the RNDR token.
Compared to Bittensor, Render operates at the infrastructure layer, solving a more immediate and tangible bottleneck with more use cases, such as gaming, milm production, enterprise AI, whereas Bittensor’s use cases are still quite small and abstract, largely centered around experimental AI model collaboration that are already ubiquitous (voice to text, image generation, etc) yet depending on sustained network effects
Akash Network
Akash is a decentralized cloud compute marketplace to provide a permissionless alternative to AWS, GCP, and Azure. Considered the “eBay of compute”, providers offer idle GPU and CPU capacity, and tenants, notably AI developers, deploy containerized workloads via blockchain-based contracts.
Unlike TAO, Akash captures revenue as 20% of compute fees go to a community pool and has a deflationary mechanism, causing AKT to appreciate.
Summary
TAO does not have direct competitors, partly due to the fact that there is no large AI competition market, and SingularlityNET’s merger into ASI may be a cause for tapering demand. Its subnet competition model for AI quality is more sophisticated than compute rental, but in a risk-off macro environment with energy-driven AI cost concerns, the decentralized compute rotation trade flows to Akash and Render, not TAO.
Not only does TAO require additional enterprise adoption steps that do not exist yet, but it also has a narrower use case compared to other open-source AI, autonomous agents, and compute rentals. Reinforced by Artemis’ sub-ecosystem developer data (−80% to −99%), this sentiment will not change in the near future.
7. Valuation
TAO has no earnings. Hence, DCFs are not applicable. Since TAO is a narrative-driven token, we can utilize multiples in a comparable analysis with the most similar tokens.
Note: There are no credible sources for FET’s revenue that are publicly available. For TAO, Artemis considers tokens recycled into the protocol via network fees.
8. Final Thoughts
Given our bearish outlook, we expect to enter short positions when TAO sits in the $270 – 290 range on a centralised exchange with a hold period of 12 weeks. We will also implement a stop loss of ~$40 (i.e. TAO reaches $320) in the event that TAO’s long-term vision materializes earlier than expected, alongside various momentum and external catalysts that could drive prices upwards.
We believe that, fundamentally, Bittensor has a very valuable project that can be developed into a truly open-source marketplace funded by TAO. However, given the short-term decrease in interest for the project, the absence of revenue generation, and the likely continuing negative macro outlooks for energy and AI costs, investors are likely to pull away from AI and anything tangentially related. In tandem with TAO’s relatively small market capitalization, increased volatility would also disincentivise investors from this protocol. In fact, since starting this paper, TAO has dropped from ~$330 to dipping below $300 despite the recent positive industry news of NVIDIA, Covenant-72B, and Grayscale’s SEC Milstones which caused prices to shoot up. Even then, prices were still far from TAO’s peak price of ~$770, illustrating hesitancy and caution within the market.
Leave a Reply