Root Reborn

What shipped

Spec version 441 is a release about a single flow: what happens to the emission the subnets owe the root network. Until this upgrade the chain enforced a bias on that flow — the dividend was sold into TAO the moment it arrived. As of 441 it is held instead. Each validator on root runs an escrowed holding of subnet alpha — a root basket — and the people who back that validator redeem their entitlement to it when they choose. The source for this page is one document: the project's own release note, titled "Root Reborn" and dated July 2026. It is an interested party's account of its own work — so what follows reports what that note says rather than repeating it as established fact. [source] [source] [source]

The release opens on the size of what it is changing, and the figures there are its own. Its headline number for the root network is 5,437,066 τ, which it presents as "$1.03B staked on root — 48.7% of every TAO ever minted"; the opening paragraph puts the same holding as nearly half of all TAO in existence and 73% of all staked TAO. Those are the note's numbers, quoted as it prints them, and nothing on this page checks them against chain state. [source] [source] [source]

The machinery the release describes is one fund per root validator, with no action required from the people staked behind it. Each epoch the dividend lands in that validator's basket. The holdings sit in chain-owned escrow — real stake entries under a pallet sub-account with no private key, which the release says means they cannot be moved or signed away, and because they are real stake they keep earning every epoch. When a dividend lands, each backer is credited a fraction of the whole fund priced at what selling the holdings would actually fetch at current pool depth rather than at a spot mark, so the release says existing holders are neither diluted nor gifted. [source] [source] [source] [source]

Two things about the launch state need saying plainly, because the headline promise and the shipped behaviour are not the same thing. Curation is switched off network-wide: every fund runs what the release calls the null strategy, in which the dividend is not sold at all but accrues in place as the alpha of the subnet that paid it, trade-free, and the call a validator would use to publish an allocation vector fails outright until governance enables it in a later upgrade. And joining root is now burn-based: registration charges the coldkey a demand-priced root burn instead of requiring stake up front. [source] [source] [source] [source]

What it changes for the ecosystem

For anyone staked on root — the position lesson 10 described as the one whose returns add up across every subnet a validator serves — the release's own instruction is that there is nothing to configure. TAO subscribed on root accrues yield automatically, btcli root list shows staked and accrued τ per validator, and btcli root claim is the exit path: omit the amount to realize accrued yield back into root stake, or pass one to withdraw to free balance. A claim is per validator and redeems your owed fraction of that one basket pro-rata. The release states that nothing is realized — and nothing is taxable — until you call it, and that per-validator payouts below the claim threshold (default 500,000 rao) are skipped and keep accruing, with no deadline and nothing that expires. [source] [source] [source] [source]

Anything already accrued moves across on its own. The per-subnet claimable state is migrated into baskets at the upgrade — previously accrued claimable alpha becomes basket holdings and entitlement on the same validator — and the release says nothing is lost and no user action is needed for past accruals. What does not survive the migration is the choice of payout form: the calls that set a claim type and a per-block claim count are removed, the Swap / Keep / KeepSubnets setting and the automatic per-block claim sweep are gone, and payouts are always TAO staked back to root. Anyone who was using that setting to end up holding subnet alpha must now get there by staking on the subnet directly. [source] [source] [source]

For the subnets the change is to a flow they were paying out every day. The release describes root's share of subnet inflation as structural sell pressure — 983 τ of subnet alpha marked and sold out of the pools daily, which it puts at 74% of the roughly 1,321 τ of fresh emission entering all subnet pools each day — and argues that this upgrade removes that drain outright, taking root's net flow at the pools from −983 τ/day to approximately zero. Its own framing of the result is a redistribution among subnets rather than a tax out of them: the subnets still pay root its proportion, and root now holds it inside them. Root proportion itself is not what changed here — what changed is what becomes of the alpha it entitles root to. [source] [source]

Lesson 09 described the slice of validator-destined pay that never reaches validators at all, because it is reserved for whoever staked TAO into the root network and accrues to them as claimable dividends. That reservation is intact; 441 changes only what happens to the reserved alpha on arrival, and therefore what a claim against it pays out. For root validators themselves the release frames the change as one in what competition on root is about: registration takes no prior stake, and the scoreboard it names is who made their backers the most TAO. Its argument is that a new entrant with a sharper read on the subnets can out-allocate an incumbent, post a better realized yield, and pull delegation away from it — an argument about what the rules now permit, not a reported outcome. Root still prunes its lowest-staked member when it is full, so the note tells a new validator to subscribe stake to its own hotkey to hold the seat. [source] [source] [source]

One consequence is easy to miss operationally, and it is about keys rather than money. Hotkey and coldkey swaps carry basket state — holdings, entitlements and claim watermarks — across to the new key, which means a hotkey with a live basket is not "clean" for reuse. If a subnet is dissolved, that subnet's basket holdings are converted into the TAO slot of each affected fund rather than disappearing. And the error that used to reject a root registration for insufficient stake is retired, so tooling that pre-funded root hotkeys before registering can drop that step and needs free balance for the burn instead. [source] [source]

Indexers and integrators get the longest list. The release asks them to regenerate metadata for the new per-validator claim call and for set_root_weights — which reuses the name of a retired emission-vote call index with new arguments and new semantics — to keep decoding the legacy claim_root, whose subnets argument still decodes but is now ignored, to drop the two retired claim-configuration calls, and to add the new basket events and the betaBasket_* RPC namespace. It also flags that set_root_weights is gated at launch and fails with a dedicated error until governance flips the flag. Operators are told to wait for the on-chain spec_version to move to 441 before upgrading nodes and clients, and SDK users to pull the matching release once the train publishes it. [source] [source] [source]

What to watch

The part of this release a reader would most want to judge is the part that is not switched on. The note's case rests on allocation — root validators sitting on the largest aggregated positions in the network and, for the first time, being paid to curate it — but the call that publishes a vector ships gated off so that the null-strategy baseline is established first, and curation opens in a follow-up upgrade. The yield claim is stated as an expectation in the same register: the release quotes a 6.6%/yr base yield at today's run-rate, says that under the old machinery it was a ceiling as much as a floor, and writes that it expects the effective yield distributed to TAO holders to rise with allocation skill. The note is equally plain that its own picture of the allocation layer is not drawn from live data: in the diagram of that layer the validator vectors are labelled illustrative, and only the destinations are real. [source] [source] [source] [source]

Unverified What that leaves measurable at launch is the flow change, not the allocation thesis. With every fund on the null strategy, the visible effect of 441 is that the daily sell of root's alpha stops and the alpha accumulates in escrow instead — which is a change in who holds subnet tokens and when they are sold, and says nothing yet about whether curated allocation beats the baseline. The note publishes no criterion for judging that comparison later.

Unverified The second thing to watch is where the new power lands. Lesson 10 made the case that concentration, not headcount, decides outcomes on this network; handing every root validator a published allocation vector over subnet alpha gives the same concentrated set a second lever, over which subnets receive a recycled dividend rather than only over which miners are scored well. The release's answer is that burn-based registration lets a better allocator displace an incumbent. Whether delegation actually moves on realized yield is an empirical question this document cannot settle.

btcli subnets burn-cost 0                                     # current root registration price
btcli subnets register --netuid 0 -w my_coldkey -H my_hotkey  # register; no prior stake needed

btcli root set-weights --weights "0:0.2,4:0.3,8:0.5" -w my_wallet
btcli root get-weights --hotkey 5F...
btcli root show --hotkey 5F...   # your fund: weights, holdings, NAV

btcli root list                                    # staked + accrued τ, per validator
btcli root subscribe --amount 100 --hotkey 5F...   # assets in
btcli root claim                                   # pick wallet → validator → claim / withdraw
btcli root claim --hotkey 5F... --amount all       # withdraw full position
btcli root claim --hotkey 5F...                    # claim accrued into stake only

import bittensor as bt
from bittensor.wallet import Wallet

wallet = Wallet(name="my_coldkey")
sub = bt.Subtensor()

owed = sub.read(
    "root_basket_owed_breakdown",
    coldkey_ss58=wallet.coldkeypub.ss58_address,
)

result = sub.execute(bt.ClaimRootWithHotkey(hotkey_ss58="5F..."), wallet)
result = sub.execute(bt.ClaimRoot(subnets=[0]), wallet)