Person. Teng Yan Yau · @tengyanAI · @0xPrismatic
The independent analyst who has written on Bittensor most consistently from outside it: a 2024 essay asking whether it is flawed, a widely shared 2025 guide to dynamic TAO, and a 2025 subnet thesis, all with positions disclosed and criticism kept in.
dTAO is a major step in the right direction … the start of a Bittensor renaissance, one that plays out over years. … forcing everyone to finally pay attention—not just to TAO, but to what each subnet is actually building.
21 Feb 2025 · in contextEvery quote is verbatim and dated. The reading above each era is written from the cited quotes only. Numbers link to the quote they cite.
In 2024 he explains Bittensor as horse racing for models and then argues the economics fall apart when unlike subnets compete for validator votes, the code is not open, and a billion dollars a year buys nothing better than open-source models [teng-yan-001, teng-yan-002, teng-yan-004, teng-yan-005, teng-yan-003]. He calls it an intriguing experiment that risks the graveyard of hype 006. By January 2025 the tone softens: the OG in the room has quietly made leaps and dTAO could be the turning point [teng-yan-007, teng-yan-008, teng-yan-010].
horse racinga decentralised incentive network for AI models … it's horse racing for AI models. The best horse (model) gets the greatest rewards.
His explainer of Bittensor for newcomers at the top of the essay Bittensor, flawed?, republished on Chain of Thought on this date after first appearing on his personal Substack in April 2024.
the economic model starts to fall apart when competition extends horizontally. … having every AAPL shareholder vote on the next product
His central criticism in the same essay: TAO holders and root validators cannot fairly compare value across dissimilar subnets such as text-to-speech and X-ray screening.
not seen Bittensor subnets achieving outcomes or capabilities that exceed those of existing open-source models
From the same essay, set against Meta's open-source release of Llama 3 reaching state of the art.
7,200 TAO paid to subnet owners + miners + validators. This equates to ~$3.2 million per day or $1.2 billion annually.
From the same essay, questioning whether the network generates enough value to justify its daily emission cost.
it is NOT open source. … What are the miners doing? What algorithms are they running? How are the final results generated?
From the same essay, on transparency: decentralised is not the same as open source, and outsiders cannot see what miners run.
one of the most intriguing experiments in the emerging Crypto x AI sector … must address its current challenges head-on to stand any chance of achieving real-world success—otherwise it will be lost in the graveyard of speculative hype.
His conclusion to the same essay, suggesting Bittensor focus on niche areas such as crypto-specific tasks rather than compete with generalist AI.
Bittensor (TAO) has been around for years. It's the OG in the room. But its token price has languished, hovering at the same level as a year ago despite all the frenzy around AI.
From his annual outlook essay, in the section on what to watch in Crypto x AI for 2025.
Beneath the surface, this 'digital hive mind' has quietly made leaps: more subnets with lower registration fees, subnets that are outperforming their Web2 counterparts in real metrics like inference speed, and EVM compatibility that introduces DeFi-like features onto Bittensor's network.
From the same outlook, listing what changed in 2024 while price stood still; the hive mind phrase is in his scare quotes.
So, why hasn't TAO soared? A steep inflation schedule and the shift in attention toward agent-oriented platforms have held it back.
From the same outlook, giving his explanation for the flat price.
However, dTAO (estimated Q1 2025) could be the big turning point. … dTAO ties block rewards directly to innovation and actual measurable performance. The better the subnet, the more valuable its token—and thus, the more emissions it gets.
From the same outlook, previewing dynamic TAO a month before it shipped.
After the February 2025 upgrade he becomes its explainer: power shifting from validators to the open market, Uniswap-style pools, buying is staking, a renaissance that plays out over years, with his own positions disclosed [teng-yan-011, teng-yan-012, teng-yan-014, teng-yan-015]. By May 2025 he calls Bittensor the most compelling place to watch Crypto x AI, the project most aligned with fair launch, and a cult with real conviction, while keeping the warnings about inflation, liquidity and leverage [teng-yan-016, teng-yan-017, teng-yan-018, teng-yan-021, teng-yan-022]. The 404-Gen profile shows the subnet-as-swarm frame 023.
marketShift From critic of the root-validator economics to holder and guide; the criticisms migrate from the design itself to the risks of the new token market.
a fundamental shift: power shifting from top validators to the open market. … emissions are no longer assigned through validator weightings. Instead, TAO emissions now flow through a market-driven system using subnet tokens.
From his guide to dynamic TAO published a week after the upgrade, explaining the change to readers without formulas.
the same pricing mechanism used in Uniswap. … When people say they're 'buying' or 'selling' subnet tokens, they're actually just staking or unstaking TAO into subnet pools.
From the same guide, correcting a common newcomer misconception about alpha tokens and constant-product pools.
Switching to an Alpha-based staking model introduces a new set of risks. … a bad actor could swoop in, buy up cheap Alpha, and use it to manipulate emissions.
From the same guide, on the risks that replace the old validator conflicts of interest.
dTAO is a major step in the right direction … the start of a Bittensor renaissance, one that plays out over years. … forcing everyone to finally pay attention—not just to TAO, but to what each subnet is actually building.
His verdict at the end of the guide, with a disclosure that he holds TAO and several subnet alpha tokens.
This changes everything. And you need to pay attention. Let me help you understand what's happening with this historic shift in the most important Crypto AI protocol today.
The lead text attached to his X post sharing the dTAO guide under the title Dynamic TAO: Your No-Nonsense Guide; read through the fxtwitter mirror, where it appears as the article card.
Bittensor was built when no one cared. It launched in 2021, before Crypto x AI was a thing.
From his subnet thesis essay Bittensor Subnets: Opportunity Aplenty, arguing why Bittensor is the place to watch Crypto x AI.
Of all the crypto-native AI projects we've seen, Bittensor is the most structurally aligned with Web3's first principles: fair launch, open participation, and live incentives for coordination.
From the same essay, stating what sets Bittensor apart from other crypto AI projects.
In short, it's a cult. There's weird stuff, sure. But also real conviction. There's actual activity on the subnets.
From the same essay, describing the community to outside readers.
A small validator set couldn't meaningfully assess dozens of fast-evolving subnets. … Some validators were running their own subnets. … dTAO hands the decision over to the market. Subnets now have to earn emissions by proving demand.
From the same essay, explaining why the root network system failed and what replaced it.
Some call TAO the Bitcoin of AI. Others say it's the internet of AI.
From the same essay, reporting the category frames in circulation without adopting either.
Alpha tokens give you leveraged exposure to subnet success. But that leverage cuts both ways. … Inflation is high. Liquidity is thin. Volatility is baked in.
From the same essay, warning investors after noting Chutes, Targon and TAO Hash at billion-dollar fully diluted valuations.
the most compelling place to watch Crypto x AI unfold in the near term … Time spent understanding subnets > trying to time the buy.
The thesis line and closing advice of the same essay, aimed at investors.
It is not one model but a swarm, competing in real time. Contributors plug in different approaches, validators score results, and tokens flow to the winners. Weak models vanish, strong ones rise.
From his profile of 404-Gen (subnet 17), co-written with Avdhesh Charjan, explaining how a subnet produces output.
A brand strategy skeleton turned on the speaker: not what Bittensor's brand is, but what this voice's Bittensor is. Slots with no evidence are left out rather than filled.
What they say it is for, the world it builds, what it stands for.
He holds it to transparency and open source, and credits it with fair launch, open participation and live incentives.
it is NOT open source. … What are the miners doing? What algorithms are they running? How are the final results generated?
Of all the crypto-native AI projects we've seen, Bittensor is the most structurally aligned with Web3's first principles: fair launch, open participation, and live incentives for coordination.
What is broken that Bittensor answers, in their telling.
Horizontal competition across unlike subnets breaks the economics, emissions cost over a billion dollars a year, the code is not open, inflation is steep, and alpha tokens invite manipulation and leverage in both directions.
the economic model starts to fall apart when competition extends horizontally. … having every AAPL shareholder vote on the next product
7,200 TAO paid to subnet owners + miners + validators. This equates to ~$3.2 million per day or $1.2 billion annually.
it is NOT open source. … What are the miners doing? What algorithms are they running? How are the final results generated?
So, why hasn't TAO soared? A steep inflation schedule and the shift in attention toward agent-oriented platforms have held it back.
Switching to an Alpha-based staking model introduces a new set of risks. … a bad actor could swoop in, buy up cheap Alpha, and use it to manipulate emissions.
A small validator set couldn't meaningfully assess dozens of fast-evolving subnets. … Some validators were running their own subnets. … dTAO hands the decision over to the market. Subnets now have to earn emissions by proving demand.
Alpha tokens give you leveraged exposure to subnet success. But that leverage cuts both ways. … Inflation is high. Liquidity is thin. Volatility is baked in.
Bittensor (TAO) has been around for years. It's the OG in the room. But its token price has languished, hovering at the same level as a year ago despite all the frenzy around AI.
If it does not address its challenges it joins the graveyard of speculative hype.
one of the most intriguing experiments in the emerging Crypto x AI sector … must address its current challenges head-on to stand any chance of achieving real-world success—otherwise it will be lost in the graveyard of speculative hype.
The tailwinds they cite.
dTAO as the turning point that ties rewards to measurable performance and starts a multi-year renaissance, on top of quiet 2024 gains in subnets and EVM.
However, dTAO (estimated Q1 2025) could be the big turning point. … dTAO ties block rewards directly to innovation and actual measurable performance. The better the subnet, the more valuable its token—and thus, the more emissions it gets.
dTAO is a major step in the right direction … the start of a Bittensor renaissance, one that plays out over years. … forcing everyone to finally pay attention—not just to TAO, but to what each subnet is actually building.
Beneath the surface, this 'digital hive mind' has quietly made leaps: more subnets with lower registration fees, subnets that are outperforming their Web2 counterparts in real metrics like inference speed, and EVM compatibility that introduces DeFi-like features onto Bittensor's network.
Who they talk to and what they tell them.
Newcomers who need the mechanism explained without formulas, and investors told that understanding subnets beats timing the buy and that leverage cuts both ways.
a decentralised incentive network for AI models … it's horse racing for AI models. The best horse (model) gets the greatest rewards.
a fundamental shift: power shifting from top validators to the open market. … emissions are no longer assigned through validator weightings. Instead, TAO emissions now flow through a market-driven system using subnet tokens.
the same pricing mechanism used in Uniswap. … When people say they're 'buying' or 'selling' subnet tokens, they're actually just staking or unstaking TAO into subnet pools.
Alpha tokens give you leveraged exposure to subnet success. But that leverage cuts both ways. … Inflation is high. Liquidity is thin. Volatility is baked in.
the most compelling place to watch Crypto x AI unfold in the near term … Time spent understanding subnets > trying to time the buy.
What they set it against.
In 2024 he set Bittensor against open-source models like Llama 3 and against its own root-validator voting, which he likened to Apple shareholders choosing products.
the economic model starts to fall apart when competition extends horizontally. … having every AAPL shareholder vote on the next product
not seen Bittensor subnets achieving outcomes or capabilities that exceed those of existing open-source models
Category frame, differentiator, reason to believe.
A decentralised incentive network for AI models, explained as horse racing; later the most important Crypto AI protocol, while he reports rather than adopts the Bitcoin-of-AI and internet-of-AI frames.
a decentralised incentive network for AI models … it's horse racing for AI models. The best horse (model) gets the greatest rewards.
This changes everything. And you need to pay attention. Let me help you understand what's happening with this historic shift in the most important Crypto AI protocol today.
Some call TAO the Bitcoin of AI. Others say it's the internet of AI.
It is not one model but a swarm, competing in real time. Contributors plug in different approaches, validators score results, and tokens flow to the winners. Weak models vanish, strong ones rise.
The OG that launched before Crypto x AI existed, the project most aligned with fair launch and open participation, and after dTAO the one where the market rather than validators allocates emissions.
Bittensor (TAO) has been around for years. It's the OG in the room. But its token price has languished, hovering at the same level as a year ago despite all the frenzy around AI.
Bittensor was built when no one cared. It launched in 2021, before Crypto x AI was a thing.
Of all the crypto-native AI projects we've seen, Bittensor is the most structurally aligned with Web3's first principles: fair launch, open participation, and live incentives for coordination.
a fundamental shift: power shifting from top validators to the open market. … emissions are no longer assigned through validator weightings. Instead, TAO emissions now flow through a market-driven system using subnet tokens.
A small validator set couldn't meaningfully assess dozens of fast-evolving subnets. … Some validators were running their own subnets. … dTAO hands the decision over to the market. Subnets now have to earn emissions by proving demand.
It was built when no one cared, which he reads as authenticity.
Bittensor was built when no one cared. It launched in 2021, before Crypto x AI was a thing.
The benefit they claim, the feeling they sell, what joining says about you.
Rewards flow to measurable performance; a subnet is a swarm where weak models vanish and strong ones rise.
However, dTAO (estimated Q1 2025) could be the big turning point. … dTAO ties block rewards directly to innovation and actual measurable performance. The better the subnet, the more valuable its token—and thus, the more emissions it gets.
It is not one model but a swarm, competing in real time. Contributors plug in different approaches, validators score results, and tokens flow to the winners. Weak models vanish, strong ones rise.
Intrigue at an experiment, tempered by warning.
one of the most intriguing experiments in the emerging Crypto x AI sector … must address its current challenges head-on to stand any chance of achieving real-world success—otherwise it will be lost in the graveyard of speculative hype.
Belonging to something he calls a cult with real conviction.
In short, it's a cult. There's weird stuff, sure. But also real conviction. There's actual activity on the subnets.
The line they repeat and the proof they point to.
This changes everything, pay attention: dTAO forces attention onto what each subnet builds, and Bittensor is the most compelling place to watch Crypto x AI unfold.
dTAO is a major step in the right direction … the start of a Bittensor renaissance, one that plays out over years. … forcing everyone to finally pay attention—not just to TAO, but to what each subnet is actually building.
This changes everything. And you need to pay attention. Let me help you understand what's happening with this historic shift in the most important Crypto AI protocol today.
the most compelling place to watch Crypto x AI unfold in the near term … Time spent understanding subnets > trying to time the buy.
In 2024 the proof cut against Bittensor: no subnet exceeded open-source models against a billion-dollar emission bill; by 2025 he cites subnets beating Web2 on inference speed.
not seen Bittensor subnets achieving outcomes or capabilities that exceed those of existing open-source models
7,200 TAO paid to subnet owners + miners + validators. This equates to ~$3.2 million per day or $1.2 billion annually.
Beneath the surface, this 'digital hive mind' has quietly made leaps: more subnets with lower registration fees, subnets that are outperforming their Web2 counterparts in real metrics like inference speed, and EVM compatibility that introduces DeFi-like features onto Bittensor's network.
Register more than claim.
Plain, punchy and irreverent: the OG in the room, it's a cult, this changes everything.
Bittensor (TAO) has been around for years. It's the OG in the room. But its token price has languished, hovering at the same level as a year ago despite all the frenzy around AI.
In short, it's a cult. There's weird stuff, sure. But also real conviction. There's actual activity on the subnets.
This changes everything. And you need to pay attention. Let me help you understand what's happening with this historic shift in the most important Crypto AI protocol today.
How they describe their own relation to it.
An outside researcher who moved from asking whether it is flawed to holding TAO and alpha tokens, disclosed at the end of the dTAO guide.
one of the most intriguing experiments in the emerging Crypto x AI sector … must address its current challenges head-on to stand any chance of achieving real-world success—otherwise it will be lost in the graveyard of speculative hype.
dTAO is a major step in the right direction … the start of a Bittensor renaissance, one that plays out over years. … forcing everyone to finally pay attention—not just to TAO, but to what each subnet is actually building.
What they call things, and refuse to call them.
He corrects the words: buying a subnet token is staking TAO into a pool, and Bitcoin of AI and internet of AI are what others call it.
the same pricing mechanism used in Uniswap. … When people say they're 'buying' or 'selling' subnet tokens, they're actually just staking or unstaking TAO into subnet pools.
Some call TAO the Bitcoin of AI. Others say it's the internet of AI.
Who they agree with, borrow from, or argue against, with the quote that shows it.
Public appearances on the record, whether or not a quote was taken.
Something wrong or missing? Quotes are corrected as they arrive; a misattribution is removed first and argued after. How to file one · GitHub issue