Arbitrum (ARB): Blockchain Expert final project

Type and goal: Arbitrum is infrastructure. Its flagship chain, Arbitrum One, is an optimistic rollup: a Layer 2 that executes transactions off Ethereum and settles them on Ethereum. The goal is to make Ethereum-compatible transactions much cheaper and faster while keeping Ethereum as the final arbiter of correctness. The same software (the Nitro stack) also powers Arbitrum Nova (an AnyTrust chain that relies on a Data Availability Committee for cheaper data) and third-party Arbitrum chains (formerly branded Orbit).

Mechanism: a sequencer (today a single one, operated by Offchain Labs) orders transactions and gives users near-instant soft confirmations. It compresses transaction batches and posts them to Ethereum (as blobs since EIP-4844), so anyone can recompute the chain. Validators then post assertions (claims about the resulting state) to Ethereum with a bond. During a challenge window of about 6.4 days anyone can dispute a wrong assertion through an interactive fraud proof: the disagreement is bisected down to a single WebAssembly instruction that Ethereum executes itself. Users who suspect censorship can force-include transactions through the delayed inbox on Ethereum.

Network and token: Arbitrum relies on Ethereum for data availability and settlement. Gas on Arbitrum One is paid in ETH. The ARB token is an ERC-20 governance token: it votes in the Arbitrum DAO, which controls protocol upgrades, the treasury, and the Security Council elections.

Security mechanism (one paragraph): Arbitrum has no consensus of its own in the Proof of Stake sense. Safety comes from Ethereum plus fraud proofs under a 1-of-N honesty assumption: as long as a single honest validator watches the chain, a wrong state cannot be finalized. Since February 2025 the BoLD protocol makes validation permissionless (anyone can bond and dispute) and bounds how long an attacker can delay finality. Liveness and ordering depend on the centralized sequencer (with forced inclusion as the escape hatch), and a 12-member Security Council can perform emergency upgrades with 9 of 12 signatures.

Industry: Ethereum scaling infrastructure, serving DeFi, stablecoin payments, tokenized real-world assets (RWA), and companies that want their own chain. Stated audiences are developers (EVM plus Stylus for Rust, C and C++), DeFi users, and institutions.

Real users today: the bulk of activity on Arbitrum One is DeFi and trading (lending, perpetuals and spot DEXs) and stablecoin flows, which is why DefiLlama tracks Arbitrum mainly through DeFi TVL and stablecoin market cap. A second user group is institutions: the Foundation's 2025 transparency report highlights more than $800M of tokenized RWAs, and DefiLlama showed about $895M of RWA active AUM on 31 Oct 2026. A third group is businesses launching their own Arbitrum chains, the most prominent being Robinhood Chain (mainnet 1 July 2026).

Team: Arbitrum is developed by Offchain Labs, founded in 2018 by Ed Felten (Princeton professor of computer science and former Deputy US Chief Technology Officer in the White House), Steven Goldfeder (CEO, Princeton PhD) and Harry Kalodner (CTO, Princeton PhD). Arbitrum began as Princeton research and was published at USENIX Security 2018. Offchain Labs also maintains Prysm, an Ethereum consensus client, after acquiring Prysmatic Labs in 2022.

Central entities: the Arbitrum Foundation (a Cayman Islands foundation company) supports the ecosystem and executes DAO decisions; Offchain Labs is the main core developer and currently operates the sequencer; the Arbitrum DAO (ARB holders and their delegates) owns the treasury and decides upgrades.

Funding: three priced rounds. Seed of $3.7M in 2019 led by Pantera Capital; Series A of $20M in April 2021 led by Lightspeed Venture Partners; Series B of $120M on 31 Aug 2021 led by Lightspeed, with Polychain Capital, Ribbit Capital, Redpoint, Pantera and others, at a reported $1.2B valuation. Total: about $143.7M (some databases list the Series B as $100M and the total as $123.7M; cite the Offchain Labs announcement as the primary figure).

Other key players: the Security Council (12 elected signers with emergency powers); Entropy Advisors (DAO-funded strategy and treasury advisors, author of the October 2026 USDG incentive proposal); the Arbitrum Developer Guild (receives 2% of Expansion Program revenue); large delegates; anchor applications in DeFi; and institutional partners such as Robinhood, which built its own Arbitrum chain.

GitHub: the live panel below counts commits and distinct commit authors over the last 30 days in Offchain Labs' core repositories (nitro, nitro-contracts, stylus-sdk-rs), fetched from the GitHub API with the check time shown. Quote the number with that date in the paper.

Validators and nodes: a rollup has no validator set like a Proof of Stake chain, so Arbiscan does not show a validator count. The relevant facts are: one active sequencer (Offchain Labs); permissionless validation under BoLD (anyone can run a node and post or challenge assertions with a bond); and any number of full nodes that anyone can run. L2BEAT lists the proposer and challenger setup and is the right source to cite here.

Two usage metrics: (1) Stablecoin market cap on Arbitrum: stablecoins are real dollars parked on the chain for payments, trading and collateral, so they measure economic trust better than token-priced metrics. (2) Fees paid on Arbitrum One over 30 days: fees are what users actually pay for blockspace, and they feed the DAO's revenue. TVL is shown too, but it is inflated or deflated by token prices.

Claim: Arbitrum presents itself as a rollup secured by Ethereum, where users do not need to trust the operator. Data that supports it: BoLD made fraud proofs permissionless on Arbitrum One in February 2025, transaction data is posted to Ethereum, forced inclusion exists, and L2BEAT rates Arbitrum One Stage 1 and says it passes the walkaway test.

Data that complicates it: the Security Council can act instantly with 9 of 12 signatures, with no exit window for users on that path. In October 2026 it used this power twice: it paused new Stylus activations (2 Oct) and moved 8,425 ETH and about 74.56M ARB tied to the Binance exploit from a specific address into a frozen wallet (30 Oct). Ordering also depends on one sequencer. Verdict: the claim holds for the technical security of state transitions, but trust in a 12-member committee remains part of the model. The picture is more complex than the slogan.

The seven events are listed in the timeline section of this page (and in the Word export), each with date, category and source. They cover a launch plus funding round (Aug 2021), a network upgrade (Nitro, Aug 2022), token launch and exchange listings (Mar 2023), a governance crisis (AIP-1, Apr 2023), a security upgrade (BoLD, Feb 2025), a strategic partnership (Robinhood Chain, Jul 2026) and emergency security actions (Oct 2026).

Most formative: the March 2023 token launch and DAO handover, including the AIP-1 crisis that followed. It created the structure that defines Arbitrum today: a DAO that owns the treasury and sequencer revenue, a constitution with two proposal types, an elected Security Council, and a Foundation that must publish budgets and transparency reports. Every later strategic move (Timeboost revenue, the Expansion Program license, incentive programs) runs through that structure.

Market reaction: Robinhood Chain is a case where usage and token price diverged. Robinhood's first-month license payment was about $360K in July 2026, roughly 35% of that month's DAO income, yet ARB fell about 24% over 30 days to around $0.16 in late October 2026. What the price reflects is that DAO income does not flow to ARB holders automatically and that team and investor unlocks continue until March 2027. Whether this is an under-reaction depends on whether the DAO ever links its revenue to the token.

Pending event: the end of the team and investor unlock schedule in March 2027. Good scenario: monthly sell pressure of about 92.6M ARB disappears, and supply becomes predictable. Bad scenario: final unlocks are sold into a weak market, and without a value-accrual mechanism demand does not absorb them. A second pending item is reopening Stylus activations: a safe reopening restores a key differentiator; a long pause weakens the multi-language story.

Platforms: the official X account (@arbitrum), the Arbitrum Discord, the governance forum (forum.arbitrum.foundation), Snapshot and Tally for votes, a YouTube channel, and developer channels on GitHub. Follower and member counts change daily, so read them on the day you submit and cite the date (the course requires a check date for every number).

Support: extensive English documentation (docs.arbitrum.io for builders and users, docs.arbitrum.foundation for governance), Stylus quickstarts and SDKs, public weekly vote roundups on the forum, recorded governance calls, an open delegate program, and a free public testnet (Arbitrum Sepolia). Documentation is primarily English; community translations exist but are not the official track.

This must be your own participation; it cannot be answered from sources. Timely threads with real substance as of the research date: (1) the on-chain vote on 100M ARB for USDG incentives (29 Oct to 12 Nov 2026) on the forum and Tally; (2) the Security Council emergency action of 30 Oct 2026 (freezing funds tied to the Binance exploit); (3) the Stylus activation pause and its reopening plan. For each screenshot write: platform, date, the question or point you raised, and the response you got.

Write this from your own notes. Useful observable checks: on the governance forum, do Foundation or Offchain Labs staff reply in proposal threads? In Discord, are support questions answered by moderators or by other users? On X, are replies about ARB price or about upgrades and votes? Answer separately from 3.4, as the course requires.

Draft to personalize (replace the bracketed parts): "For my final project in the Blockchain Expert track at the Blockchain and Digital Currency College, I researched Arbitrum, Ethereum's largest optimistic rollup. Three takeaways: (1) security is a spectrum: BoLD made fraud proofs permissionless, yet a 12-member Security Council can still act in an emergency, as it did in October 2026; (2) usage and token value are different questions: institutions like Robinhood now build on Arbitrum technology, but DAO revenue does not flow to ARB holders automatically; (3) governance is real work: reading forum threads and votes taught me more than any dashboard. Taking part in [name the discussions] showed me [your insight]. Next: [your next step]. #Blockchain #Arbitrum #Ethereum #BDCC"

Image: use your own screenshot (for example your Tally vote or forum post) or a photo from your studies; do not reuse Arbitrum brand assets as if they were your own. If you publish, add the link and tag the college page.

Who decides: the Arbitrum DAO, meaning ARB holders voting directly or through delegates. Offchain Labs writes most protocol code and proposes upgrades, the Arbitrum Foundation executes and administers, and the Security Council handles emergencies. Chain upgrades, treasury spending and changes to the constitution all require a DAO vote.

Where: a hybrid. Discussion starts on the forum, followed by an off-chain Snapshot temperature check, then an on-chain vote on Tally through one of two Governor contracts. On-chain results execute automatically after timelocks.

Who and thresholds: any ARB holder can vote or delegate. Submitting an on-chain proposal requires 1,000,000 ARB of delegated voting power (lowered from 5M by AIP-1.2). Two proposal types: Constitutional (changes to the constitution, core contracts, chain software or the Security Council) and Non-Constitutional (funding and other decisions). Quorum was historically 5% of votable supply for Constitutional and 3% for Non-Constitutional; the Constitutional quorum was cut to 4.5% in 2025, and in 2026 the DAO adopted a quorum based on delegated voting power (DVP). A Constitutional proposal takes about 37 days end to end because it passes waiting periods on L2 and L1 plus the withdrawal delay.

Participation: the DVP quorum proposal itself documents that participation fell from around 8% to 4 to 5% of votable supply over the preceding year, close to the quorum line. A recent example: the October 2026 Snapshot vote on 100M ARB for USDG incentives drew about 199.5M ARB in favor. Note that the DAO treasury's own ARB cannot vote.

Concentration: analyses posted to the DAO forum report that the ten largest delegates hold roughly 40% of delegated voting power, and Curia's study found voting power heavily concentrated (Gini above 0.9). Hundreds of delegates receive delegation but rarely vote. Quote the latest Tally delegate table with its date for the current numbers.

Conflict and change: disputes are argued on the forum and decided by votes; the clearest precedent is AIP-1 (April 2023), where a failed off-chain vote forced the Foundation to restructure its request. The governance model itself changes only through a Constitutional AIP, as with the 2025 quorum reduction, the 2026 DVP quorum, and the 2026 extension of Security Council terms.

Decision: adopting Timeboost (activated on Arbitrum One and Nova in April 2025). Initiator: Offchain Labs, which designed it as a replacement for pure first-come-first-served ordering. Mechanism: a sealed-bid, second-price auction each minute sells an express lane with a short time advantage; bids are paid in ETH and the proceeds go to the DAO. Critics on the forum argued it favors sophisticated searchers and adds a centralized auction step. It passed as a Constitutional AIP through the full Snapshot then Tally process. Outcome: Timeboost returned more than $6M to the DAO in 2025, its first year. Fairness: procedurally fair (public discussion, two votes, timelocks), but the proposal came from the core developer that also runs the sequencer, which shows how agenda-setting power sits with Offchain Labs.

Suggested rating: 2 out of 5. For: real on-chain votes that bind upgrades and treasury, a successful community revolt (AIP-1), elected Security Council. Against: low participation (4 to 5% of votable supply), roughly 40% of delegated power in ten delegates, one sequencer operator, and a Security Council that acted unilaterally twice in October 2026. This rating is a defended opinion; adjust it if your own reading of recent votes differs.

Type and role: ARB is an ERC-20 governance token. It is not used for gas (gas is ETH) and is not staked to secure the chain. Its role is voting on upgrades, treasury and Security Council elections, and funding ecosystem programs from the DAO treasury.

Supply: 10,000,000,000 ARB minted at genesis (March 2023). There is no hard cap: the DAO may mint up to 2% per year by a Constitutional vote. No burn mechanism. Circulating supply is shown live below.

Initial distribution: DAO treasury 35.28%; Offchain Labs team and future team plus advisors 26.94%; investors 17.53%; user airdrop 11.62%; Arbitrum Foundation 7.5%; airdrop to DAOs building on Arbitrum 1.13%.

Unlocks and changes: team and investor tokens had a one-year cliff (March 2024) followed by 36 monthly releases of about 92.6M ARB, ending in March 2027. After the AIP-1 crisis, the Foundation's remaining 700M ARB was put under a four-year smart-contract lockup (AIP-1.1). The economics changed mainly on the revenue side (Timeboost from 2025, Expansion Program licenses) and through large treasury spending programs; the 2% inflation option has not been part of those changes.

Directly: it does not. There is no fee switch, burn, buyback or staking yield paid from protocol revenue to ARB holders. Indirectly: revenue (net sequencer fees from Arbitrum One and Nova, Timeboost auction proceeds, 8% of net revenue from Arbitrum chains settling outside One and Nova under the Expansion Program, and treasury yield) accumulates in the DAO treasury, which ARB holders control by vote. Scale: about $23.5M gross revenue in 2025 and $6.19M of income in H1 2026, with $125M of non-ARB treasury assets at the end of June 2026. Any future link to the token (buybacks, staking rewards) would require a DAO decision.

Short term the chain keeps running: users pay gas in ETH, the sequencer keeps ordering, batches keep posting to Ethereum, and BoLD validators (bonded in ETH) keep securing withdrawals. What breaks is governance: both Governor contracts count ARB votes, so no Constitutional upgrade or treasury payment could pass, and Security Council elections (also ARB-weighted) would stop. Control would drift back to the Security Council and the Foundation, and the treasury's ARB (its largest asset) would vanish.

Lesson: ARB is not needed for the product to work; it is needed for the project to be owned and upgraded by a community instead of a company. The token economics serve the decentralization vision, not the usage of the chain, which is exactly why usage growth does not automatically show up in the token price.

Entities: the Arbitrum Foundation is a Cayman Islands foundation company that supports the ecosystem and holds the ARBITRUM trademark; Offchain Labs is a privately held US company that develops the software and runs the sequencer; the DAO itself has no legal wrapper. Status: as of the research date we found no SEC or other enforcement action against Arbitrum, the Foundation or Offchain Labs, and ARB is listed on major regulated exchanges (including Coinbase). The offshore foundation plus US developer structure is common and has not been tested in court for Arbitrum.

Trade-offs: Arbitrum gains scalability by executing off Ethereum, and keeps security close to Ethereum's by posting all data there and using fraud proofs. It pays for this mainly in decentralization of operations (one sequencer, an emergency council) and in finality for withdrawals (about 6.4 days to Ethereum). Arbitrum Nova trades further: it keeps data with a committee to be cheaper.

Failure points: (1) sequencer downtime halts fast transactions (outages occurred in 2021, 2022 and on 15 Dec 2023); (2) Security Council collusion or key compromise (9 of 12 can upgrade instantly); (3) proof-system or VM bugs, which is why Stylus activations were paused in October 2026 and a guard was added to the one-step proof; (4) dependence on Ethereum for data and settlement, including its fees and censorship resistance.

Interfaces: the canonical Arbitrum bridge to Ethereum (deposits in minutes, withdrawals after the challenge period); Arbitrum chains that settle on Arbitrum One as L3s; Circle's CCTP for native USDC; and third-party messaging and liquidity bridges (Chainlink CCIP, LayerZero-based bridges, Across and others). ARB itself exists on both Ethereum and Arbitrum One.

Volumes: the authoritative figure is L2BEAT's Total Value Secured (canonically bridged, natively minted and externally bridged assets); a 2026 compilation of L2BEAT data put it at about $12.3B. Stablecoins on the chain are shown live below (DefiLlama). For transfer volumes over a recent period cite a Dune bridge dashboard with its date.

Uses and risks: most flows are users moving ETH and stablecoins from Ethereum into Arbitrum DeFi and back, institutions minting RWAs, and liquidity moving to and from other L2s. Risks: third-party bridges add their own validator or multisig trust; the canonical bridge inherits the rollup's own risks (Security Council, proof bugs); and Arbitrum One is exposed to whatever hacked funds flow in, as the 30 Oct 2026 freeze of assets tied to the Binance exploit showed.

Thinking question: interoperability is mostly an advantage, because Arbitrum's value proposition is being Ethereum's liquidity hub, and Arbitrum chains (like Robinhood Chain) extend its technology reach. It becomes a dependency where activity happens on chains that only license the technology: those chains bring license fees but can take users and liquidity away from Arbitrum One.

Competitors: Base (built by Coinbase on the OP Stack) and OP Mainnet (Optimism). Criterion 1, economic gravity: DeFi TVL and stablecoins show where capital chooses to live (live table below). Criterion 2, trust model: the L2BEAT stage and proof system determine how much users must trust operators. All three are Stage 1 with permissionless proofs (Arbitrum via BoLD; OP Mainnet and Base via OP Stack fault proofs), so the differences sit in proof design, council powers and distribution.

Advantages: deep DeFi liquidity, a mature proof system with bounded delay (BoLD), Stylus for non-Solidity developers, and new DAO revenue lines (Timeboost, licenses). Disadvantages: no built-in consumer funnel comparable to Coinbase for Base, a token with no direct value capture, and October 2026 security actions that highlighted operational risk.

Next: Arbitrum ships through numbered ArbOS upgrades approved by the DAO. In 2026 these were ArbOS 51 "Dia" (January, Ethereum Fusaka compatibility and gas pricing changes) and ArbOS 61 "Elara" (activated 20 Aug 2026, multidimensional gas pricing and a 96 KB Stylus contract size limit). Upcoming items: compatibility with Ethereum's Glamsterdam upgrade, a decision on reopening Stylus activations, continued growth of Arbitrum chains under the Expansion Program, and the long-discussed path to Stage 2 (restricting the Security Council to provable on-chain bugs).

Most critical: the Arbitrum chains business (Expansion Program). It is the only revenue line that scales with adoption beyond Arbitrum One's own fees, and Robinhood Chain already contributed about 35% of the DAO's July 2026 income. Its success depends on the security reputation of the stack, which ties it directly to the Stylus and Security Council questions above.

Suggested position (yours to defend): 3 out of 5, a small allocation of about 2 to 3% of a crypto venture portfolio. Reasoning: strong technology, real institutional adoption and growing DAO revenue, offset by a token with no direct value capture, a price that fell from an all-time high of about $2.39 (January 2024) to about $0.17 (October 2026), and continuing unlocks until March 2027.

Realistic failure in 3 years: large institutions follow the Robinhood model but choose other stacks or negotiate away license fees; activity on Arbitrum One migrates to app-specific chains and to Base; DAO revenue stays in single-digit millions while the treasury is spent on incentives; and a serious proof-system or VM bug forces a long Security Council intervention, damaging the security brand that the licensing business depends on.

What would change the view: upward, a DAO decision that ties revenue to ARB (buybacks or staking rewards), several more large institutions launching Arbitrum chains, or reaching Stage 2. Downward: a sustained drop in stablecoin supply on Arbitrum One, a quarter of falling license revenue, or another emergency action caused by a protocol bug.

Recommended path on Arbitrum Sepolia (free, no financial risk): (1) add Arbitrum Sepolia to your wallet (chain id 421614); (2) get Sepolia ETH from a faucet listed in the Arbitrum docs; (3) bridge it from Ethereum Sepolia to Arbitrum Sepolia with the official Arbitrum Bridge (testnet mode), or send a small transfer between two of your addresses; (4) open the transaction on the Sepolia Arbiscan explorer and copy its hash. Note how long the deposit took versus how long a withdrawal back to Sepolia would take: that gap is the challenge period from question B.1 in practice.

Interactive proving: Arbitrum never re-executes a whole block on Ethereum. When two parties disagree, they bisect the disputed execution until a single WebAssembly (WAVM) instruction remains, and only that one step is executed on L1 by the one-step prover. Execution runs natively (Nitro compiles a modified Geth core), while the same code compiled to WASM is used for proofs. This keeps L1 cost small and lets the same engine that runs the chain prove it.

BoLD (Bounded Liquidity Delay), live on Arbitrum One and Nova since February 2025: earlier Arbitrum disputes were one-against-one between allow-listed validators, so an attacker with many identities could delay finality by forcing sequential challenges. BoLD is all-against-all and permissionless: a single honest party can defend the correct state against any number of attackers, and the worst-case delay is bounded (about one extra challenge period) regardless of how many attackers join. Top-level assertions require a large ETH bond, which makes delay attacks expensive. OP Stack chains also have permissionless fault proofs, but BoLD's distinctive property is this bounded delay against many adversaries.

Stylus: a second virtual machine next to the EVM (live on mainnet since September 2024). Developers write contracts in Rust, C or C++ compiled to WebAssembly, and these contracts are fully interoperable with Solidity contracts. Compute- and memory-heavy logic (cryptography, on-chain games, verification) becomes much cheaper. ArbOS 61 raised the Stylus contract size limit to 96 KB. On 2 Oct 2026 the Security Council paused new Stylus activations on One and Nova over risks from hand-crafted WebAssembly programs; existing Stylus contracts keep running and Solidity is unaffected.

L2BEAT stage: Arbitrum One is Stage 1 and passes the walkaway test. What blocks Stage 2: the Security Council can act on issues that are not provable on-chain bugs, and its emergency path (9 of 12) takes effect immediately with no exit window. Regular upgrades give users only about 10 days to exit (8 days on L2 plus 3 days on L1, adjusted for the forced-transaction delay), while Stage 2 requires at least 30 days for upgrades unrelated to provable bugs.

Thinking question: the roughly 6.4-day wait (about 45,800 Ethereum blocks) is the challenge period: it must be long enough that an honest validator can get a dispute included on Ethereum even if an attacker censors or congests L1, or if validators need time to coordinate. With a one-hour window, an attacker who can censor Ethereum blocks or knock validators offline for an hour could finalize a false state and drain the bridge, so security would fall from one honest party to whoever controls a short window of L1 inclusion. Users bypass the wait with liquidity bridges and exchanges (for example Across or a centralized exchange): a liquidity provider pays them on Ethereum within minutes for a fee and waits out the challenge period itself.

Orbit: the permissionless framework (now marketed as "Arbitrum chains") for launching your own chain with the Nitro stack: rollup or AnyTrust data mode, custom gas token, custom permissions, settling either on Arbitrum One or Nova (as an L3) or on Ethereum or another chain. AEP: the license under which any chain using Arbitrum technology that settles somewhere other than Arbitrum One or Nova pays 10% of its net protocol revenue to the ecosystem: 8% to the DAO treasury and 2% to the Arbitrum Developer Guild. Chains settling on One or Nova pay nothing extra because they already pay fees to the parent chain.

Robinhood Chain: public testnet on 10 Feb 2026 (Consensus, Hong Kong); public mainnet on 1 July 2026 at Robinhood's event in London. It is a dedicated Arbitrum chain that settles to Ethereum, built for tokenized real-world assets: Stock Tokens of US equities and ETFs tradable 24/7 and usable in DeFi as collateral. Because it settles to Ethereum, it pays the AEP share: about $360K in July 2026, roughly 35% of that month's DAO income.

Thinking question: both. It is a technology win (a top US broker picked the Arbitrum stack and pays a license) and a signal that large brands want their own chain: control of sequencing, fees, compliance and branding. The DAO gains most directly (license revenue that scales with Robinhood's activity). Arbitrum One gains least and may lose activity and liquidity to the new chain. ARB holders gain only indirectly, through a richer treasury they govern, unless the DAO creates a mechanism that links revenue to the token.

What it was: days after the March 2023 airdrop, the Foundation posted AIP-1, a "ratification" of the governance documents and structure that also allocated 750M ARB (worth about $1B at the time) to the Foundation for grants and operations, plus a budget for legal and administrative costs.

The storm: delegates discovered the Foundation had already started using the tokens before any vote: 10M ARB had been sold for fiat to cover operating costs and 40M ARB had been lent to a market-making counterparty. The DAO was being asked to approve a decision already executed. The off-chain vote failed decisively (about 100M ARB against versus 16M for).

Response and resolution: the Foundation apologized publicly and split the request into AIP-1.1 (the remaining 700M ARB locked in a smart contract and released over four years, with defined budget categories and mandatory transparency reports) and AIP-1.2 (amending the governance documents, removing AIP-1 references, and cutting the proposal threshold from 5M to 1M ARB). Both passed their temperature checks with 98.13% and 99.15% support and were then confirmed on-chain.

Thinking question: the crisis showed that formal power was asymmetric (the Foundation held the keys and could move tokens) but that the DAO's legitimacy was decisive: an off-chain vote with no binding force still forced a full retreat. Long term it strengthened governance: lockups, budgets and transparency reports became norms (the Foundation still publishes them, for example the 2025 Transparency Report), and the lower proposal threshold opened the agenda to more delegates. It also left a lasting trust deficit that shows up whenever the Foundation asks for funding.

Members and elections: 12 members split into two cohorts of six; one cohort is elected every six months by ARB-weighted DAO votes after a nomination phase and a Foundation compliance check, with limits on how many members may come from one organization. In the March 2026 election six members were elected (reported winners include Michael Lewellen, Certora, bartek.eth and yoav.eth; cite the forum results post for the full list). In July 2026 the DAO extended member terms from one year to two, and the council removed one member and appointed Patrick Collins (Cyfrin) by 9-of-12 votes.

Powers: through emergency multisigs on Ethereum and Arbitrum, 9 of 12 members can execute any action immediately, including contract upgrades and parameter changes, with no DAO vote and no delay. Non-emergency actions by the council go through the same delays as DAO proposals. October 2026 examples: on 2 Oct it set the Stylus activation gas to the maximum, effectively pausing new activations; on 30 Oct it moved 8,425 ETH and about 74.56M ARB tied to the Binance exploit into a frozen wallet that only further council action or a DAO vote can release.

Constitutional versus Non-Constitutional: Constitutional AIPs change the constitution, core contracts, chain software or the Security Council; they need a higher quorum and pass through L2 waiting, the L2-to-L1 delay and L1 waiting (about 37 days), so users can exit before a change they reject takes effect. Non-Constitutional AIPs (grants, treasury, guidelines) need a lower quorum and execute faster. The split matters because it gives the strongest protection exactly to changes that can affect user funds.

Thinking question: it is a compromise, not a contradiction, as long as three conditions hold: the council is elected and removable by the DAO, its emergency actions are published and justified (both October 2026 actions were announced on the forum), and its scope is narrowed over time. The October freeze shows the cost: the same power that blocked stolen funds can freeze any address. It is right to limit it to provable on-chain bugs (the Stage 2 bar) once the proof system and Stylus have run for a sustained period without emergency fixes, and to remove it only when the protocol no longer needs upgradability for safety.

Revenue sources: (1) Arbitrum One and Nova net sequencer fees (users' gas minus the cost of posting to Ethereum); (2) Timeboost express-lane auctions, more than $6M in 2025, its first year; (3) AEP licensing, led by Robinhood Chain (about $360K in July 2026); (4) treasury management returns on stablecoins and tokenized treasuries, with $125M of non-ARB assets at the end of June 2026. Totals: about $23.5M gross revenue in 2025; $6.19M in H1 2026, with gross margin above 97%. The Foundation's unaudited report gives the H1 2026 total; ask for the per-line split in the report itself when citing exact shares.

Metrics versus price: the live chart below indexes ARB price, DeFi TVL and stablecoin supply on Arbitrum to 100 at the start of the last year, so their paths can be compared directly. The pattern documented across sources: stablecoins and RWAs on Arbitrum grew (the Foundation reported stablecoin supply up about 80% year on year in 2025 and more than $800M of RWAs), while ARB fell from its January 2024 high of about $2.39 to about $0.17 by late October 2026. Conclusion: adoption and token value have decoupled, consistent with a token that has no direct claim on revenue and with ongoing unlocks.