# NEAR Toll Audit: the memo

Chain data through {{m:as_of:text}}; prices and fees pulled {{m:market_as_of:text}}. The numbers in this file are frozen on that snapshot, the same one the [dashboard](https://apps.swiftbotlab.workers.dev/near-toll-audit/) and the README use. Kill rules T6 and A5 were added on 2026-11-12 after review and graded on it too. This is research, not investment advice, and we make no claim about holdings.

## The thesis

NEAR is turning into a toll road for cross-chain settlement, and the road is real. The token does not own that road yet, and at {{m:price:usd2}} it is priced as if it mostly did.

We split this into two claims, because they can fail separately.

- **The toll road claim:** NEAR Intents stays the leading native swap rail through 2027 and keeps a gross fee above 8 bps. We think this is more likely than not.
- **The token claim:** the price is ahead of what can reach holders. Our probability-weighted value is {{model:weighted}}, {{model:gap}} below the market on a {{model:bull_weight}} bull weight. That result leans on the weight: it is {{model:weighted_bull_10}} at 10% and {{model:weighted_bull_30}} at 30%, and about {{model:implied_bull}} gets to the market price.

## What has to be true at {{m:price:usd2}}

The price needs a mix of these:

1. Volume compounds fast. At base-case inputs that is {{model:implied_cagr}} a year for five years, to {{model:implied_y5}} a year, {{model:implied_multiple}} today's run rate and {{model:implied_y5_share}} of what every DEX does in a year now. A 40x exit lowers that to {{model:implied_multiple_40x}}, and a $3.9B chain value lowers it to {{model:implied_multiple_low}}.
2. NEAR's share of gross fees rises from {{m:capture_30d_pct:pct0}} to 45%, while the gross fee holds near 12 bps.
3. Most captured revenue reaches holders. No rule requires that today.

## What the evidence says

**The base layer did not grow into the valuation.** One account, `{{m:top_signer_mar26:text}}`, signed {{m:top_signer_share_mar26:pct0}} of March 2026 transactions. It stopped within 32 hours on 1 and 2 April, and transactions fell {{m:tx_drop_apr26_abs:pct0}} in a month. About {{m:kaiching_deletes_oct26:num}} of its accounts were deleted in October, so we leave them out of signer counts. Engaged signers are down {{m:signers_engaged_change_abs:pct0}} since July 2025, mostly Telegram game accounts (down {{m:tg_engaged_change_abs:pct0}}). Fee burn covers {{m:burn_share_of_issuance_pct:pct1}} of issuance.

**One signal cuts the other way.** Engaged self-custody signers fell {{m:self_engaged_change_abs:pct0}} over the same period, but rose {{m:self_engaged_rise_since_aug:x1}} from August to October, to {{m:signers_engaged_self_last:n}}. We do not know why. Named `.near` accounts also retain far better ({{m:ret_named_m1:pct0}} back after a month) than hex accounts ({{m:ret_hex_m1:pct0}}).

**Settled flow is real.** The official dashboard and DefiLlama match to the dollar only since {{m:official_llama_match_since:text}}, so we rebuilt flows from raw token events. One-way on-chain flow runs at {{m:flow_ratio_median:pct_from_x}} of reported volume, between {{m:flow_ratio_low:pct_from_x}} and {{m:flow_ratio_high:pct_from_x}} across 26 weeks. That includes solvers rebalancing through the same contract, so it is not user demand. Intents holds {{m:rail_share_13w:pct0}} of volume among four native swap rails with public data over 13 weeks, down from {{m:rail_share_jun26:pct0}} in June.

**End-user demand is a range, not a number.** {{m:wd_small_count_share:pct0}} of withdrawals were under $1,000, but the top 10 addresses took {{m:wd_top10_value_share_prev:pct0}} to {{m:wd_top10_value_share:pct0}} of value in the two weeks we priced. So end-user demand sits between about {{m:wd_floor_share_of_reported_pct:pct0}} and {{m:wd_ex_top10_share_of_reported_pct:pct0}} of what is reported: the first leaves out the 100 largest addresses, the second only the 10 largest. One-way flow, {{m:flow_ratio_median:pct_from_x}}, is a ceiling that still includes rebalancing.

**The fee base is narrow.** Swap-level data from the Intents Explorer covers {{m:ex_coverage_pct:pct0}} of reported volume and {{m:ex_fee_coverage_pct:pct0}} of reported fees on the same days. The remaining volume pays at most about {{m:ex_rest_bps:bps1}}. Inside 1Click, `{{m:ex_top_fee_partner:text}}` carries {{m:ex_top_fee_partner_vol:pct0}} of volume and {{m:ex_top_fee_partner_fee:pct0}} of app fees, at about {{m:ex_top_fee_partner_bps:bps0}}. Zcash sits on one side of {{m:ex_zec_pct:pct0}} of that volume.

**Wallets keep most of the toll, but less than they did.** The protocol's own fee is {{m:intents_fee_pips:n}} pip. Everything else is commercial. By NEAR's count, NEAR kept {{m:capture_30d_pct:pct0}} of gross fees over 30 days, against {{m:capture_all_time_pct:pct0}} all time; partners, wallets and solvers kept the rest. Most of that gain came from NEAR's own front end. Of every $100 swapped, {{m:per100_gross:usd2}} is charged ({{m:gross_bps_30d:bps1}}), and NEAR's {{m:capture_30d_pct:pct0}} of it is {{m:per100_captured:usd3}}.

**What reaches the token is discretionary.** The official page says buybacks "permanently remove NEAR from circulation". Its FAQ says revenue "can be deployed" for buybacks, locking or staking. So far the chain shows {{m:revenue_wallets_near:nearm}} held in two wallets, {{m:revenue_held_share_pct:pct0}} of official inflows, with the largest daily burn since the buyback began at {{m:largest_daily_burn_since_buyback:near0}}. A {{m:buyback_confirmations:n}}-confirmation multisig can move the buyback wallet. In {{m:window_month_name:text}}, burn plus revenue equalled {{m:offset_30d_pct:pct0}} of issuance at best. Separately, `treasury.near` sent {{m:treasury_moved_near:nearm}} out in April and May, {{m:treasury_moved_multiple:x1}} what those wallets hold. We can see the first hop and not what followed.

**Pricing power, our new test, came back inconclusive.** A toll road is worth what it can charge, so we tried to measure it. The take did not step when the partner revenue share went live in February. It rose in May and again from September. Rail share was unchanged in one window ({{m:switch_share_before:pct0}} to {{m:switch_share_after:pct0}}) and lower in the other ({{m:step_share_before:pct0}} to {{m:step_share_after:pct0}}), and neither interval rules out a 25% loss against rivals. Across {{m:fee_events_rises:n}} lasting rises in a partner's posted fee, its share changed {{m:fee_rise_share_change_pct:signed0}} on average, with a 95% interval of {{m:fee_rise_lo_pct:signed0}} to {{m:fee_rise_hi_pct:signed0}}. Most of the capture gain is mix, not price. So the fee inputs below are assumptions, not findings.

**Volume does not lead usage yet.** Over {{m:leadlag_weeks:n}} weeks, Intents volume beat price as a leading indicator in {{m:leadlag_volume_wins:n}} of 6 forward tests. Out of sample, volume helped in {{m:leadlag_oos_beats_own:n}} of 6 and price in {{m:leadlag_oos_price_beats_own:n}} of 6.

**The market has already moved.** NEAR rose {{m:near_chg_1y_pct:pct0}} in a year while the median of seven peer chains fell {{m:peer_chg_1y_median_abs:pct0}}. It trades at {{m:near_multiple:x0}} captured revenue (October's {{m:rev_run_rate:usd}} a year), or {{m:near_multiple_ex_chain:x0}} after subtracting a peer-median chain value, against {{m:comps_median_multiple:x0}} for ten app tokens with holder revenue. Against gas fees on other chains it looks ordinary, at {{m:near_mcap_to_kept_fees:x0}} what it keeps once L1 fees are added ({{m:near_kept_annual:usd}} a year). Neither comparison is clean.

**Also measured.** On the public dataset, unlocks are largely over: {{m:locked_bq:nearm}} is not yet circulating, down from {{m:locked_bq_2023:nearm}} in January 2023. Staking holds {{m:stake_ratio_pct:pct0}} of supply and pays about {{m:staking_yield_net_pct:pct1}} after commission. {{m:nakamoto_third:n}} validators hold a third of stake. Contracts with AI or agent names took {{m:ai_named_share_pct:pct2}} of October transactions, though `escrow.ai.near` signed {{m:escrow_ai_share_pct:pct0}} of direct calls into `intents.near`.

## Valuation

Five years of holder earnings, an exit multiple, and a stated value for NEAR as a chain apart from Intents. All inputs are editable on the live page.

| Case | Weight | Volume growth | Year-5 gross fee | Year-5 capture | Reaches holders | Exit | Chain value | Value per NEAR |
|---|---|---|---|---|---|---|---|---|
| Bear | {{s:bear:probability:pct0}} | {{s:bear:cagr:pct0}} | {{s:bear:gross_bps_end:bps0}} | {{s:bear:capture_end:pct0}} | {{s:bear:claim:pct0}} | {{s:bear:exit_multiple:x0}} | {{s:bear:l1_value:usd}} | {{model:bear}} |
| Base | {{s:base:probability:pct0}} | {{s:base:cagr:pct0}} | {{s:base:gross_bps_end:bps0}} | {{s:base:capture_end:pct0}} | {{s:base:claim:pct0}} | {{s:base:exit_multiple:x0}} | {{s:base:l1_value:usd}} | {{model:base}} |
| Bull | {{s:bull:probability:pct0}} | {{s:bull:cagr:pct0}} | {{s:bull:gross_bps_end:bps0}} | {{s:bull:capture_end:pct0}} | {{s:bull:claim:pct0}} | {{s:bull:exit_multiple:x0}} | {{s:bull:l1_value:usd}} | {{model:bull}} |

In the base case, chain value is {{model:base_chain_share}} of the total ({{model:base_chain_value}} of {{model:base_equity}}, or {{model:base_chain_per_near}} a NEAR over {{model:supply}}), and it is an assumption: the median fully diluted value of seven peers.

**How the weights were set.** Bear gets 35% because the lead is young: Intents took it in late 2025, the weekly leader among rails has changed {{m:rail_leader_changes:n}} times since January 2025, and share has slipped since June. Bull gets 20% because it needs about 14x volume, 60% capture and a binding rule together. If each is 50% to 60% likely, the joint odds are 13% to 22%. These weights are judgement, and the bull case is most of the result, so change them on the page.

## Kill criteria

Graded by code at every build, with no human step.

- **The toll road claim is dead if:** rail share falls below 40%, gross fee below 8 bps, or the three-month captured rate below 2.5 bps, by 2027-12-31; on-chain flow falls below 35% of reported volume; or annualised volume falls below $26B by 2027-06-30. Added on 2026-11-12: an exploit leaves more than $10M of user funds unrecovered within a year (today: $0, the $3.8M taken on 1 October 2026 was returned).
- **Our caution on the token is dead if:** captured revenue passes 50% of issuance value by 2027-12-31; annualised revenue passes $100M by 2027-06-30; a burn above 100,000 NEAR in a day shows a binding rule; or the multiple on revenue falls below 60x by 2027-12-31. Added on 2026-11-12: the market's premium over our base case widens past 100% (today {{model:premium_over_base}}).

## What we could not measure

Confidential intents, off-chain AI usage, Foundation holdings and sales, and who sits behind large addresses. Relay, Across and CCTP volumes are behind a paid API. Volume has about 22 months of history and fees about 18, in one market regime.

## The strongest case against this thesis

These are the best arguments we can make against ourselves, tested in the same model.

**1. You are valuing a chain like a stock.** Chains trade on monetary premium, not cash flow. Seven peers trade at a median of {{m:peer_mcap_to_fees_median:x0}} fees, and NEAR is at {{m:near_mcap_to_kept_fees:x0}} what it keeps. Count all Intents gross fees and it is {{m:near_mcap_to_all_fees:x0}}, though most of those are not NEAR's. This is the strongest objection, which is why chain value is a separate input and the largest bar in the sensitivity chart. Give NEAR Avalanche's and Sui's level, $3.9B, and the base case rises to {{model:if_peer_chain_value}}.

**2. Capture went from single digits to {{m:capture_30d_pct:pct0}} in a year. Extrapolate the trend, not the level.** Captured revenue is rising far faster than volume, and it was {{m:rev_run_rate:usd}} annualised on October alone. If that continues, row A2 trips and we are wrong. Our reply is that capture has a ceiling, and that most of the gain so far is NEAR's own front end winning share. The evidence on that limit is thin.

**3. Discretionary is how every buyback starts.** House of Stake is passing real changes: the issuance ramp stands at {{m:issuance_vote_for_pct:pct1}} in favour, and removal of the gas rebate has been approved. Bought NEAR sits off the market today whether or not it is burned. A binding rule could be one vote away, and it would lift the base case to {{model:if_binding_rule}}. Against that, {{m:issuance_vote_voters:n}} accounts voted, and veNEAR is {{m:venear_supply:nearm}} on a supply of {{model:supply}}.

**4. Reported volume understates the business.** Confidential flow is invisible to us, and volume that pays little today is an option on a future fee. That option is in our bull case. It has not been exercised, and charging it would test the same pricing power we could not establish.

**5. Flows set the price, not your model.** The Bitwise ETF holds {{m:etf_near:nearm}}, {{m:etf_to_buyback_ratio:x1}} what the revenue wallets hold, after about a month. Together they equal {{m:absorbed_pct_of_year_mint:pct0}} of a year's new NEAR. Another ETF is pending. Price can stay above fair value for years. We agree. This is a view on value, not on timing, and it is not a price target.

**6. Your frame leans bearish.** Five years is short for a toll road, and 15% is a stiff rate. Use 12% and a 40x exit and the base case is {{model:if_long_runway}}. Add a $3.9B chain value and it reaches {{model:if_chain_and_runway}}, against a market price of {{m:price:usd2}}, with no binding rule at all. Our caution therefore depends on chain value and the exit multiple at least as much as on anything we measured.

**Where this leaves us.** The measured facts are solid: settled flow is real, fees are concentrated, and the claim is discretionary. The valuation is softer, because the assumptions that move it most are the ones the chain cannot settle. If we are wrong, the likeliest reason is that the market keeps paying a chain premium a cash-flow frame cannot see. Row A5 now watches that premium: if it widens past 100% of our base case, our caution is wrong.