Arbitrum’s economic engine is ETH, not ARB

Arbitrum runs a very “Ethereum-aligned” playbook. Users pay transaction fees in Ether, and the protocol’s recurring cash flows are denominated in ETH. ARB sits off to the side as governance weight. That split matters because it makes ARB’s value story mostly political, not mechanical. The chain can be wildly useful and still deliver weak token value accrual if governance does not translate into durable, defensible revenue rights. For a contrasting L2 token model, compare it with POL tokenomics.

Arbitrum’s fee plumbing is also unusually explicit about who gets what. The L1 posting cost component is routed to the sequencer operator (currently the Arbitrum Foundation) to “break even” over time, while most other fee components on Arbitrum One route to the DAO. This creates a real treasury inflow in ETH. It still does not create an automatic claim for ARB holders. It creates a governance-controlled pool of money.

What ARB actually does (and what it doesn’t)

ARB is a governance token native to Arbitrum One. It is used to vote on Arbitrum DAO proposals and to delegate voting power to delegates. If you want a refresher on the basics, see our tokenomics FAQ.

Two negative statements are the crux of ARB tokenomics.

First, ARB is not the gas token. The governance documentation is direct: ETH pays transaction fees, ARB does not.

Second, there is no protocol-level “fee-to-ARB” loop. Fees accrue in ETH to a treasury address the DAO can spend via governance. That can be powerful. It is still discretionary. A token holder is effectively betting on governance behavior, not on a hardcoded capture function.

This is why “burn” narratives struggle here. Even if the DAO someday chooses to use ETH revenue to buy ARB and burn it, that is a policy choice that can be reversed. It is not a structural sink like “fees must be paid in ARB and a fixed portion is burned.” Arbitrum did not build that.

Supply, allocations, unlocks, and the inflation switch

At launch, ARB’s initial supply cap was 10,000,000,000 tokens, with maximum inflation of 2% per year described in the airdrop allocation docs.

The onchain governance repository goes one step further and describes an explicit minting capability on the Arbitrum One token instance: up to 2% of total supply once per year. That is the 2% minting lever. It is not an emission schedule in the classic sense. It is governance-controlled optional dilution.

The other supply dynamic is unlock-driven, not inflation-driven. Team and investor allocations were locked with a one-year cliff and then monthly unlocks across the following three years. The governance docs pin the token generation event at March 16, 2023, with first team and investor unlocks March 16, 2024, then monthly unlocks through the remaining three years of the four-year lockup.

Initial distribution (post AIPs 1.1 and 1.2) is documented as follows.

On circulating supply, Arbitrum publishes a live integer feed. As of March 7, 2026, the circulating supply feed reports 5,939,074,958 ARB circulating.

Some dashboards label total supply and max supply as 10B, but governance and contract capabilities can define different long-run constraints. Treat this as a reminder to separate “what dashboards label” from “what the contracts and governance allow.”

Fees, treasury inflows, and why “burn narratives” miss the point

Arbitrum’s fee distribution rules split a transaction fee into four components and route them to specific recipients. The key facts are straightforward.

Users pay fees on Arbitrum One and Arbitrum Nova in Ether.

One portion, the L1 Base Fee, goes to the sequencer operator (currently the Arbitrum Foundation for both chains), and is designed to compensate the sequencer for L1 posting costs so it “breaks even” over time.

On Arbitrum One, the remainder of the transaction fee routes to the DAO. On Arbitrum Nova, most routes to the DAO, while a portion of the L2 base fee routes to third parties running critical infrastructure. Specifically, Nova’s L2 base fee is split 80% to the DAO and 20% to parties running Nova validators and the DAC.

This is the real “value capture” surface in Arbitrum today: ongoing ETH flows into a DAO-controlled treasury timelock on Arbitrum One. The DAO can spend from that pool via treasury governance proposals.

Two implications follow.

ARB does not need burns to look scarce. It needs sustainable net inflows to something the token can credibly control. Arbitrum has that, in ETH, through fee routing to the DAO. But the token still has no automatic right to those inflows. The system is “governance capture,” not “token capture.”

Net issuance still dominates optics. ARB has large unlock-driven supply expansion through March 16, 2027 for team and investors, and a Foundation unlock schedule running four years from April 17, 2023. Any hypothetical buyback-and-burn program would have to outrun (1) those unlocks and (2) the governance-controlled ability to mint up to 2% annually. That is a high bar unless fee generation is strong and persistent.

Arbitrum also built a separate, explicit capture mechanism for the broader ecosystem via Orbit-style chain deployments and the Arbitrum Expansion Program (AEP). The AEP structure is described as requiring 10% of a chain’s profit to be paid back, with 8% to the ArbitrumDAO treasury and 2% to an Arbitrum Protocol Developer guild.

The AEP fee calculation formalizes this as 10% of Net Protocol Revenue, and defines it in terms of fee components. It also states fees are deterministic and that the sequencer cannot tweak them in real time.

Governance control surface: who can change what

Arbitrum governance is not “a token and a forum.” It is a fairly heavyweight onchain system with defined delays and a Security Council circuit breaker. For a framework to map these levers, see our design components.

The governance constitution defines a 12-member Security Council with authority to execute Emergency Actions with 9-of-12 approval and no delay, and to execute certain Non-Emergency Actions with 9-of-12 approval while still passing through later delay phases. It also makes the Security Council’s scope modifiable or removable by Constitutional AIP.

The governance contracts are documented publicly, including the ARB token contract address on Arbitrum One (0x912CE59144191C1204E64559FE8253a0e49E6548) and the core governor and treasury governor addresses.

Quorum and participation are where the token’s “governance premium” either becomes real or evaporates.

Quorum thresholds are described as 3% for non-constitutional proposals and 4.5% for constitutional proposals, and only FOR and ABSTAIN count toward quorum.

The constitutional quorum itself was explicitly proposed to be reduced from 5% to 4.5%, reflecting a practical reality: circulating supply can rise while turnout stays flat, and then governance becomes fragile.

Arbitrum also documents a “vote exclusion” mechanism in its governance repository. Token holders can delegate to a designated exclude address so those tokens are removed from quorum calculations. The treasury is an example of tokens that should not vote, and therefore should not inflate quorum.

As a design choice, this is a double-edged sword. It makes governance workable at scale. It also concentrates practical power in the slice of supply that is actively delegated. That can be great when delegates are aligned and competent. It can be a problem when delegation is thin or captured.

Risk register: where ARB value accrual strains

The optimistic take on ARB is simple. Arbitrum is a high-usage scaling system. The DAO captures meaningful ETH fee surplus. ARB governs that surplus and controls upgrade and parameter rights. That can be enough.

The skeptical take is also simple. ARB is not required to use the chain. ARB does not automatically receive fees. Supply expands via unlocks, and governance can optionally mint up to 2% annually. That is a lot of dilution pressure for a token whose “cash flow” is a vote.

Top 3 risks

  1. Governance value capture stays discretionary, so usage does not translate into ARB demand. Trigger: sustained growth in transactions and DAO ETH fee inflows without a corresponding increase in ARB participation or demand. Mechanism: fees accrue in ETH to a DAO-controlled treasury, but token holders have no automatic claim, and governance may prioritize grants, ops, and incentives over buybacks or holder-directed value return. Who bears it: ARB holders, especially passive holders relying on “ecosystem growth” narratives. Measurable indicators: DAO treasury ETH inflow and spend patterns, frequency of proposals that create explicit ARB sinks, and the share of supply delegated to active voting.

  2. Governance capture via low delegation and quorum engineering. Trigger: declining participation rates relative to votable supply, paired with repeated quorum parameter changes or heavy reliance on vote exclusion. Mechanism: when delegated voting power is small, a motivated minority can dominate outcomes even if formal quorum thresholds look high, especially when quorum is defined as a percentage of “votable tokens” and large holders can move tokens in or out of quorum calculations through delegation choices. Who bears it: users and builders on the chain (governance decisions can alter fee recipients and upgrade paths), plus minority delegates. Measurable indicators: total delegated voting power, proposal vote totals vs thresholds, and the pace of constitutional changes to quorum settings.

  3. Net issuance overwhelms any scarcity narrative. Trigger: continued unlock-driven circulating supply expansion through the documented lockup schedules, combined with any future governance decision to mint up to 2% annually. Mechanism: marginal buyers need to absorb steady new liquid supply, while ARB’s utility remains governance-only and does not scale directly with transaction demand. Who bears it: ARB holders and DAO incentive recipients paid in ARB. Measurable indicators: circulating supply trend (including Arbitrum’s published circulating feed), remaining locked allocations, and any executed mint events under the annual 2% allowance.

Dominant risk: Governance-controlled ETH revenue does not automatically make ARB a productive asset, and the gap is structural.

The fee system gives Arbitrum something many L2 governance tokens never get: ongoing protocol revenue routed to a DAO address. On Arbitrum One, the non-L1-base-fee components route to the DAO’s treasury timelock in ETH. That is real economic bandwidth.

But ARB holders are not entitled to that ETH. They are empowered to decide what happens to it, subject to governance process, turnout realities, and the Security Council’s role in upgrades. This distinction is why ARB behaves less like an “equity proxy” and more like a political instrument. Political instruments can hold value. They are just harder to underwrite.

The most common attempt to close this gap is a buyback-and-burn program funded by protocol revenue. As a burn skeptic, I care less about whether a burn “works” in a single epoch and more about whether it can be sustained without starving the protocol of reinvestment. With Arbitrum, the math has two headwinds that do not go away:

1) Unlocks are a multi-year supply overhang. The governance docs describe a one-year cliff and then monthly unlocks for team and investor allocations after the March 16, 2023 token generation event, continuing across the remaining three years of a four-year lockup. Foundation tokens unlock linearly over four years from April 17, 2023. If your burn rate is not consistently above the market’s net new liquid supply, “deflation optics” are just optics.

2) There is an explicit annual dilution lever. The governance repository describes that the Arbitrum One token instance allows minting up to 2% once per year. Even if that lever is never used, its existence affects how confidently the market can price long-duration scarcity. If it is used, it resets the hurdle rate for any burn program. The relevant metric is net issuance, not whether some tokens were burned on a given month.

The honest bull case is still viable. A governance token can be valuable when governance controls something valuable, and when that control is credibly exercised. Arbitrum governance controls upgrades, fee recipients, and a treasury that receives ETH flows, plus it has an ecosystem-level capture mechanism via AEP that can expand the revenue base if Orbit-style chains scale.

The bear case is also coherent. The token is not required for usage, governance participation can be thin, and the DAO may rationally choose to reinvest ETH revenue into growth, security, and grants rather than engineering direct ARB value return. That decision can be good for the chain and bad for the token. It is not hypocrisy. It is a trade-off.

If you are building governance incentives or evaluating treasury policy, a short engagement with a tokenomics design team can be worthwhile. Tokenomics consulting is most useful here when it is grounded in verifiable control surfaces, budget constraints, and onchain participation data, not slogans about scarcity.



This article is part of our Tokenomics Deep Dive series.