NEAR Intents leads the native-swap venues, but too little of it reaches the NEAR token yet. First of four native-swap venues with 58% over 90 days (46% over 30, 38% in the latest week); third of eight, with 18%, once four bridges count. The call is to watch, not buy, until buybacks cover 50% of new NEAR and the 30-day share is back above 50%.

The call: WATCH. The business is real and leading; the token's cut of it is not yet large enough, and the lead is slipping in the latest weeks.

Is Intents the leader?

58% of four-venue swap volume over 90 days, up from 40% a year earlier; THORChain is second with 30%.
First on every consistent way of counting: 58% as reported, 56% counting one side of each swap, 60% on volume rebuilt from the chain. Only halving Intents alone, against the rivals' own indexers, puts THORChain ahead.
When THORChain halted for 37 days, Intents took 41% of the flow it lost, but growth that size was ordinary for Intents (52nd percentile of the year before), so it shows capacity, not cause.
Add the four large bridges and Intents has 18%, rank 3 of 8. When THORChain halted, swap venues behaved as its substitutes and no bridge did.
It does not win on price: 15.5 bps of volume in gross fees against a rival median of 13.0 bps.

Does it reach the NEAR token?

27% of gross Intents fees reach NEAR's revenue wallets, rising every 90 days: 8% → 12% → 20% → 27%; the rest stays with solvers and the front ends that distribute Intents.
Buybacks took 1.00M NEAR against 7.96M newly issued over 90 days, 13%. Covering half would take 4.0× today's captured revenue.
The buybacks are purchases, not transfers: 99.6% of the NEAR the wallet received from intents.near over 90 days was the output of a swap into NEAR, 51% through a 1Click router and 48% swapped by the paying account itself; nothing left the wallet in the window, so it is held.
Twelve months out, buybacks would cover 4% of new NEAR in the bear case, 13% in the base and 36% in the bull; even the bull case needs 70% of fees reaching NEAR to get to half.

What is priced in

NEAR's price is not paying for Intents. Valued as a perpetuity at 13.5% discount and 7.5% growth, the cash Intents sends NEAR is worth $0.26 a token, 7% of the $3.50 price (4% to 22% across the range), and 21% in the bull case. On its own it would need 13× today's cash to justify the price. New NEAR costs holders $114M a year in dilution against $14M of buybacks.
So the call does not rest on a re-rating: an investor in NEAR today is buying the L1 and an option on Intents' cut rising. ACCUMULATE needs that cut to rise until buybacks cover half of new NEAR, which in the model means about 30% of today's price backed by Intents cash; AVOID fires if the cut starts to fall.

Where the lead could break

Distribution is rented. SwapKit brings 38% of 1Click volume and 55% comes through front ends that also route to THORChain, Chainflip or Maya. They pick the best quote each swap; nothing locks the flow to Intents.
It is not the cheapest. Intents charges 15.5 bps of volume; THORChain and Maya Protocol charge less. An aggregator that routes on price moves flow the day a rival undercuts.
The front ends earn more than NEAR. Front ends add 43.0 bps in their own fees on top of the protocol's 15.5 bps, and 73% of the protocol's fees stay with solvers and front ends. Raising NEAR's cut means taking from the parties that bring the volume.
THORChain is back. THORChain took 49% of the four venues in the latest week against Intents' 38%, up from 19% four weeks earlier; the first week it has led since the week of 2026-04-20.
Per dollar of volume, NEAR is priced far above its rivals. Intents keeps 4.2 bps of revenue per dollar of volume, THORChain 3.5 bps, Chainflip 5.8 bps; the market pays 21× revenue for THORChain's token and 223× for NEAR's, because NEAR's price is for the whole L1.
Less of it can be checked. As settlement moves to the private shard, the public chain shows the value moving but not who filled it; a loss of share to internal flow would be invisible here.

What the public chain cannot see

96% of last week's on-chain Intents volume shows only as money entering and leaving 1Click routers, with no public fill behind it, as settlement moves to NEAR's private shard; the public chain verifies the value moved, not who filled it.
The reported series matches what the public chain sees (1.07×), so private flow does not inflate the 58% share; swaps that stay inside the shard are missing from both, which can only understate it. Capture could be off either way.

Why these numbers can be trusted

Intents volume rebuilt from the chain, every day: 1.04× reported. Every log line intents.near wrote (NEAR's public BigQuery dataset); within 20% on 69 of 71 days since the settlement method changed.
Rivals' double counting measured from their own indexers: 8,886 swaps read. THORChain, Maya and Chainflip volume is reduced to one side of each swap with factors read from Midgard and the Chainflip indexer.
Revenue matches the revenue wallets: 98.5%. Transfers into NEAR's three revenue wallets against the revenue figure used.
Buybacks traced to the swaps that bought them: 99.6% swapped. Every payout from intents.near to the buyback wallet is matched to the wNEAR a 1Click router delivered, or to a swap the paying account signed.
Supply from block headers, activity from two indexers: 454 of 455 days agree. NEAR issuance is read from total_supply in block headers; daily L1 counts from NearBlocks and BigQuery agree to within 1%.
Headline recomputed in a TEE, checked to Intel's root: 13/13 checks. The same WASM recomputed the share inside an Intel TDX enclave; its quote is verified here down to Intel's root CA, TCB UpToDate.
Thresholds, the call's rule and predictions sealed: 3 seals, 2 on NEAR. Hashes timestamped on Bitcoin and written on NEAR mainnet; the build stops if any sealed value changes.
Every number rebuilds, byte for byte: build.py --offline. The offline build regenerates the page, THESIS.md and METHODS.md from the committed snapshot, byte for byte, checked on every push; a full build refreshes the data but keeps the thesis on its fixed window.

14 claims

C1 SUPPORTED: NEAR Intents is the largest swap venue with more than 50% of volume, whether volume is taken as reported, one side of each swap is counted for every venue, or Intents volume is rebuilt from intents.near itself.
C2 REFUTED: Most of the year-on-year volume gain came from taking share, not from market growth.
C3 SUPPORTED: The peer-set market itself grew more than 25% year on year.
C4 SUPPORTED: Intents charges at least the median take rate of its rivals (it is not winning on price alone).
C5 REFUTED: NEAR keeps more than 50% of the gross fees Intents generates.
C6 REFUTED: The lead is stable: weekly share never fell below 40% in the last 26 undistorted weeks.
C7 SUPPORTED: When its largest rival stopped trading, Intents picked up more than 33% of the flow that rival had carried.
C8 SUPPORTED: Intents daily volume is steadier than every rival's (lower standard deviation relative to its mean).
C9 SUPPORTED: The share of fees reaching NEAR rose in every one of the last 4 consecutive 90-day windows.
C10 REFUTED: No single distribution channel carries more than 33% of 1Click swap volume.
C11 REFUTED: Intents is also the largest venue when the four big bridges are added to the peer set.
C12 SUPPORTED: The captured-revenue figure used here matches the transfers into NEAR's three revenue wallets within 5%.
C13 SUPPORTED: The reported Intents volume counts both sides of each swap, like the rivals' figures: on every day rebuilt from intents.near token_diff events (through August, where the method applies) it sits inside the both-sides interval and outside the one-side interval.
C14 REFUTED: The NEAR bought back with Intents revenue covers at least half of the NEAR newly issued in the same window.

Dated predictions

P1, due 2027-02-28, PENDING: NEAR Intents still holds at least 50% of peer-set volume over the trailing 90 days.
P2, due 2027-02-28, PENDING: NEAR's captured revenue is at least 30% of gross Intents fees over the trailing 90 days.
P3, due 2027-05-31, PENDING: The Intents take rate has not been competed below 12 bps over the trailing 90 days.
P4, due 2026-11-12, MET: NEAR Intents holds at least 40% of the four swap venues' volume over 2026-11-11 and 2026-11-12, days after the data this thesis was built on.
P5, due 2026-11-12, MET: The Intents take rate stays at or above 12 bps over 2026-11-11 and 2026-11-12.
P6, due 2026-12-10, PENDING: NEAR Intents keeps at least 40% of the four swap venues' volume over 2026-11-13 to 2026-12-10 (the thesis's kill line).
P7, due 2026-12-10, PENDING: NEAR Intents is still the largest of the four swap venues over 2026-11-13 to 2026-12-10 (its share minus the largest rival's is above zero).
P8, due 2026-12-10, PENDING: Intents volume rebuilt from intents.near events with the sealed BigQuery method stays within 0.90 to 1.20 times the reported volume over 2026-11-13 to 2026-12-10, so the headline keeps an on-chain check.

What this cannot say

Confidential Intents settle on a private shard: the public chain verifies the value that crosses intents.near, not the fills behind it, and swaps that stay inside the shard are not counted at all.
The router rule that rebuilds Intents volume after August was found by looking at the data; it is sealed now and tested on days not yet seen (P8).
Which market is the right one rests on one natural experiment, the THORChain halt.
Claim thresholds and the call's rule were set after a first look at the data; they are sealed so any later change shows.

The short thesis is THESIS.md; every method and check is METHODS.md. Turn on JavaScript for the charts.

Share/Shift