Bittensor Grows Up: Quantum Keys, Revenue, a Listed Proxy

Three bids for legitimacy in one week

Bittensor spent most of 2026 answering the same two questions. Is the revenue real, and is the network actually decentralized? In the first half of November it tried to change the subject three times at once.

In the space of a few days the Opentensor Foundation shipped post-quantum signing keys into the chain runtime, an independent research group reported that paying customers were now sending the network roughly $70 million a year (double the late-summer figure), and the largest publicly traded holder of TAO prepared to report quarterly earnings as a Nasdaq-listed company. Put together, they look like a network growing up: custody-grade security, real revenue, and a stock-market proxy for people who want the exposure without holding the token.

The price did not cooperate. TAO changed hands near $217.57 on 11 November, for a market value around $2.55 billion and a rank just inside the top 40, down about 71% from its March 2024 high of $757.60 and well below the roughly $329 it touched in late September when an AI-token rally was still running, according to CoinGecko. The tape is telling one story; the fundamentals are trying to tell another.

The argument of this piece is simple. Each of the three bids is real, and each carries an asterisk. The quantum keys are optional and not yet switched on. The revenue still covers under a quarter of what the network pays out in tokens. The listed proxy is small and loss-making. And all three ultimately route back through the same concentrated control the project has promised to give up by the end of 2027.

Bittensor in one paragraph

For readers meeting the network for the first time: Bittensor is a proof-of-stake chain, built on Substrate, whose block rewards pay for machine intelligence instead of raw hashing. The work happens inside subnets, independent markets where miners compete to deliver a service (model inference, fine-tuning, data collection, even distributed pre-training) and validators score how well they do it. A consensus mechanism called Yuma turns those scores into pay. Since a 2025 upgrade known as dTAO, every subnet also has its own alpha token and automated market-maker pool, and the alpha price sets the subnet’s share of emissions. The base asset, TAO, is capped at 21 million like Bitcoin and halves on a supply schedule, with the first halving in December 2025. The network runs on the order of 130 subnets today, with a hard cap rising toward 256, and the core team has sketched a full-stack intelligence network in which subnets buy data, compute and inference from one another. The Opentensor Foundation and cofounder Jacob Steeves built it; both founders stepped back from executive roles in February 2026 but still shape the roadmap.

Post-quantum keys: what actually shipped

The most technical of the three changes is also the most novel. On 5 November, runtime 488 went live on Bittensor’s Finney mainnet at block 9,442,681, carrying three post-quantum signature schemes and an optional key-rotation feature; the Opentensor Foundation announced it the next day, according to tao.media and Crypto Briefing.

The three schemes are ML-DSA-65, Falcon-512 and Falcon-1024. ML-DSA-65 is the lattice-based scheme the US National Institute of Standards and Technology standardized in 2024 as FIPS 204 (its research name is Dilithium); Falcon is the compact lattice scheme NIST picked as the basis for a forthcoming companion standard. In Bittensor’s own benchmarks, ML-DSA-65 signs fastest, in about half a millisecond, while Falcon-1024 targets the highest security category and Falcon-512 produces the smallest keys and signatures.

What matters for users is the coverage. The new keys reach everything that holds value on the network: both coldkeys (the cold-storage key that custodies and transfers TAO) and hotkeys (the operational key validators and miners sign with), across transfers, staking, subnet registration and validator weight-setting, plus multisig and proxy accounts. There is also an optional automatic key-rotation mode. After a transaction is recorded, the wallet can roll to a fresh signing key while the account’s address, balance and stake stay the same; the old key can no longer authorize anything. Turning rotation on costs a one-time registration lock of 0.2 TAO. The goal is to shrink the window in which any single exposed signature could be abused.

For a store-of-value chain, quantum risk is mostly about protecting balances. Bittensor has a second exposure. Validators sign the weight vectors that decide how emissions are shared, and hotkeys authorize staking and registration, so a compromised operational key could redirect real cash flow, not just drain a wallet. That is why extending post-quantum coverage to hotkeys and validator weight-setting, rather than only to cold storage, is the part of this upgrade that fits Bittensor specifically, even if the wider quantum threat is still years away.

The catch: optional, pending, and not quantum-proof yet

Three caveats keep this from being the Bitcoin-killer some TAO holders billed it as.

First, it is not switched on. The schemes sit in the runtime, but post-quantum signing stays disabled until the chain’s root authority flips a privileged setting (in the code, a sudo call named for enabling hashed accounts). Until then, a post-quantum address can receive TAO but cannot spend it, per tao.media. The upgrade is a loaded option, not a live defense.

Second, it is additive and opt-in. The classical Sr25519 and Ed25519 schemes Bittensor has always used are unchanged and need no migration. Anyone who does nothing keeps elliptic-curve keys, which are exactly what a quantum computer would eventually target.

Third, the threat is still theoretical. The danger is Shor’s algorithm: a large, fault-tolerant quantum computer could derive a private key from the public key that an elliptic-curve signature exposes. No such machine exists today. The reason to move early is captured by the phrase ‘harvest now, decrypt later’, the worry that an adversary records public keys now to crack them whenever the hardware arrives. Bittensor is buying insurance, not responding to a break-in.

There is a sharper irony buried in the mechanics. The quantum hardening is gated behind a single switch controlled by root, the same concentrated authority critics say makes the network centralized in the first place. The most future-proof security upgrade in Bittensor’s history will go live when one privileged key decides it should.

Revenue finally grew, and a DCG founder noticed

The second bid is about money, and it answers 2026’s loudest criticism most directly. For most of the year, skeptics argued that Bittensor’s subnets farmed token emissions and sold almost nothing to real customers. The emissions were generous; the invoices were thin.

That gap is narrowing quickly. SubConnect’s Bittensor Revenue Index, which counts only subnets with verifiable external income, put revenue-generating subnets at roughly $70 million in annualized revenue in early November, up from about $50 million on 1 October and about $35 million when the index launched on 1 September, as reported by PANews and detailed by Crypto Briefing. That is close to a doubling in two months, roughly 40% month over month.

The cheerleading came from a notable corner. Barry Silbert, the founder of Digital Currency Group, told his followers on X on 4 November that the Bittensor ecosystem was ‘on fire’ and that revenue was growing 40% a month, sharing the SubConnect numbers, per cryptonews.net. Silbert’s enthusiasm is not disinterested, as the next section explains, but the underlying trend is corroborated by the index’s method, which tracks paying customers rather than token flows.

Compute and infrastructure subnets do most of the earning, roughly 78% to 82% of the verified total. The leaders by external revenue are serverless-compute and inference shops: Lium (SN51), Targon (SN4) and Chutes (SN64), the serverless-inference subnet run by Rayon Labs. Clients surfaced in the index work have included PwC France, Dropbox and an unnamed NYSE-listed real-estate investment trust. SubConnect projects the line clears $100 million by year-end.

What these subnets actually sell is compute and inference: serverless model hosting, confidential GPU capacity, fine-tuning and data pipelines. The catch is that the product is commoditizing on both sides at once. Centralized inference prices have fallen hard through 2026 as the big model makers cut rates, so the discount a decentralized network offers is a moving target, and the gross margin on reselling rented GPUs is thin before token incentives. Growing revenue 40% a month is impressive; doing it while the price of the underlying service keeps dropping is harder than the headline suggests.

The gap that did not close

Here is the asterisk. The network’s emissions, the TAO it mints and pays out every day, run north of $300 million a year at current prices. Against that, $70 million of external revenue covers under a quarter of the payout, up from under a tenth in late summer but still a heavy subsidy. Bittensor is paying far more to produce intelligence than customers are paying to consume it.

Independent trackers are harsher than the index. Analysts at ownyourmind have argued that only a sliver of self-reported subnet revenue is verifiable on-chain or through third parties such as the routing marketplace OpenRouter, and that the biggest inference subnet, Chutes, has at times run an emissions-to-revenue subsidy on the order of 20 to 40 times, with a break-even cost per million tokens well above what centralized providers charge. The revenue-versus-emissions question is not unique to Bittensor; HOGE Wire has traced the same tension on Solana, where a high-revenue chain still struggles to turn fees into holder value, in our look at real yield and the payout problem.

The distinction that matters is between claimed throughput and verified revenue. A subnet can self-report well over a hundred billion tokens a day; what the index tries to count is dollars from people who are not being paid in alpha to show up. The honest read in November 2026 is that real demand exists, is growing fast, and is still a fraction of the token budget keeping the lights on. Whether that curve crosses, revenue catching emissions, is the entire investment case.

A Wall Street proxy for TAO

The third bid is the most literal form of legitimacy: a stock. TAO Synergies (Nasdaq: TAOX) is a pure-play Bittensor treasury company, the former biotechnology firm Synaptogenix, which rebranded in 2025 and rebuilt itself around a single idea, accumulating and staking TAO, according to a company release.

By early November it held 54,058 TAO, which it says makes it the largest publicly traded holder of the token. Unlike a US spot ETF, it stakes those tokens to the Bittensor root network and earns rewards; root staking has run around 10% a year, and the company has reported trailing revenue near $770,000 from the program, per stockanalysis.com. That is a small number, and the business is deeply unprofitable, carrying a large net loss and a share price near $2.58, far below its highs. It is due to report quarterly results later in November, its first set of numbers as a scaled treasury.

TAOX is not alone. A second listed vehicle, xTAO, pursues a similar staked-treasury strategy. The pattern echoes the corporate-treasury trade that defined 2025 for Bitcoin, now ported to a decentralized-AI token, with one twist: a TAO treasury can earn native staking yield that a passive holder of the token (or of a non-staking fund) cannot. The appeal is the yield, which puts these vehicles in the same debate as liquid-staking tokens, where the hard part is not generating a rate but sustaining demand for it, a problem HOGE Wire examined in Lido vs Rocket Pool vs Frax.

The treasury trade carries its own reflexivity. A vehicle like TAOX raises cash, buys TAO, and ideally sees its shares trade at a premium to the tokens it holds, which lets it raise more and buy again; the loop runs in reverse just as easily when the premium flips to a discount and the token falls. The companies that loaded up on Bitcoin learned both halves of that lesson in 2025. Adding staking yield cushions the downside a little, but it also layers equity-market risk, dilution and management costs on top of a token that is already volatile. The proxy is convenient; it is not a free lunch.

The DCG thread

Silbert’s ‘on fire’ post deserves a second look, because Digital Currency Group now touches Bittensor from at least three directions. DCG’s asset-management arm, Grayscale, is the sponsor trying to turn its Bittensor Trust into a US spot ETF. DCG’s validator subsidiary, Yuma, stakes across the network and has become one of the loudest institutional voices in Bittensor governance. And DCG was among the backers of an $11 million private placement into TAO Synergies, the treasury company, alongside its digital-strategy adviser James Altucher, as The Block reported.

One conglomerate is therefore long Bittensor through a fund, a validator and a listed proxy at the same time. That is a vote of confidence and a concentration risk in a single breath, and it means the network’s most visible revenue bull is, to a degree, talking his book. The bets are not even monolithic inside DCG: Yuma, the validator (not to be confused with Yuma Consensus, the scoring algorithm), publicly warned that the Root Reborn overhaul could ‘fundamentally alter the role of validators’, a rare case of one DCG entity flagging risk in a change another corner of the network cheered, per crypto.news.

The ETF that is still waiting, and the shutdown that is lifting

For a year the obvious institutional on-ramp, a US spot TAO ETF, has been stuck. Grayscale filed to convert its Bittensor Trust (ticker GTAO, destined for NYSE Arca, with a fee near 2.5% and custody split between Coinbase and BitGo) in December 2025 and has amended the registration through 2026; the latest amendment reached the SEC in the autumn, under CIK 2029297 on EDGAR. No US spot product has launched.

Part of the delay was Washington. The federal shutdown that began on 1 October became the longest in US history and left the SEC running a skeleton operation for roughly six weeks, freezing hundreds of registrations. It was only lifting in mid-November, and the Commission signaled it would work through the backlog with fresh guidance; filings submitted during the shutdown can take effect automatically if they omitted a delaying clause, the mechanism that let the first spot XRP ETF list, as CoinCentral noted. With generic listing standards in place since September 2025, a conforming crypto product can reach an exchange in roughly 75 days. The broader countdown, from the ETF queue to a December policy cliff, is the subject of our read on the post-election calendar.

The twist ties the three threads together. Even if GTAO lists, its registration bars staking, so the ETF cannot pass the network’s roughly 10% yield to shareholders. A GTAO share would be exposure to TAO’s price and nothing more. The staked-treasury proxy TAOX and Europe’s staked STAO exchange-traded product both do pass yield, which is why, for now, the clumsier-looking vehicles give holders something the clean spot ETF cannot. Custody is its own fault line here; the role BitGo plays for GTAO is the same institutional plumbing we covered when the custodian itself went public.

Strip it down and an institution now chooses among three shapes of the same bet. A US spot ETF like GTAO would offer clean, regulated price exposure and nothing else. A listed treasury such as TAOX offers price exposure plus staking yield, wrapped in the risks of a small, loss-making equity. Europe’s STAO product stakes its TAO and rolls the yield into its value, but sits outside US market hours and rules. None is strictly better; they trade convenience, yield and regulatory comfort against one another, and the fact that the yield-bearing options are the ones that exist today is the clearest sign of how unfinished the US onboarding still is.

dTAO, alpha tokens and the halving math

Underneath the institutional story is the same token engine. TAO is hard-capped at 21 million and, like Bitcoin, pays a block reward that halves on a supply schedule; the first halving in December 2025 cut emissions to roughly 3,600 TAO a day, per Taostats. Each day’s rewards split by a fixed rule, 18% to the subnet owner and 41% each to miners and to validators and their delegators, and they are paid in each subnet’s alpha token rather than in TAO directly.

Since dTAO, those alpha tokens trade in their own automated market-maker pools, and the alpha price sets how much emission a subnet pulls, a continuous popularity contest priced in real time. If the idea of a token’s pool price steering real cash flows sounds familiar, it is the same machinery we unpacked in our primer on how automated market makers work. Around 70% of TAO supply is staked, earning the roughly 10% a year the treasury companies are chasing. The design rewards attention and liquidity as much as raw output, which is why ‘is the revenue real’ and ‘is the emission deserved’ have been the right questions all year.

Decentralization is still a 2027 promise

Every November headline eventually hits the same wall. Bittensor is governed, today, by a small center. The Opentensor Foundation runs the chain’s privileged root functions through a multisig, the quantum flag waits on a sudo call, and the network’s biggest economic experiments of 2026, Root Reborn (which turned validators into something like fund managers) and the proposed Gamma inter-subnet credit, were designed, gated and rewritten by the core team. Independent analysts also point out that root sits behind a two-of-three multisig and that the median subnet is still controlled by a single coldkey, a reminder that concentration exists at the subnet layer too.

The founders say they will let go. Steeves and cofounder Ala Shaabana stepped down from executive roles in February 2026 and have framed full decentralization as a journey ending around December 2027, arguing the network must stay fast to keep pace with AI; both points are paraphrase, not a claim they have disputed. The critics are blunter. Sam Dare, whose Covenant AI team built and then abandoned a flagship subnet in April 2026, called the promise that no single entity controls the network ‘a lie’ and the governance ‘decentralization theatre’, as he told The Block. Mark Creaser, who runs the DSV Fund, framed the year’s repeated rule changes this way: ‘a casino at least keeps the same rules all night; Bittensor changes the table mid-hand and calls it an upgrade’, while his partner Siam Kidd called the token ‘basically uninvestable’, in comments to tao.media.

The uncomfortable synthesis is that the same concentrated control that lets Bittensor ship a quantum upgrade overnight, rewrite emissions three times in a year and court Wall Street is the thing standing between it and the decentralization it sells.

How the three bids connect

It would be easy to read the quantum keys, the revenue and the stock as three unrelated press releases. They are not. Each answers a specific objection an institutional allocator would raise before touching a decentralized-AI token: is it safe to custody, does anyone actually pay for the output, and can I buy it inside a regulated wrapper. Hitting all three in the same month is a coordinated pitch to the kind of buyer Bittensor did not have a year ago.

The weakness of the pitch is that the answers are partial in the same direction. The keys are shipped but off. The revenue is real but subsidized. The proxy exists but is tiny, and the clean ETF is still waiting. In every case the network has built the mechanism and left the last, hardest step, activation, profitability, approval, decentralization, for later. That is a defensible way to build. It is also why the asterisks matter more than the headlines right now.

What to watch

• Whether root flips the switch that turns post-quantum signing on, and how many validators and treasuries actually migrate their keys once it does.
• Whether SubConnect’s revenue line clears $100 million and, more importantly, whether the share of emissions it covers keeps climbing past a quarter.
• The SEC’s post-shutdown backlog: whether GTAO (or a rival) lists via the generic standards, and whether any US product finds a lawful way to offer staking yield.
• TAO Synergies’ first earnings as a scaled treasury, including whether its shares trade above or below the value of the TAO it holds.
• Any move to formalize Gamma or to switch on the gated parts of Root Reborn, and what either does to sell pressure.
• The December macro calendar, an FOMC decision and a federal funding deadline, which will set risk appetite for every AI token, TAO included.

The bottom line

Bittensor in November 2026 is a more serious object than it was in the summer. It has quantum-ready keys in the runtime, a revenue line that is doubling, and a listed proxy that pays yield. It is also still a network where a handful of keys decide when the future arrives, and where the revenue, impressive as its growth rate is, remains a minority of the token budget. The three bids for legitimacy are genuine. So are the asterisks, and for now they are the more useful thing to track.

Frequently Asked Questions

Is Bittensor now quantum-proof?

Not yet. Runtime 488 added optional post-quantum signature schemes (ML-DSA-65, Falcon-512 and Falcon-1024) to the chain in early November 2026, but post-quantum signing stays disabled until the network’s root authority enables it, and the existing elliptic-curve keys are unchanged. It is a forward defense against a future quantum computer, not a live one, and no machine capable of breaking the current keys exists today.

How much revenue do Bittensor subnets actually make?

SubConnect’s revenue index estimated that revenue-generating subnets were earning about $70 million a year from external customers in early November 2026, up from about $35 million in September. That still covers under a quarter of the network’s $300-million-plus in annual token emissions, so most subnet rewards remain a subsidy rather than paid demand.

What is TAO Synergies (TAOX)?

TAO Synergies is a Nasdaq-listed company, formerly the biotech firm Synaptogenix, that rebuilt itself as a pure-play Bittensor treasury. It holds 54,058 TAO, says it is the largest publicly traded holder, and stakes those tokens to earn network rewards. It is small and unprofitable, and a second listed vehicle, xTAO, runs a similar strategy.

Is there a US spot Bittensor ETF?

Not yet. Grayscale has filed to convert its Bittensor Trust into a spot ETF (ticker GTAO) on NYSE Arca, but no US spot product has launched, partly because the record government shutdown froze the SEC through much of the autumn. Even if it lists, the filing bars staking, so the ETF would not pass the network’s staking yield to holders.

How is TAO’s supply structured?

TAO is capped at 21 million tokens like Bitcoin and pays a block reward that halves on a supply schedule; the first halving in December 2025 cut issuance to roughly 3,600 TAO a day. Rewards are split 18% to the subnet owner and 41% each to miners and validators, and are paid in each subnet’s alpha token under the dTAO system launched in 2025.

Marcus Okafor covers decentralized AI and crypto markets for HOGE Wire.