TAO Equals Scarcity

The Barry Shillbert stack

There is a particular kind of asset that lives in the corner of every cycle where the loud money refuses to look, and Bittensor’s TAO is sitting in that corner right now, mispriced not because the mechanics are hidden but because the people who would benefit most from understanding them are too busy chasing whatever launched on a meme launchpad this afternoon. The cleanest description of what TAO is borrows its cadence from Travis Kling’s well-known framing of Bitcoin as “a non-sovereign, hard-capped supply, global, immutable, decentralised, digital store of value,” and the parallel framing for TAO, with every qualifier earned rather than borrowed, is that TAO is a non-sovereign, hard-capped, programmatically scarce, halving-disciplined, decentralised, digital settlement asset for the open AI economy. My claim is that TAO can plausibly deliver a hundred-x from current levels across this cycle and the next, and the rest of this piece is the defence of every qualifier in that second line.

The Scarcity Pillar

The first thing to understand about TAO, and the reason the scarcity framing is not marketing language but a description of the actual protocol, is that Bittensor is hard-capped at 21 million tokens, which is the same number as Bitcoin and not a coincidence but a deliberate echo of the supply discipline that made Bitcoin work as a monetary asset. The network runs a four-year halving cycle that mirrors Bitcoin’s, with the first halving having already executed on December 14, 2025, cutting daily emissions from roughly 7,200 TAO per day to 3,600 TAO per day in a single block transition, and the subsequent halvings will compound the supply tightening at each four-year interval until emissions taper toward zero around 2032 to 2036. The mechanism is not governance-controlled and cannot be voted on by any subset of holders or developers, because the halving is hard-coded into the protocol and activates automatically when the circulating supply crosses each predetermined threshold, which is the same property that made Bitcoin’s monetary policy credible to institutional buyers who otherwise would not have touched a digital asset. The deflationary curve alone is not what makes TAO interesting, because plenty of capped-supply tokens exist and most of them go nowhere, but the curve becomes interesting in combination with the second mechanical detail that almost nobody outside the Bittensor community is talking about, which is that over seventy percent of the circulating supply is currently staked into validators and subnet positions and therefore removed from the immediately tradeable float. The combination of a hard cap, a freshly executed halving that cut new issuance in half, and a staking ratio that locks up the majority of what does exist produces a supply setup that is structurally tighter than almost any other major crypto asset. This in itself compounds quietly in the background regardless of what the price is doing on any given day.

The Institutional Pillar

The second pillar of the thesis is the one that most retail participants on this platform underweight badly, because they have been trained by (half) a decade of crypto culture to treat institutional involvement as either bearish or irrelevant, when in fact institutional accumulation in a hard-capped asset is the variable that moved Bitcoin from a thousand dollars to a hundred thousand dollars over a single decade and is the variable most likely to do the same kind of work for TAO. Digital Currency Group, which is Barry Silbert’s holding company and one of the most consequential capital allocators in the history of crypto, first invested in Bittensor in 2021 when the network was effectively unknown outside a small circle of decentralised-AI researchers, and that early position has since been operationalised into a full subsidiary called Yuma that launched in November 2024 with Silbert himself stepping in as CEO, which is the first time Silbert has taken a hands-on CEO role in four years and is the strongest possible signal that DCG is treating Bittensor as the headline allocation for its next investment phase rather than as a small line item in the portfolio. Yuma is not a passive holder, which matterssignificantly and is something the market has not yet fully internalised. Yuma operates as one of the largest validators on the network, runs subnets directly, accelerates and incubates new subnet projects, and as of October 2025 has launched a dedicated asset management arm anchored by a ten-million-dollar commitment from DCG that gives institutional and accredited investors a structured entry point into the network. On top of the Yuma layer, Grayscale and Bitwise have filed with the SEC for spot TAO ETFs, with decisions potentially arriving as early as August 2026, which would create the same kind of regulated capital on-ramp that turned Bitcoin from a retail asset into a sovereign-grade allocation across a two-year window after spot ETF approval. The base case here is not that institutional flows arrive and prices triple in a quarter, because that is the impatient retail framing of an institutional thesis, but that the institutional infrastructure is being built right now during the unfashionable phase of the cycle and will be sitting in place when the AI narrative re-accelerates and the broader market needs a clean institutional vehicle for exposure to decentralised AI.

The Product Pillar

The third pillar is the one that separates Bittensor from the long graveyard of AI tokens that promised infrastructure and delivered nothing, and it is the pillar that most thesis pieces about TAO either skip entirely or hand-wave through with vague references to subnets, which is a disservice to the reader and to the asset. Bittensor is a network of over one hundred and twenty independent subnets, each one a specialised marketplace where AI models, datasets, and compute resources compete for TAO emissions based on the quality of the work they produce, and the subnets cover everything from text-based AI tooling to confidential GPU compute to code generation tools that compete directly with Claude Code and Cursor, to computer vision pipelines that process raw video at scale, to protein-folding and financial-services applications that would be sitting inside walled gardens at OpenAI or Google in the centralised alternative. The metric that tells you whether this network is producing real value rather than merely producing emissions is the ratio of aggregate subnet token market cap to TAO’s own market cap, and according to Yuma’s State of Bittensor report published in March 2026 that ratio has now hit a record twenty-seven percent, which is a structural indicator that value is genuinely accruing to the subnet layer and flowing back to TAO holders through the staking and emission mechanics rather than evaporating into a marketing budget. This is the part where the cynical reader is allowed to remain cynical, because plenty of subnets will fail and most experimental projects on any open network do not survive contact with the market, but the relevant question is not whether every subnet succeeds but whether the platform as a whole produces enough successful subnets to justify the TAO token as the settlement layer for an entire emerging category of decentralised AI activity. The answer to that question, on the evidence available right now in early 2026, is that it is producing them faster than any other AI-native blockchain.

The Asymmetry

Pulling the three pillars together produces a thesis that is structurally cleaner than almost anything else in the market right now. The thesis rests on what TAO actually is rather than on what it could be made to resemble through marketing. What TAO actually is, on the evidence assembled across the previous three sections, is the dominant settlement asset of the decentralised-AI category at exactly the moment when that category is transitioning from a research curiosity to an investible thesis, with a hard supply cap that mirrors Bitcoin’s, a halving that has already executed in the way the mechanism was designed to execute, an institutional layer in DCG and Yuma that is positioned to absorb the next wave of capital, and a subnet network that is producing measurable value flows back to the base token at a rate that is accelerating rather than decelerating. A hundred-x from current levels on that combination of factors is a mathematical possibility that follows from the standard pattern of how scarce assets with maturing demand profiles get repriced when the consensus arrives, and the people who will collect that outcome are the ones who buy now while the room is still arguing about whether decentralised AI is real, hold through the inevitable drawdowns that come with any thesis trade that has not yet been consensus-validated, and unwind when the ETF flows and the institutional narrative have done the work the protocol designed itself to do.

The discipline required to hold this position through the noise is the same discipline that the previous article in this publication described at length, which is patience applied to a thesis nobody else has bothered to do the work to verify. The scarcity is real, the institutional accumulator is moving in plain sight, and the subnet network is producing real value flows back to the base token. Once those three facts are accepted as facts rather than debated as marketing, the position takes care of itself from the moment the trade is entered to the moment the consensus arrives to validate it.

May the FOMO be with you (at the price you deserve),

Baitman