The structural inputs, measured on chain, each placed in its own measured range rather than against anyone's assumed scenarios. A row says whether this figure is high or low for this asset, lately β not what it will do next. The notes record how one circulating valuation writeup labelled the same inputs; that writeup prompted these questions but its cases are assumptions, and nothing on this page is scored against them.
Priced on chain against USDG, a USD stablecoin, so this is the pool's own execution price rather than an aggregator's opinion. Market cap is that price times live supply.
Every other measure on this page is backward-looking: volume that happened, fees that were taken, flow
that already crossed. This one faces forward. Concentrated liquidity sitting below spot is money
already committed to buying AI; AI sitting above spot is supply already committed to selling. The
gap between them says which direction is cheaper to push, before anyone trades.
Rebuilt from every ModifyLiquidity event the pools have ever emitted, so it is measured
rather than quoted β two checks confirm the replay: net liquidity across all ticks must come to exactly
zero, and the liquidity active at the current tick must equal what the pool reports in its own last swap.
Liquidity is not a promise. A position can be withdrawn in one block, so read this as the shape of
the book right now, not as support that will be there later.
Who is absorbing whom. Buy/sell imbalance turns before price does, because it is what moves it.
All three fee legs, annualised. This is the protocol's revenue, and it is paid in AI, so it is also the burn engine.
NVDA accumulating in the vault. Unlike the burn, this is an asset that cannot be inflated away.
Cross-routed AI volume Γ· direct AI volume. Swaps are grouped by transaction: one AI swap in a
transaction is direct β someone wanted AI. Two or more, where the trader receives AI in one leg
and spends it in another, means AI was a pass-through hop (BONER β AI β MEME); the routed
amount is min(received, spent) and any residual counts as direct.
It separates βpeople want to own AIβ from βAI is the plumbing everything else moves throughβ. Only the
second is the liquidity-hub thesis, and it is the more durable, because it does not need anyone to hold
an opinion on AI at all.
Share of indexed AI volume crossing the tolled AI/NVDA pool. Revenue scales with this, so the hub's own success can cannibalise it.
AI destroyed or immobilised, cumulatively. A slow, one-way squeeze on what can actually trade.
Market cap divided by the annualised fee run-rate. Both scale with the AI price, so it cancels: this multiple is price-invariant and needs no oracle. Lower is cheaper.
Share of AI volume on fee-bearing (hooked) pools versus hookless ones. Anyone can open a v4 pool for AI with no hook and a lower fee; routers will prefer it, and the vault gets nothing. Measured across the venues indexed in depth rather than all ~5,600 pools that hold AI: only the deep set has hourly volume, and it is chosen as the busiest, so it carries most of the flow but is not a census.
Every indexed venue for AI, its fee, and whether it pays the vault. Sorted by recent volume.
Fee run-rate capitalised across a band of discount rates. Only the run-rate is measured; every rate is an assumption, and the spread between them is wider than most of the figures on this page β read the range. USD figures use a public aggregator's price and are a convenience, not an on-chain measurement.
The thresholds worth watching, stated in advance so they are not rationalised after the fact.
Bars above the axis are AI bought (leaving the pool); below are AI sold. Derived from Uniswap v4 Swap logs β every trade, not a sample.
Running sum of AI bought minus AI sold. A rising line means traders are net absorbing AI.
Pool price from sqrtPriceX96, in quote units per AI. Shown on its own axis β never overlaid on flow.
AI sent irreversibly to 0x0 by the fee splitter. Burn rate is a direct read on fee-paying volume.
Cumulative NVDA stock-token balance accumulated by the community vault.
The table equivalent of the two charts above, most recent first.
Imbalance is (bought β sold) Γ· (bought + sold) in AI terms. The table is the accessible view of the charts above.
Most recent swaps across the indexed AI pools.
Genesis supply decomposed into what is gone for good, what is immobilised, and what can actually trade. Every share here is of the genesis mint, so the segments add to 100%; the tiles above measure holdings against current supply instead, which is why the float reads slightly higher there.
Cumulative AI burned and AI locked in the vault. Both legs are irreversible.
Measured from transfers, not assumed. Each leg is a share of total AI fees taken by the hook.
Cross-routed AI as a share of direct volume, counted from transactions where AI is a genuine pass-through hop rather than assumed. Markers are this series' own 10th, 50th and 90th percentiles, so the scale comes from what the token has actually done.
Daily AI volume split by whether AI was the destination or merely the bridge.
Pools opened against AI each day that are still trading now; hover for how many opened in total. Formation rate is the clearest read on whether AI is being adopted as a venue β but the surviving count is survivorship-filtered, so older days under-report by however many bridges have since gone quiet.
For each token, the share of its total measured on-chain volume that settles on its AI bridge rather than its other venues. Tokens the launchpad created against AI are marked native and sit near 100% by construction β that is a fact about how they were minted, not evidence of AI winning flow. The organic rows, where a token had its own venues first and grew an AI pool later, are the ones that carry any hub claim. Each rowβs two volume columns are in that tokenβs own units, so they are comparable down a row but never across rows β summing them produces a number with no meaning. The population figure is therefore each tokenβs own share weighted by the AI moving through its bridge, with the median shown beside it.
Observed pass-through paths, by routed AI volume.
Direct (non-routed) AI volume by paired token.
Every number on this site comes from Robinhood Chain logs via public RPC. Nothing is modelled or estimated unless it says so.
Each address was established by tracing real transfers, not from a directory.
Assertions that must hold if the indexing is correct.