Originally published at mrnasdog.com/research/arb/inflation by MrNasdog.
Arbitrum has never created a single new ARB — the token was minted once, in full, at launch, and a search of every transfer the contract has ever emitted finds exactly one issuance event — and yet the Pressure Framework reads ARB at +6.40% over the trailing 90 days and +4.81% over the next 90. All of it comes from two calendars rather than from issuance: monthly team and investor unlocks of 92.6M ARB running to Mar 16 2027, and 230.0M ARB the Arbitrum DAO treasury handed to the Arbitrum Foundation on Jun 28 2026. Sell pressure is 427.6M ARB, buy pressure is 0, and the ceiling is 10,000M ARB — which a DAO vote can lift by 2% a year, and never has.
The verdict, in one paragraph
Against a circulating base of 6,678M ARB, the framework books 427.6M ARB of sell pressure and 0 of buy pressure over the trailing 90 days — a net of +6.40% — and projects +4.81% for the next 90 days, the difference being the one-off Arbitrum Foundation grant dropping out of the forward window while the unlock calendar continues untouched. The inflation monitor reads +6.48% for the same window, a gap of 0.08 percentage points, comfortably inside the framework's 0.5pp tolerance, so this page ships with no data-conflict warning: two independent measurements of Arbitrum supply agree. The label for ARB is a capped token diluting on a schedule: nothing is minted, nothing is burned, and the float still climbs every month because coins that already existed stop being restricted.
Sell pressure: where new ARB comes from
It does not come from minting, and the proof is unusually clean. Every ARB in existence was created in one transaction at Arbitrum One block 70,398,215 — 10,000M ARB to a single distributor — and the full history of the contract holds no second issuance. Since then 1,022 ARB have been destroyed by hand, which is why the count of ARB in existence reads 9,999,998,977 rather than a round ten billion; the two figures reconcile to the eighth decimal place. So Sell #1, protocol inflation, is 0. It is not, however, closed. The ARB token still carries a live mint function restricted to the Arbitrum DAO's own executor, capped at 2% of supply once every 365 days, and that permission unlocked on Mar 15 2024 and has never been exercised — the contract's own counter has not moved since. Asked to mint one coin from the DAO's address, the contract agrees; asked for 200M ARB, one unit above the cap, it refuses in its own words. Docs are policy and bytecode is mechanism, and here they say the same thing: a single vote stands between this row and roughly 200M ARB of genuine issuance. The row is watched, never marked permanent.
The largest line is Sell #2, vesting unlocks, at 277.9M ARB. Team, advisors and investors received 4,448M ARB at launch under a four-year lockup: a cliff in Mar 2024, then one forty-eighth on the sixteenth of every month until Mar 16 2027. Three of those tranches fell inside this window — Jun 16, Jul 16 and Aug 16 2026 — at about 92.6M ARB each. It matters how Arbitrum implements this, because it decides which number is correct. The two original allocation contracts hold 0 ARB today: the team allocation passed through a routing address and the investor allocation went straight to 228 individual wallets at launch. There is no escrow, no vesting contract and no lock holding the unvested balance, so an Arbitrum unlock date lifts a written restriction rather than moving a coin, and realised on-chain outflow measures nothing. The published schedule is therefore the right measure, and the error it can make is one-directional: it overstates sell pressure if recipients sit on their coins, and can never understate it. 648.5M ARB remains on the calendar across seven more tranches — a figure the Arbitrum Foundation's own mid-August report independently corroborates at 0.77B still vesting.
Sell #3, Foundation and unscheduled unlocks, is 149.6M ARB, and it is where the Arbitrum DAO treasury matters. That treasury sits outside the circulating float, so anything leaving it reaches the market by definition. In the whole 90-day window it made exactly one ARB payment: 230.0M ARB to the Arbitrum Foundation on Jun 28 2026, funding a year of operations under an on-chain vote that closed Jun 25 2026. Running the other way, 129.3M ARB came back to the treasury as an older funding vehicle wound down and returned capital, and that leaves the float again. Alongside it, the Arbitrum Foundation's own 700M ARB lockup released 48.9M ARB on its straight line to Apr 17 2027. Net of returns, 149.6M ARB left team control. Sell #4, long-term locked or bankruptcy, is 0: ARB has no bankruptcy estate, no trustee and no court-ordered distribution.
Buy pressure: where new ARB goes
Nowhere at all, and not by accident. Buy #1, programmatic buyback, is 0 — the Arbitrum DAO runs no programme that spends treasury money repurchasing ARB, and the DAO's revenue lines (Arbitrum One transaction fees, the Timeboost priority-ordering auction, and Arbitrum Expansion Program licence fees) are collected and kept rather than recycled into the token. A bond-funded ARB buyback has been floated on the Arbitrum governance forum and never reached a vote. Buy #2, protocol fee burn, is 0, and unlike most chains this is not an idle mechanism but an absent one: Arbitrum charges gas in ETH, so a transaction never consumes ARB. Both destruction surfaces were read at both window ends regardless — the unspendable address held 1,533 ARB on both dates, and the count of ARB in existence moved by two hundredths of one coin. Buy #3, Foundation buy, is 0: the Arbitrum Foundation's ARB balance grew hard, from 85.2M to 324.1M, but every coin came from the DAO treasury and from its own lockup rather than from a purchase, so the framework tracks it as sell-side overhang instead of crediting it as demand. Buy #4, new long-term lock, is 0 — ARB staking is approved in principle with a liquid receipt token designed for it, but it is not live, no contract holds locked ARB, and staked tokens would count as float anyway.
Foundation and overhang
The ARB overhang is enormous, fully enumerated, and mostly discretionary — the opposite shape to a pure vesting story. The dominant item is the Arbitrum DAO treasury at 2,556.1M ARB, down from 2,656.9M at the window open; it is read from the chain at every rebuild. Second is the Arbitrum Foundation's vesting wallet at 105.9M ARB, down from 154.8M, emptying on its straight line by Apr 17 2027. Third is the 648.5M ARB still on the team and investor calendar. Those three buckets plus a residual of 11.4M ARB reconstruct the entire excluded supply — the difference between the 10,000M ceiling and the 6,678M circulating base — which is how the framework knows where the boundary falls: the DAO treasury and the Foundation's vesting wallet are outside the float, and the Foundation's operating wallet, now holding 324.1M ARB, is inside it. That distinction is the whole reason the 230.0M ARB grant counts as sell pressure even though the Arbitrum Foundation has not sold a coin of it. The trigger sentence applies to all three: if any of these balances falls between refreshes by more than the schedule accounts for, the outflow enters Sell #3 at the next refresh. Worth watching alongside them, because it is where those coins actually go: the Arbitrum Foundation operating wallet has sent exactly 10.0M ARB to an exchange deposit relay every month without a break, four times inside this window — on Jun 30, Jul 31, Aug 31 and Sep 8 2026. It is not a supply row, because both ends of that hop already sit inside the counted float, but it is the clearest read on what the grant is for.
It is also worth saying what the enumeration proves about announcements. Across the trailing year the Arbitrum DAO treasury made exactly one ARB payment. So the DeFi Renaissance Incentive Program, whose first season opened in early Sep 2026 with a budget of up to 24M ARB, releases nothing new — its 80M ARB left the treasury on Jun 23 2025, a year before this window, and already sits inside the float. An announcement is not a transfer.
How ARB compares to other capped governance tokens
ARB belongs to the layer-2 governance-token class: a hard, fixed ceiling, no issuance running, no fee burn, and a treasury large enough to be the real supply mechanism. That is a different animal from an uncapped continuous-emission layer-1, where a staking-linked emission of a few percent a year is the whole story and the schedule is written in protocol code. Arbitrum mints nothing at all — and still reads higher than most of those chains, because a cap constrains total supply while a vesting calendar and a treasury vote both move tradable float, and those are different quantities.
The sharper contrast is with the other big rollup governance tokens. Several of them carry a dated, capped inflation permission of the same shape as ARB's — a governance-controlled mint of roughly 2% a year that has never been used — so on the issuance axis this class is genuinely tight. Where they differ is the treasury. Arbitrum's DAO holds 2,556.1M ARB, more than a third of the circulating base, and deploys it by vote rather than by schedule, which means the largest single source of future ARB supply is a decision nobody has made yet. A holder reading a hard cap as protection is reading the wrong number.
The final comparison is to exchange tokens that run quarterly buybacks and burns. Those offset issuance with a demand-linked removal that scales with usage, and their readings can go genuinely negative. Arbitrum has no equivalent, and cannot easily build one, because gas is paid in ETH and the DAO's revenue arrives in ETH and stablecoins — the token is not in the fee path at all. Arbitrum DAO income for the first half of 2026 was about $6.19M against a float worth roughly a billion dollars, so even a full revenue-to-buyback conversion would not move this reading. That is the gap between having a treasury and having a sink.
What to watch in the next 90 days
First, the unlock calendar, which is the bulk of the number: tranches of about 92.6M ARB land on Sep 16 2026, Oct 16 2026 and Nov 16 2026, and the final one arrives Mar 16 2027 — after which Sell #2 goes to zero permanently and this page's forward reading roughly halves. Second, the Arbitrum DAO treasury balance at 2,556.1M ARB, the only pot on this token with a spender rather than a schedule; the annual Arbitrum Foundation grant has just fired, so the next comparable request is not due until roughly mid-2027, and anything sooner is a new release. Third, the mint permission: the contract's counter has read Mar 15 2024 since it was set, and any Arbitrum governance proposal to call mint would put up to 200M ARB of real issuance into Sell #1 for the first time in the token's life. Fourth, the Timeboost proceeds split, whose constitutional vote opened Aug 20 2026 and routes 97% of priority fees to the DAO treasury: it is an ETH flow today, and the thing to watch is any successor proposal that points sequencer revenue at ARB itself. Fifth, ARB staking, which is approved but not live — a launch would not change this reading, because staked ARB stays float, but a lock-up with a real cliff would.
Summary
The MrNasdog Pressure Framework reads ARB at +6.40% over the trailing 90 days and +4.81% projected forward: supply growing, projected to keep growing. The mechanism is not inflation but release — Arbitrum minted its entire 10,000M ARB once at launch and has issued nothing since, burned nothing, and has no fee burn or buyback that could remove anything, while a four-year unlock calendar pushes 92.6M ARB a month into the float and the DAO treasury sent 230.0M ARB to the Arbitrum Foundation in a single vote. The key risk is the treasury, not the calendar: the calendar ends on Mar 16 2027 and is fully known, while 2,556.1M ARB sits outside the float waiting on votes that have not happened. The genuine comfort is the ceiling — no vote is needed to stop ARB being minted, only to start it, and in three and a half years that vote has never been called.
MrNasdog Pressure Framework analysis of ARB, Metric 1 — Inflation. Data + explanation only. Not financial advice. Updated Sep 10 2026.
Top comments (0)