We remain in the very early days of Bittensor subnet valuations.
Most subnets have a market cap in the $5M to $10M range. The biggest subnet, Chutes, currently sits at an $85M market cap, with an ATH of $140M.
No subnet has yet crossed the chasm. While there is rampant ‘early signal’, there is not yet a breakout success. We remain in the ‘2013’ stage of Bitcoin or ‘2016’ Ethereum.
New money is not yet entering Bittensor in force, though new accessibility pathways are rapidly developing (see Kraken, OKX, THORChain developments further down). The same relatively small pool of wallets is trading subnets against each other in a zero-sum game.
The entire subnet marketplace currently has a market cap of $700M, with an all-time high in March 2026 of $1.5B. It has all largely moved in lockstep with the broader Bitcoin and crypto market cap and sentiment, which has been bearish.
For us, the name of the game right now is to accumulate as much $TAO as possible by making the right subnet bets. Even though the broader market is slumping, significant ‘denominated in $TAO’ gains can still be made.
Our thesis remains that $TAO will be a breakout ‘third great coin’ alongside Bitcoin and Ethereum at some point. Early subnet bets are the path to accumulating $TAO during this market development phase, even though, as denominated in dollars, valuations may be flat or down, we view this as normal early-stage volatility that cannot be predicted or controlled.
In our view, the biggest mistake that can be made here is to be out of the subnet market when one or more subnets break out in a big way. We can’t control the timing of this, or the broader macro sentiment, but we can control our positioning and exposure and preparation for this moment.
In a certain sense, nothing has happened yet.
This is all still the calm before the storm.
Year One Of Subnets Is Over. This Is Year Two.
2025 was the first year in which Bittensor had subnets and subnet alpha tokens. There were a lot of questions around whether a successful subnet token marketplace would even arise and what form it would take.
The first year was indeed successful, but restrained. The good news is that we saw strength, but we did not see wild appreciation of subnet tokens followed by the inevitable crash and inability to ever recover (like what the agent tokens launched on Virtuals experienced). Instead, we saw slow, steady growth, but without any true breakouts.
Year one is over. Year two is about revenue and market success. Stillcore was at two Bittensor-heavy conferences in Europe recently: Digital Currency Group (DCG) Summit in Spain and Proof of Talk in Paris, and this was a common refrain we heard.
Now is where the rubber meets the road. Which subnets can sell products and produce real revenue? And which ones will cross the chasm first?
The State of Centralized AI
Centralized AI is increasingly under new pressure.
OpenAI and Anthropic are growing revenue rapidly but remain deeply unprofitable due to enormous compute and R&D costs. OpenAI generated ~$13B in 2025 revenue yet posted massive losses (operating ~$21B, net attributable up to ~$38B in leaked figures), burning billions quarterly while spending far more than it earns.
Anthropic has shown stronger momentum (~$30B annualized run-rate at points) and potentially earlier profitability, but both face heavy ongoing burn and capex pressures, with true profits likely years away.
Key concerns include disappointing ROI and eroding moats. Chamath Palihapitiya and others note that while AI boosts productivity (e.g., 10x for skilled users), broader economic delivery and enterprise returns lag hype, with companies pushing back on high token costs.
Open-source Chinese models like DeepSeek and especially GLM 5.2 (mid-2026 release) are closing the gap fast, competitive on coding/agentic tasks at ~1/6th the cost under permissive licenses, while accusations of distillation from Western models accelerate this.
On top of this, there are new concerns that AI may be censored. Anthropic released Mythos to the public as Fable. And then, three days later, the US Government banned it. The ban lasted eighteen days before being lifted. This used to be a purely theoretical fear. The off-switch is real, and it has now been used.
Intelligence is rapidly commoditizing and closing off.
In this macro environment, we believe the best place to be is in a Bitcoin-like open source AI provider marketplace.
How Bittensor Stacks Up in Mid-2026
Bittensor produces a superior AI product, and offers it to the market for the best price.
How? It does this by using Bitcoin’s ‘contest-o-nomics’.
The cost of actually running a subnet, colocation, machines, electricity, is pushed to the edges: to the miners. Miners spend their own money to compete. Only the best win. And only the best are awarded subnet tokens. Subnet tokens provide an ownership layer which lets miner-contributors participate in subnet market cap appreciation.
Note that the subnet operator never pays miners. The Bittensor chain does. The chain mints new $TAO and new subnet tokens with every block, and awards it to the best contributors.
Because of this, the subnet operator has no production cost.
This is exactly what Bitcoin does. Same idea. There is no entity paying for Bitcoin’s ongoing costs. The miners collectively pay for it.
However, the subnet operator needs to keep miners interested, or they will cease to mine the subnet. The ‘free labor’ and ‘free electricity’ and ‘free innovation’ will leave.
This means that a subnet operator needs to keep the subnet token value rising.
Usually, this means the subnet operator needs to sell the AI product produced by the subnet and produce revenue. Some portion of that revenue goes towards purchasing the subnet’s tokens off the open market, putting a persistent bid under the token.
Now. Why would a subnet operator do that?
Because it is much cheaper: a pattern we’ve seen is 1/10th the cost of hiring in-house. And because the product created is better: 256 of the best minds on earth competed to produce it. And because the product is produced much faster: a crucible of fierce flash-evolution is set up by the subnet. The company happily buys its own subnet tokens: 90% off (plus faster, plus better) is a great deal!
Also, in many cases, it is simply not even possible to create the same quality of product through hiring.
It wouldn’t be for Bitcoin. One organization could not pay for all the hash rate of the entire BTC network. It currently costs $8 to 15 billion/year to sustain 100% of current network hash rate. This is assuming a single organization could even get the electricity and data center space. The annual electricity consumption of the Bitcoin network is ~150 to 200+ TWh (terawatt-hours). This equates to ~200 billion kWh per year, which is comparable to a mid-sized country’s usage (e.g., Thailand or Argentina).
Instead, Bitcoin gets all of this for free.
Bittensor uses this exact same mining dynamic, with a similar economic result.
By pushing the risk AND the reward out to the edges, the cost of failure is absorbed by the market. But by providing the ownership layer of subnet tokens, the potentially very large rewards are also pushed to the market, as they were with Bitcoin.
What Happens If A Subnet Is Acquired?
We get this question a lot. The above discussion also answers the riddle of ‘what happens if a subnet is acquired?’ Won’t the acquiring company shut the subnet down and bring everything in house?
Or: Once a subnet becomes successful, won’t the company just kill it off?
A properly constructed subnet is producing the product at no cost to the operator. It is also producing a better product in many instances. And the subnet product is always evolving: it must always trend up in quality, and it must continuously improve with great speed: something Bittensor ‘contest-o-nomics’ provides.
Thus, if a company running a subnet is acquired, the acquiring entity is greatly incentivized to keep the subnet operational, as it is ‘the product department’ of the company they just acquired, producing high-quality product quickly and cheaply.
If Google could somehow acquire Bitcoin, they would keep Bitcoin mining operating rather than lose $8B to $15B per year running the machines and hash rate themselves.
How Subnets Win: Small, Specialized Models
Instead of very large generalized models, like Mythos or Claude Opus or Grok, Bittensor subnets produce smaller but much more specialized models, custom-made for a specific vertical.
A specialized subnet, trained and optimized around one problem, can outperform a general model that was not built for that task. Some examples (by their own published benchmarks):
- Score (SN44): World’s best AI vision
- Trishool (SN23): World’s best AI safety
- Bitsec (SN60): Beat Mythos at finding cybersecurity holes
- Vidaio (SN85): World’s best video compression
You can think of ‘Big AI’ (Anthropic, OpenAI) models as like mainframes, and Bittensor subnets as like PCs.
Local, sovereign, focused models are where AI is headed. Think Mythos on your mobile, fully owned and operated by the user, with full AI sovereignty and full data privacy.
Think swarms of small expert models, networked, micro-paying one another in crypto for services and expertise.
Think decentralized crucible which trains, serves and refines these models faster and cheaper than any centralized entity can keep pace with. That’s Bittensor.
The centralized players today: Anthropic, OpenAI, xAI, are wildly unprofitable. They lose billions annually, with their costs 2x to 3x greater than revenue.
At the same time, their frontier models are distilled and replicated nearly instantly into free or fractional cost (1/6th) models. DeepSeek. And more recently, GLM 5.2.
The moat, increasingly, isn’t the model.
On top of these dynamics, once you apply Bitcoin’s ‘contest-o-nomics’ to every AI layer, pre-training, post-training, inference, applications, you get (on average) another 10x boost in cost efficiency to what the open-source models are already providing.
As AI commoditizes, as the cost of intelligence dives towards zero, we believe Bittensor economics and decentralization and open source are going to be where the biggest value capture occurs.
None of this means the frontier labs are going away, or that their work doesn’t matter. It does. They are spending billions to find the edge of what’s possible, and someone has to. Our bet is simply about where the value of that work accrues once it exists, and we think more of it than people expect will flow to the open, decentralized layer that distributes and specializes it.
The Successful Subnet Thesis
As a result of Bittensor’s ‘Bitcoin-like contest-o-nomics’, a pattern has begun to emerge in subnet operation that suggests a template for success.
A subnet replaces a company’s product department. You do not hire or pay for the ongoing production of your product. The subnet takes care of all of that.
Your job as a company is to first define your product as a series of Bittensor incentive mechanisms. This guides the activity of your subnet. And then, your job is to sell the product.
You produce revenue.
Some portion of that revenue then goes into purchasing your own subnet tokens off the open market. You might burn them (destroy them forever) or you might hold them in reserve. Either way, this reduces circulating supply, all else being equal, that’s upward pressure on the token.
Ideally, you sell enough of your product to fully offset all subnet tokens being awarded (emitted) to miners by the chain. This is called being Subnet-Profitable. Two subnets report having achieved this to date, Bitcast (Subnet 93) and Synth (Subnet 50).
We also have significant revenue traction from Score (Subnet 44), which produces a subnet-powered AI vision product that sells to enterprise clients through a distribution partnership with PricewaterhouseCoopers. Founder Max Sebti has publicly projected $40M in annualized revenue, with a crossing into subnet-profitability later this year.
Because Bittensor uses Bitcoin-like ‘contest-o-nomics’, it converts energy into intelligence more efficiently than anything else on earth.
What Bitcoin did for stranded energy, Bittensor does for stranded talent.
The chain is the VC. Subnets are the startups. Miners are the workers.
Miners can sell winnings immediately or they can hang on to them long-term if they believe the subnet flywheel will kick in and grow the token’s value. Bittensor thus offers the ownership layer to workers. And it pushes the option for speculation to them also. Like early Bitcoin miners, some will sell immediately to cover costs, and some will choose to hold, speculating that the subnet will be worth much more in the future. Because miners, not subnet owners, cover electricity and other costs to compete, this lowers the cost of product production across the entire Bittensor network.
So to put it all together:
- Small, specialized models win: they’re better
- Bittensor produces these models cheaper (1/10th the cost of any other method)
- Bittensor produces better models faster
- Bittensor recruits stranded talent and gets the best brains and best results
- Companies then sell the subnet product and use the revenue to buy and burn subnet tokens, making them scarcer and more valuable
- Repeat. This produces a flywheel that pushes subnet token valuation higher
- All subnet flywheels in concert make TAO more valuable
Recent Subnet Success Highlights
Now you’ve read the theory. But what about specific, practical results? Where is this ‘early signal’? Below, we’ve collected some highlights of early success and achievement from Bittensor subnets. Figures are as reported by the teams.
| Subnet | Category | Recent achievement |
|---|---|---|
| Score (SN44) | AI Vision | Partnered with PwC France to deploy its Business Operations World Model for enterprise physical AI, targeting multi-year contracts and projecting $40M+ in annualized revenue through decentralized computer vision services. |
| Bitsec (SN60) | Cybersecurity | Beat Anthropic’s Mythos at discovering security issues. Bitsec claims to have jailbroken Fable 5 (turning it into full Mythos) and pointed it at a paying client’s codebase (with permission), then ran Bitsec against the same code. Mythos found around 60 vulnerabilities, five of them high or medium severity. Bitsec found over 160 in total, including five criticals and ten highs that Mythos missed entirely, and it caught everything Mythos caught. |
| Vidaio (SN85) | Video | Achieved Earth’s best video compression through its decentralized AI-powered models, delivering up to 80% smaller file sizes with superior perceptual quality that outperformed traditional and commercial solutions. |
| IOTA (SN9) | Training | Macrocosmos completed an Orion-100B parameter decentralized training run, achieving up to 65% of datacenter efficiency for a 100-billion-parameter model distributed across globally available GPUs. |
| ENGY (SN53) | Inference | Announced the availability of GLM 5.2 at the lowest inference cost tier in the world, serving it on consumer GPUs powered by the cheapest power on earth. Leadership hails from name-brand frontier AI labs. |
| Chutes (SN64) | Inference | Processes trillions of tokens monthly, has ranked first among open-source inference providers on OpenRouter, and broke $1M in monthly revenue. Launched the Parallax training framework and announced an attempt at a 1-trillion-parameter MoE model across globally distributed GPUs. |
| Actual (SN95) | Inference | Released a test inference engine that runs on heterogeneous compute clusters, including low-end home MacBooks and PCs. Amazingly, it even works on Windows 95. Full product release expected this summer. |
| Bitcast (SN93) | Marketing | A decentralized creator ad network managed by AI, became the first profitable Bittensor subnet, fully offsetting miner emissions through brand sponsorship revenue recycled into alpha token buys and burns. |
| Zeus (SN18) | Weather | Consistently outperformed state-of-the-art centralized models like ECMWF IFS HRES and AIFS, with reported improvements such as ~23.55% better 2m temperature accuracy and ~75% faster delivery (~1.5 hours vs. ~6 hours). |
| Synth (SN50) | Finance | A price-prediction subnet, became the second profitable Bittensor subnet. Recently announced a $100,000 SN50 alpha token buy and burn. |
| Trishool (SN23) | AI Safety | Released HaloGuard 1.0. The 4B version claims first place across seven established prompt-safety benchmarks, and the 0.8B version outperforms models several times its size. |
| 404Gen (SN17) | 3D / Gaming | AI text-to-3D, now has plugins for Unity, Blender and Unreal Engine, enabling paid in-engine generation of scenes and assets for immersive worlds. |
| Targon (SN4) | Compute | A decentralized confidential compute marketplace, co-authored an AI security whitepaper with Intel and has $5.5M in annualized revenue to date. |
| Minos (SN107) | Genomics | A decentralized genomic variant-calling and benchmarking platform, completed over 37,000 variant-calling evaluations in ~11 days (about 1/5th the time of a conventional lab setup). |
| Vanta (SN8) | Finance | A decentralized prop firm, launched a major marketing campaign with WallStreetBets, resulting in 2,600 traders and $500M in volume. |
| Teutonic (SN3) | Training | Formerly Templar, delivered Covenant-72B decentralized training success, with NVIDIA’s Jensen Huang publicly referencing the milestone. After a founder exit in April, it rebranded under Bittensor’s Const, kicked off an 80B king-of-the-hill run, and the community is using the new Conviction tools to take the subnet back. |
Bittensor macro
Comparison of Subnet Market Caps vs. Non-Bittensor Equivalents
Now let’s have a look at some of the actual market caps of Bittensor subnets versus centralized and non-Bittensor competitors. Let’s quantify what the delta is, and what the opportunity is.
Inference & GPU compute: Chutes, Targon, Lium. Chutes (SN64, ~$84.6M) is one of the top open-source inference providers on OpenRouter and processes trillions of tokens a month. Targon (SN4, ~$58.5M) runs confidential compute with recently reported $5.5M in annualized revenue and a security whitepaper co-authored with Intel. Lium (SN51, ~$46.7M) is a similar GPU marketplace. Their decentralized peers: Render at ~$834M and Akash at ~$185M. Even Venice, Erik Voorhees’ private inference platform running a similar buy-and-burn playbook, trades at ~$545M. And the centralized peers? RunPod hit $1 billion in June. Baseten raised at a $5 billion valuation. Fireworks was priced at $4 billion in October and is reportedly in talks at $15 billion. And on July 1, Together AI closed an $800 million round at an $8.3 billion valuation, to serve open source models, which is literally Chutes’ job. One funding round, bigger than the entire subnet asset class. Chutes trades at about 1% of Together AI.
Training: IOTA (SN9, ~$32.5M). This subnet shipped the largest permissionless training run on Earth (Orion-100B). Now look at what the market pays the labs training open models in a centralized way. Zhipu, maker of the GLM 5.2 model that’s all over this report, listed in January at about $7 billion and is worth north of $100 billion today. DeepSeek just raised at a reported $45 billion. Moonshot, maker of Kimi, is valued at more than $20 billion. Qwen comes from Alibaba, a company worth hundreds of billions. Even Nous Research, a startup chasing a similar decentralized training thesis, raised from Paradigm at $1 billion. Yes, the labs’ models are bigger. For now. But it’s the same category of work, and the market prices one side in the tens of billions and the other side at $30M.
Storage: Hippius (SN75, ~$16.3M). Versus Filecoin (FIL, ~$580M) and Arweave (AR, ~$150M): the two decentralized storage incumbents are worth about 45x Hippius combined. And that’s with Filecoin down 67% over the past year. Hippius is priced as a rounding error on an already beaten-down sector.
Code security: Bitsec (SN60, ~$6.2M). Versus Snyk ($7.4B peak valuation): Snyk peaked at $7.4 billion doing this centrally. Bitsec, which just beat Anthropic’s Fable 5 head-to-head on a live codebase, trades at less than 0.1% of that.
New Bittensor Support Across Exchanges And Bridges
Now you’ve read the subnet achievements. You’ve seen what the delta looks like in subnet market caps versus other AI companies.
Clearly, the broader market is not getting this yet.
But many crypto ecosystem providers are.
Exchanges. Bridges. These parties are now actively preparing for a day when new money suddenly understands Bittensor and demands to get into subnets. These providers are actively building ways to accommodate this:
- Kraken, a tier-one exchange, added seven Bittensor subnet alpha tokens to its listings roadmap, including Chutes (SN64), Targon (SN4), Score (SN44), Lium (SN51), Ridges AI (SN62), Hippius (SN75), and Vanta (SN8), marking the first major CEX expansion beyond TAO. Stillcore predicted this in our January ‘State of TAO’ report.
- OKX, also a tier-one exchange, listed TAO/USDT (Bittensor) for spot trading, with deposits and trading live June 30, 2026. It also launched OKX AI, an on-chain marketplace for autonomous agents, with the Opentensor Foundation among its launch partners.
- Wormhole activated a canonical TAO bridge to Solana on May 5, 2026, enabling native trading on Jupiter and Meteora and compatibility with Phantom/Solflare wallets for broader DeFi access. We expect that canonical subnet tokens will also be made available on Solana soon.
- THORChain announced impending mainnet integration of Bittensor (TAO), enabling native, non-custodial cross-chain swaps with assets like Bitcoin and expanding decentralized liquidity for the AI network.
The High Speed of Change In Bittensor
There have been several fast-implemented changes to the core operations of the Bittensor chain and incentive alignment mechanisms since our last report. These include programmable subnet tempo, Root Reborn, shorting, and the rescinding of TAO Flow.
Each has thrown prior assumptions in the air, and required a re-think of investment strategy across the board. This reinforces the idea that Bittensor is still a highly dynamic environment that requires constant attention.
Our view is that these changes have been largely good and have improved the Bittensor ecosystem’s chance of success, disruptive though they may be. Bittensor is managed with a predisposition to action: identify the largest problem, fix it quickly, then move to the next problem and repeat. There are no committees or lengthy discussions, just an announcement, feedback (which is heeded in some cases), and then fast implementation.
We applaud this.
Root Reborn
The chain used to sell subnet tokens for $TAO with every new block (this was done to produce $TAO yield for root stakers).
Now, it doesn’t do that. Instead, that flow now stays invested in subnets, with the validators dictating which ones. Chain-based subnet token sell pressure has been removed.
Instead of kneecapping subnets, the Bittensor chain now invests in subnets. This is of course bullish for subnet valuations moving forward.
TAO Flow Rescinded
TAO Flow was meant to allocate new chain $TAO emissions to subnets that had most recently had a flow of new buys, which theoretically showed where market sentiment thought chain-based support should flow to.
However, this was gameable. So now a moving average of subnet alpha price is used.
Conviction
Conviction ties subnet ownership to locked, at-risk stake: conviction = stake × time. Lock more, for longer, and you can take a subnet from an owner who walked away. Proposed days after the Covenant founder’s exit, it’s being tested first on the old Covenant subnets (3, 39, 81) and isn’t fully live network-wide, but it answers the exact failure that exit exposed. A recent upgrade made conviction build roughly 3x faster, cutting a takeover like SN3’s from ~100 days to ~33.
Subnet-Programmable Tempo
Every subnet used a fixed global tempo, a default block interval (typically ~360 blocks, or roughly every 72 minutes) between Yuma Consensus epochs. This is when the subnet calculates performance scores and distributes alpha/TAO rewards to miners, validators, and the owner. Now, subnet owners can set a custom tempo.
Shorting
Several designs to make on-chain shorting of subnets possible are currently being discussed. One of them looks likely to be approved.
Conclusion
It’s mid-2026. Bitcoin is 50% off all-time highs. Crypto sentiment is bearish, despite a new Fed chairman and what appears to be impending Clarity Act passage.
This has all hindered the growth of anything in crypto, period. This is true despite the fact that Bittensor operates in the hottest growth sector on earth: artificial intelligence.
We think that will change. Soon. Nobody knows exactly when, but our guess is 2026 or 2027 at the latest.
Until then, Stillcore will continue studying subnets, meeting founders and positioning ourselves to capitalize on breakout subnet success.
Disclosures
About Stillcore Capital
Stillcore Capital serves as investment manager to private investment funds (the “Funds”) structured to qualify for exclusion from the Investment Company Act of 1940 under both Sections 3(c)(1) and 3(c)(7). The Funds are not registered under the Investment Company Act or the Securities Act of 1933. Stillcore Capital Management, LLC is not registered as an investment adviser under the Investment Advisers Act of 1940 or any state securities laws, and operates in reliance on exemptions from registration available to advisers to qualifying private funds.
Purpose of This Report
This report represents our views on the Bittensor ecosystem and decentralized AI markets. It is published for informational and educational purposes. Nothing in this report should be interpreted as an offer to sell, or solicitation to purchase, interests in any Stillcore Capital fund or any other investment product. The Funds are offered exclusively through private placement to qualified investors who satisfy applicable accreditation and suitability standards, subject to minimum investment thresholds and redemption limitations set forth in the governing documents.
Not an Offer
This report is market research and ecosystem analysis. It is not an offer to sell, or a solicitation of an offer to buy, interests in any Stillcore Capital fund. Fund interests are offered only to verified accredited investors under Rule 506(c) of Regulation D, and only through confidential offering documents.
No Fund Performance Disclosed
This report contains no information about the performance, returns, or specific holdings of any Stillcore Capital fund. Investment views, position sizing concepts, or strategies referenced herein reflect general market perspectives and may not correspond to actual fund portfolios. The Funds’ actual investments and performance may vary significantly from any themes discussed in this report.
Not Investment Advice
Nothing in this report constitutes investment advice, legal advice, tax advice, or a recommendation to purchase or sell any asset. Reading this report does not establish an advisory relationship or any professional engagement between you and Stillcore Capital. We do not act as your fiduciary or agent. This content is not tailored to any individual’s financial situation, objectives, or risk parameters. You should evaluate any investment decision in light of your own circumstances and consult appropriate professional advisors.
Digital Asset Risks
TAO, Subnet tokens, and other digital assets discussed in this report are speculative instruments with substantial volatility and the potential for complete loss of value. Risks include but are not limited to: severe price swings, thin liquidity, evolving and uncertain regulatory treatment, smart contract vulnerabilities, protocol failures, cybersecurity threats, custody challenges, and absence of traditional investor protections. Regulators may determine that certain digital assets constitute securities, which could impose registration requirements and trading restrictions that materially impair value or transferability.
Forward-Looking Statements
This report includes projections, forecasts, and statements about anticipated future developments. These forward-looking statements rest on assumptions that may prove incorrect, and actual outcomes could diverge materially from what is described. Contributing factors include shifts in regulation, technology setbacks, competitive disruption, execution challenges, and general market conditions. All information reflects our understanding as of the publication date and may become stale. We have no obligation to revise or update this report.
Valuation Methodology Caveats
Any price targets or valuation frameworks presented rely on subjective inputs and modeling choices. Alternative assumptions would yield different conclusions. Revenue growth estimates, comparable company multiples, network activity projections, and token economic assumptions are inherently speculative. Small changes in key variables can produce dramatically different outputs. These models are illustrative, not predictive.
Conflicts of Interest
Stillcore Capital, its principals, employees, and related parties maintain positions in TAO and various Subnet tokens, including tokens discussed in this report. We stand to benefit financially from appreciation in these assets. This economic interest creates inherent bias in our perspective. We may execute trades, adjust stakes, or otherwise transact in any asset discussed herein at any time without notice. We may also advise, invest in, or maintain relationships with Subnet teams, validators, and other Bittensor ecosystem participants. We owe no fiduciary duty to readers. Our interests may not align with yours, and we may be incentivized to present information favorably to our positions.
Sources and Data Reliability
Information presented derives from public blockchain data, team disclosures, industry sources, and our own research. We believe these sources to be generally reliable but have not independently audited or verified all data points. Metrics such as revenue, user counts, and performance figures may be self-reported, unaudited, or calculated using non-standard methodologies. Readers bear sole responsibility for independent verification and due diligence.
Disclaimer of Warranties and Liability
THIS REPORT IS PROVIDED ON AN “AS IS” BASIS WITH NO WARRANTIES OF ANY KIND. TO THE FULLEST EXTENT PERMITTED BY APPLICABLE LAW, STILLCORE CAPITAL DISCLAIMS LIABILITY FOR ANY LOSSES OR DAMAGES, DIRECT, INDIRECT, CONSEQUENTIAL, OR OTHERWISE, ARISING FROM USE OF OR RELIANCE ON THIS REPORT.
Copyright and Use Restrictions
This report is the intellectual property of Stillcore Capital. Reproduction, redistribution, or creation of derivative works requires prior written consent, except for limited personal, non-commercial use permitted by law.
Third-Party References
References to external websites, platforms, or content do not imply endorsement. Third-party information is accessed at your own risk.
Geographic Limitations
This report is not directed at persons in jurisdictions where its distribution would violate local law or require regulatory authorization. Recipients are responsible for understanding and complying with applicable legal requirements in their jurisdiction.
Hypothetical Projections
Valuations, price scenarios, and return projections are hypothetical exercises with well-known limitations. They do not reflect real trading, do not account for transaction costs or market impact, and benefit from hindsight. Actual investment results would differ.
Key Assumptions
Analysis in this report assumes: continued Subnet revenue growth consistent with early-stage network trajectories; valuation comparables drawn from high-growth technology sectors; staking and liquidity dynamics that persist in line with current patterns; stable or favorable regulatory developments; and proper functioning of Subnet token economic mechanisms.
Analytical Limitations
Readers should recognize that: assumptions embedded in this analysis may fail to materialize; projections exclude fees, expenses, and tax consequences; selection of highlighted examples involves judgment calls that may reflect favorable bias; comprehensive risk assessment is beyond the scope of this report; model outputs are highly sensitive to input assumptions; and past performance of any referenced asset or comparable does not guarantee future results.
Selective Coverage
Subnets highlighted in this report were chosen to illustrate aspects of the Bittensor ecosystem. This selection is inherently subjective. We have not profiled every Subnet, including those that have underperformed, failed, or been removed from the network. Success of featured Subnets provides no indication of likely outcomes for other Subnets or for the ecosystem as a whole.
