Arbitrum’s governance token jumped 135% over the course of September 2026, dwarfing Ethereum’s 8% gain over the same stretch, as a brokerage-run Layer-2 network built for tokenized stocks turned into the sector’s biggest growth story. The move pushed the combined market cap of major Ethereum Layer-2 tokens up roughly 65% from their 2026 low to about $11.5 billion by September 22, according to data from CryptoRank reported by CryptoTimes. At the center of it: Robinhood Chain, a network most crypto traders had barely noticed three months earlier.
The rally wasn’t a clean story, though. The same week ARB and STRK posted double-digit daily gains, a lending protocol on Starknet lost $3.5 million to a manipulated price feed, and a smaller Ethereum scaling network called Silicon quietly began shutting down, leaving close to $10 million stuck behind a withdrawal deadline. Together, the two threads describe where Ethereum’s Layer-2 scaling market actually stands heading into October 2026: a handful of well-funded chains pulling ahead on real financial activity, and a long tail of smaller networks running out of runway.
What’s driving the Ethereum Layer-2 rally this week
Ether itself had a comparatively quiet September. The token closed September 18 at $2,611.34, up 6.8% on the day, part of an 8% gain for the month according to 247wallst’s tracking of the period. Layer-2 tokens moved on a different scale entirely. ARB gained roughly 17% on September 18 alone and 26% over a trailing 24-hour window, touching about $0.2212. STRK, Starknet’s token, rose 18% the same day, hitting its highest level since June 19, 2026.
That single day set the tone for the month. By September 22, the combined market capitalization of the largest Layer-2 tokens, including Arbitrum, Optimism, Starknet, zkSync, and Mantle, had climbed to roughly $11.5 billion, a 65% recovery from where the group bottomed earlier in 2026. CryptoRank’s figures, cited by CryptoTimes, put the sector within reach of where it started the year, a notable reversal after a rough first half for L2 valuations.
Separately, L2Beat’s Total Value Secured metric, which tracks assets locked across Ethereum’s scaling systems through canonical bridges and related contracts, hit $34.34 billion on September 30, 2026. That figure sits apart from the token-price rally and reflects a steadier, longer-running accumulation of deposited value rather than a single month’s price action.
Arbitrum’s 135% run: inside the Robinhood Chain effect
Arbitrum’s token move stands out even against that broader L2 rally. ARB rose 135% over the month ending September 24, according to 247wallst, while Optimism’s OP token rose about 22% over a comparable window, a strong result in its own right but far behind ARB’s climb. The single catalyst analysts kept returning to was Robinhood Chain, a network the brokerage launched on its own infrastructure but built using Arbitrum’s technology.
The connection runs through revenue, not direct token demand. Chainstack reported that a $1.92 million revenue day on Robinhood Chain on September 1 helped send ARB about 30% higher that same day. That’s a striking ratio: a chain generating under $2 million in daily revenue moving a multi-billion-dollar token by nearly a third in 24 hours. It says less about Robinhood Chain’s absolute scale and more about how thin trading conditions and sentiment can amplify a single piece of good news in the Layer-2 token market.
By late September, CryptoTicker reported that Robinhood Chain held about $1.02 billion in its contracts, more than the amount deposited on Polygon. That’s a fast ramp for a network whose public mainnet only went live on July 1, 2026. Separately, Yahoo Finance reported the chain had attracted roughly $146 million in tradeable tokenized stocks before Robinhood’s initial free-gas promotion expired on September 29.
What Robinhood Chain actually is (and isn’t)
Robinhood Chain is an Ethereum Layer-2 network built on the Arbitrum Orbit stack, the same toolkit Arbitrum licenses to third parties for building their own application-specific chains. L2Beat classifies it as an Arbitrum Orbit Layer 2 operated by Robinhood, focused on tokenized real-world assets such as stocks and ETFs and on-chain financial services including 24/7 trading, lending, and borrowing. The chain settles its security back to Ethereum and charges transaction fees in ETH, not in ARB or any Robinhood-issued token.
That last detail matters more than it might seem. Robinhood Chain has no native token of its own. Robinhood had already sold more than 200 tokenized U.S. stocks and ETFs to European customers back in 2025. The new chain gives that business a dedicated settlement venue instead of routing everything through Arbitrum’s main network. The purpose, in other words, is tokenized securities infrastructure for a regulated brokerage, not a new speculative asset.
Built on Arbitrum Orbit, not a new token
Because Robinhood Chain has no token, its effect on ARB’s price runs entirely through a revenue-sharing mechanism rather than direct buy pressure. That distinction gets lost in a lot of market commentary, where “Robinhood Chain is growing” and “ARB should rise” get treated as the same statement. They aren’t. One measures usage of a brokerage product. The other measures demand for Arbitrum’s governance and staking token. The two are connected, but loosely, and the strength of that connection is exactly what analysts are now trying to pin down.
The free-gas promotion ends
Robinhood’s subsidized-gas period for Robinhood Chain ran out on September 29, 2026. Up to that point, the brokerage had been covering transaction costs for users trading tokenized stocks on the network, a common tactic for bootstrapping activity on a brand-new chain. Yahoo Finance framed the end of that subsidy as the real test: once wallets start paying their own ETH gas fees, does the $146 million in tokenized-stock activity hold up, or does it turn out to be incentive-driven volume that fades once the discount disappears? That answer wasn’t available as of October 1, but it’s the single most important number to watch over the next month.
The Arbitrum Expansion Program: how revenue flows back to ARB
The mechanism tying Robinhood Chain’s growth to ARB’s price is called the Arbitrum Expansion Program. Under that arrangement, any Orbit chain, including Robinhood Chain, returns 10% of its net revenue to the broader Arbitrum ecosystem: 8% goes to the Arbitrum DAO treasury and 2% funds a developer guild. The DAO treasury allocation is the part markets price in, since treasury inflows can eventually support buybacks, grants, or other value-accrual mechanisms for ARB holders.
A simplified version of that split looks like this:
orbit_chain_net_revenue = 1_920_000 # Robinhood Chain, Sept 1, 2026 (reported)
dao_treasury_share = orbit_chain_net_revenue * 0.08 # = $153,600
developer_guild_share = orbit_chain_net_revenue * 0.02 # = $38,400
total_to_arbitrum_ecosystem = dao_treasury_share + developer_guild_share # = $192,000 (10%)Even a single high-revenue day only routes a few hundred thousand dollars to Arbitrum’s treasury under this formula. That ARB still moved 30% on the news shows how much of the rally has been driven by narrative and expectations about future Orbit-chain adoption, rather than revenue actually landing in Arbitrum’s accounts that day. Arbitrum’s broader protocol upgraded in parallel: the ArbOS 61 “Elara” release activated on August 20, 2026, adding compliance tooling for Orbit chains and expanding smart contract capacity, groundwork that made it easier for a regulated brokerage like Robinhood to build on the stack in the first place.
That upgrade landed alongside routine token mechanics that could have worked against the rally. Roughly 92.65 million ARB, about 2% of circulating supply, unlocked on September 16, 2026. In a market already nervous about L2 token supply overhangs, that unlock could easily have capped the rally. It didn’t, which is itself a signal of how much buying pressure the Robinhood Chain narrative generated that month.
Layer-2 token performance in September 2026
The table below lines up the month’s biggest moves against Ether’s own performance, using the figures reported by 247wallst and CryptoTimes through September 24.
| Asset | Single-day move (Sept 18, 2026) | Monthly move (through late Sept 2026) | Primary driver cited |
|---|---|---|---|
| Arbitrum (ARB) | +17% (26% over 24 hrs, to ~$0.2212) | +135% over the month | Robinhood Chain revenue and Orbit adoption narrative |
| Starknet (STRK) | +18%, highest since June 19, 2026 | Part of sector-wide 65% climb from 2026 low | Sector rally; same-day Nostra exploit was a separate, negative event |
| Optimism (OP) | Not separately broken out | +22% over the month | Superchain interoperability upgrade progress |
| Ether (ETH) | +6.8%, closed at $2,611.34 | +8% over the month | Broad market conditions, not L2-specific |
| Major L2 sector (combined) | N/A | +65% from 2026 low to ~$11.5B market cap | Arbitrum, Optimism, Starknet, zkSync, Mantle combined |
The gap between Arbitrum’s 135% monthly gain and Ethereum’s 8% is the headline number here, and it’s a reminder that Layer-2 tokens trade with far more volatility than the base-layer asset they’re meant to scale. A single piece of adoption news on one Orbit chain moved ARB by a magnitude Ethereum itself rarely sees outside of major protocol announcements.
Optimism’s Superchain interoperability upgrade goes live
Arbitrum wasn’t the only ecosystem shipping real infrastructure that month. Optimism’s governance completed a vote on September 16, 2026, clearing the way for an upgrade meant to enable Superchain interoperability, the long-promised ability for chains in the Optimism ecosystem, including OP Mainnet, Base, and other OP Stack chains, to pass messages and assets between each other more directly. Test deployments went live across OP Sepolia, Ink Sepolia, Soneium Minato, and Unichain Sepolia starting September 17, with a contract upgrade applied the same day.
Mainnet execution was targeted for September 24, contingent on a healthy seven-day testnet soak period, the standard precaution before pushing consensus-level changes to a network holding billions of dollars in user funds. OP’s 22% monthly gain tracked alongside this rollout, though it trailed ARB’s move by a wide margin, a reflection of how much more aggressively the market rewarded Arbitrum’s financial-application narrative over Optimism’s infrastructure narrative that particular month.
Ethereum’s Glamsterdam upgrade heads to testnet
Underneath all the L2-specific news, Ethereum’s base layer kept moving on its own roadmap. The Glamsterdam upgrade, which follows the earlier Fusaka upgrade, is scheduled for its first public testnet activation on Sepolia at epoch 353,024, slot 11,296,768, which the Ethereum Foundation’s September 17 blog post pinned to October 6, 2026, at 13:53:36 UTC. A second test fork on the Hoodi testnet was set to follow on October 27.
Glamsterdam’s centerpiece is enshrined proposer-builder separation, known as ePBS, alongside block-level access lists and revised gas pricing meant to better reflect the real cost of execution and state growth. According to Ethereum’s own roadmap documentation, ePBS could expand the network’s block-propagation window from roughly 2 seconds to about 9 seconds by separating the deadlines validators use to attest to a consensus block versus an execution payload. A longer propagation window, in theory, lets Ethereum support more throughput and more blob data, the resource Layer-2 rollups depend on to post their transaction data cheaply back to the base chain.
Glamsterdam remains a Q4 2026 target with no confirmed mainnet date yet. For Layer-2 operators like Arbitrum and Optimism, though, the upgrade is directly relevant: any change to blob capacity or gas pricing flows straight through to how much L2 users pay per transaction, which in turn affects whether chains like Robinhood Chain can keep offering, or stop subsidizing, cheap trades.
The other side of the ledger: Silicon Network’s shutdown
While Arbitrum and Optimism posted their biggest gains of the year, a smaller Ethereum Layer-2 called Silicon Network was heading in the opposite direction. Silicon stopped accepting new bridge deposits on September 2, 2026, and entered a formal shutdown process. Users have until December 31, 2026, at 12:00 UTC+9 to withdraw their remaining assets before the network and its block explorer go permanently offline.
Roughly $9.75 million was still sitting on the network when the shutdown began, commonly rounded in coverage to nearly $10 million. Reported balances broke down to about $2.66 million in USDC, $2.54 million in WBTC, $2.08 million in ETH, and $1.85 million in USDT, according to figures reported by Blockonomi and CryptoTicker. Because Silicon is non-custodial, assets left behind after the December 31 deadline may not be recoverable at all, and any tokens issued natively on Silicon rather than bridged in from Ethereum face a separate risk: shrinking on-chain liquidity as users rush for the exits, which could make those tokens hard to convert even before the deadline hits.
Silicon’s operators have not published a detailed account of what went wrong financially, so it’s safer to describe this as a network winding down amid insufficient usage rather than attribute it to one documented cause. Still, the timing is instructive. The same month one Layer-2 attracted over a billion dollars by partnering with a household-name brokerage, another quietly ran out of road, with ten million dollars of user funds now subject to a hard countdown clock. For a deeper look at how fragile cross-chain infrastructure can be even without an outright shutdown, see our coverage of the Kelp DAO lawsuit against LayerZero following its $292 million rsETH exploit.
September’s exploit tally: Nostra and the $766 million month
The Layer-2 rally also played out against a brutal month for crypto security broadly. Blockchain security firm CertiK estimated that platforms and users lost approximately $766.4 million to exploits and phishing across all of September 2026, the highest monthly total and highest incident count recorded all year. Two incidents accounted for more than 92% of that figure on their own, a reminder of how concentrated crypto’s loss events tend to be, a pattern we also examined in our report on the Liquid Network’s $320 million Bitcoin-linked hack.
One of September’s incidents touched the Layer-2 story directly. On September 18, the exact day ARB and STRK posted their sharpest gains, Nostra, a lending protocol built on Starknet, lost about $3.5 million to a manipulated price feed. Starknet’s core network and sequencer weren’t affected. This was an application-layer failure in how Nostra priced collateral, not a flaw in Starknet itself. We covered the mechanics of that exploit, including the role of Pragma’s oracle feeds, in our prior analysis of the Nostra incident. What’s worth noting here is the contrast: STRK rallied 18% the same day a protocol on its network got drained, an unusual divergence between token-market momentum and the underlying security conditions of the applications actually running on that chain.
Testing and auditing practices for the kind of smart contracts behind these exploits have become their own discipline. Teams increasingly rely on tools like Foundry to simulate attack paths before deployment, the same approach we walked through in our guide to cross-chain bridge exploit testing and our breakdown of ERC-4626 inflation attack testing. Oracle manipulation, the category Nostra fell into, remains one of the harder bug classes to catch in pre-deployment testing because it depends on live market conditions rather than a static code flaw.
Competitive landscape: Arbitrum, Optimism, Starknet, zkSync, Base, and Polygon
Ethereum’s Layer-2 field has consolidated around a handful of ecosystems that each took a different bet on what would drive adoption. The table below summarizes where each stood as September 2026 closed out.
| Network | Technology | Native token | Signature September 2026 development |
|---|---|---|---|
| Arbitrum | Optimistic rollup; Orbit stack for app-chains | ARB | Robinhood Chain (Orbit-based) passes $1.02B in contract value; ARB up 135% for the month |
| Optimism (OP Mainnet) | Optimistic rollup; OP Stack; Superchain framework | OP | Superchain interoperability upgrade approved Sept 16, targeted for mainnet Sept 24 |
| Starknet | Validity (ZK) rollup; Cairo VM | STRK | Token up 18% in a day, even as Nostra, an app on the network, lost $3.5M to an oracle exploit |
| zkSync Era | Validity (ZK) rollup; zkEVM | ZK | Counted among the five tokens driving the sector’s 65% rally to $11.5B combined cap |
| Base | Optimistic rollup; OP Stack; operated by Coinbase | No native token | Part of the OP Stack ecosystem affected by the Superchain interoperability rollout |
| Robinhood Chain | Arbitrum Orbit app-chain | No native token; fees paid in ETH | Mainnet live since July 1, 2026; ~$1.02B in contracts; free-gas promo ended Sept 29 |
The pattern that emerges is less rollup versus rollup and more a question of which ecosystem landed a real-world financial use case first. Arbitrum’s answer was tokenized securities through a licensed brokerage. Optimism’s answer, at least this month, was infrastructure: making its own network of chains talk to each other more smoothly. Starknet’s zero-knowledge architecture is technically differentiated from both, but that didn’t stop an oracle bug on one of its applications from undercutting confidence within the same 24-hour window the token rallied.
Historical context: from 2021 rollup hype to 2026 consolidation
Ethereum’s Layer-2 category started as a scaling fix, not a product category in its own right. Arbitrum and Optimism both launched mainnet in 2021, built to solve one problem: Ethereum’s base layer was too expensive for everyday transactions. Starknet and zkSync followed with a different approach, using zero-knowledge proofs instead of optimistic assumptions to validate transactions faster and, in theory, more cheaply.
By 2023 and 2024, the category had exploded into dozens of competing chains, many nearly identical and differentiated mostly by incentive programs and airdrop farming. That proliferation is what produced casualties like Silicon Network: chains launched into a crowded field without durable fee revenue or a distinct enough use case to hold users once incentives ran dry.
September 2026 shows a market moving past that phase. Arbitrum’s Orbit framework turned the build-your-own-chain pitch from a developer tool into a revenue-sharing product that attracted a regulated brokerage. Optimism’s Superchain vision, years in the making, is finally shipping the interoperability layer meant to make its multi-chain approach coherent rather than fragmented. L2Beat’s $34.34 billion Total Value Secured figure, tracked since 2021, shows the category’s overall size holding steady even as chains like Silicon fall away. Consolidation, not proliferation, defines the current cycle.
Market impact: what this means for ETH, DeFi, and institutional adoption
For Ether holders, the Layer-2 rally is a mixed signal. A thriving L2 ecosystem is Ethereum’s scaling thesis playing out as designed: more transactions, more blob usage, more fee revenue eventually flowing back to the base chain. But the fact that L2 tokens gained 65% while ETH gained 8% highlights a persistent problem for Ethereum’s own valuation story. Much of the economic activity Ethereum enables ends up accruing value to L2 tokens instead of ETH itself.
For DeFi protocols on these chains, the month was a reminder to treat token-price momentum and security posture as separate questions. Nostra’s $3.5 million loss the same day STRK rallied 18% is the clearest illustration available: a token can rise sharply on sector enthusiasm while an application on that same network stays exposed to oracle manipulation, a risk that has caused DeFi losses for years.
For institutional adoption, Robinhood Chain is the more consequential development. A publicly traded brokerage choosing to build its own Layer-2 rather than simply issuing tokenized assets on an existing chain signals that regulated financial firms want dedicated infrastructure, with their own compliance tooling and control over the network, rather than sharing a general-purpose chain with speculative DeFi activity. If that pattern holds, expect more brokerages, banks, or asset managers to follow Robinhood’s playbook: license an existing rollup stack, operate a branded chain, and route a share of revenue back to the underlying ecosystem.
Risks and red flags analysts are watching
A few open questions will determine whether September’s rally holds or reverses. First, Robinhood Chain’s activity was heavily subsidized through September 29, so whether tokenized-stock trading volume survives paying real gas fees is the clearest test of durable demand versus incentive-driven usage. Second, the September 16 ARB unlock of 92.65 million tokens, about 2% of supply, adds ongoing sell pressure that a slowdown in Robinhood Chain’s growth could expose. Third, Optimism’s Superchain interoperability upgrade still needed to clear its mainnet rollout as of late September, and any delay or bug discovered during that process could undercut OP’s momentum just as quickly as Nostra’s exploit dented confidence in Starknet applications.
There’s also a structural risk that gets less attention: the Silicon Network pattern could repeat elsewhere. A handful of large, well-capitalized L2s are pulling in most of the new activity and revenue, while smaller chains with thinner treasuries face the same slow bleed that left Silicon shutting down with $9.75 million trapped behind a withdrawal window. Anyone holding assets on a lesser-known L2 should treat that shutdown notice as a prompt to check the chain’s financial health and bridge mechanics, not assume Ethereum’s base-layer security automatically extends to every network built on top of it.
Predictions: where Layer-2 tokens go from here
- Robinhood Chain’s growth will slow, not collapse, once gas subsidies end. Expect a meaningful drop in daily transaction counts after September 29 as casual users fall away, but the $1.02 billion already deposited in contracts suggests enough committed capital to keep the chain relevant into 2027.
- More Orbit-chain announcements from regulated financial firms are likely within six to twelve months. Robinhood’s approach, a dedicated Arbitrum Orbit chain with revenue-sharing back to the DAO, is a template other brokerages and asset managers can copy quickly given Arbitrum’s existing compliance tooling from the ArbOS 61 upgrade.
- Optimism’s Superchain interoperability rollout will become the comparison point for cross-chain UX through early 2027. If the mainnet execution targeted for September 24 lands cleanly, expect Arbitrum and other ecosystems to accelerate their own interoperability roadmaps in response.
- Smaller, undercapitalized L2s will keep shutting down. Silicon Network’s wind-down is unlikely to be the last of 2026. Chains without a distinct financial use case or a revenue-sharing partner like Robinhood face the same slow attrition.
- Ethereum’s Glamsterdam upgrade, once it reaches mainnet, will matter more for L2 fee economics than for ETH’s own price. A wider block-propagation window and more blob capacity primarily benefit the rollups building on top of Ethereum, Arbitrum and Optimism chief among them, rather than driving direct demand for ETH itself.
Frequently asked questions
Why did Arbitrum’s ARB token rise 135% in September 2026?
The main driver was Robinhood Chain, a Layer-2 network Robinhood built on Arbitrum’s Orbit stack for tokenized stocks and ETFs. A reported $1.92 million revenue day on the chain on September 1 helped push ARB about 30% higher that day alone, part of a broader 135% gain for the month, according to 247wallst and Chainstack.
Does Robinhood Chain have its own cryptocurrency token?
No. Robinhood Chain is an Arbitrum Orbit Layer-2 that settles to Ethereum and charges transaction fees in ETH. It has no native token of its own, which is why its relationship to ARB’s price runs through revenue-sharing under the Arbitrum Expansion Program rather than direct token demand.
What is the Arbitrum Expansion Program?
It’s the revenue-sharing arrangement under which Orbit chains, including Robinhood Chain, return 10% of their net revenue to the Arbitrum ecosystem: 8% to the Arbitrum DAO treasury and 2% to a developer guild.
What happened to Silicon Network?
Silicon Network, an Ethereum Layer-2, stopped accepting new bridge deposits on September 2, 2026, and began a shutdown process. Users have until December 31, 2026, at 12:00 UTC+9 to withdraw roughly $9.75 million in remaining assets before the network and its block explorer go offline permanently.
Was the Nostra exploit connected to the Layer-2 token rally?
Only by timing. Nostra, a lending protocol on Starknet, lost about $3.5 million to a manipulated price feed on September 18, 2026, the same day Starknet’s STRK token rallied 18%. The exploit affected the Nostra application, not Starknet’s core network or sequencer.
What is Ethereum’s Glamsterdam upgrade and when does it launch?
Glamsterdam is Ethereum’s next major protocol upgrade after Fusaka, centered on enshrined proposer-builder separation (ePBS), block-level access lists, and revised gas pricing. Its first public testnet activation is scheduled for October 6, 2026, on Sepolia, with a Q4 2026 target for mainnet, though no confirmed mainnet date has been set.
How big is the Ethereum Layer-2 sector right now?
By two separate measures: the combined market cap of major L2 tokens, Arbitrum, Optimism, Starknet, zkSync, and Mantle, reached roughly $11.5 billion by September 22, 2026, per CryptoRank. Separately, L2Beat’s Total Value Secured metric, which tracks deposited assets across Ethereum’s scaling systems, hit $34.34 billion on September 30, 2026.
Should I worry about my funds on a smaller Layer-2 network?
Silicon Network’s shutdown is a useful reminder that Ethereum’s base-layer security guarantees don’t automatically make every Layer-2 financially sustainable. It’s worth periodically checking whether a smaller L2 you use has adequate funding, active development, and a clear bridge-withdrawal process, rather than assuming it will operate indefinitely.
