In 2020, when Vitalik Buterin proposed the "rollup-centric" scaling roadmap, few questioned its fundamental logic. Ethereum’s mainnet would remain "slow and expensive" for the long term, while Layer 2 (L2) solutions would serve as efficient, low-cost transaction layers, handling most user interactions. The mainnet would retreat to serve as the foundational infrastructure for security and settlement. This narrative has driven tens of billions of dollars in capital flows over the past five years and led to the emergence of more than 100 L2 public chains.
However, 2026 is shaping up to be a pivotal year that challenges this logic.
In February 2026, Vitalik Buterin publicly stated that Ethereum’s previously rollup-centric roadmap "no longer applies." Almost simultaneously, the Ethereum Foundation released the Strawmap roadmap covering 2026 to 2029, placing native L1 scaling—not just reliance on L2—at the core of its technical planning for the first time. According to the plan, the Glamsterdam upgrade is expected to activate on the mainnet around June or Q3 2026. Its key features—parallel transaction execution and encapsulated proposer-builder separation—will raise the mainnet gas limit from the current 60 million to 200 million, targeting a TPS of 10,000.
For the L2 ecosystem, which has relied on "Ethereum is too expensive" as its survival foundation for the past five years, this is nothing short of a structural shock.
The Technical Substance of the Glamsterdam Upgrade
The Glamsterdam upgrade is not just another technical iteration. It marks a critical strategic shift: from "L1 serving L2" to "L1 possessing independent competitiveness."
This upgrade carries several headline features across three core tracks—Scale, Improve UX, and Harden the L1. Its technical architecture involves two major changes:
First, block-level access lists and parallel execution. Currently, Ethereum L1 operates in a single-channel mode, with all transactions queued for sequential execution. The block-level access list mechanism will pre-label the accounts and storage slots accessed by each transaction in a block, allowing nodes to process non-conflicting transactions in parallel. According to Galaxy Research, this mechanism is a key engineering prerequisite for achieving the "Gigagas L1" goal—10,000 TPS at the base layer. While current L1 throughput is only about 15 to 30 TPS, the path toward parallel execution is clear and has already been validated in testnets.
Second, encapsulated proposer-builder separation. Currently, Ethereum block production relies on external relay networks. Encapsulated proposer-builder separation writes this process directly into the consensus layer, removing reliance on external relays and eliminating risks of centralized censorship. Public reports indicate that after the Glamsterdam upgrade, gas fees are expected to drop by approximately 78.6%.
It’s important to note that Glamsterdam is not an isolated event. Its follow-up upgrade, Hegotá, is already in the planning stages, aiming to further strengthen FOCIL anti-censorship capabilities and foundational technologies like Verkle trees. Overall, from 2026 to 2029, Ethereum is expected to iterate with hard forks roughly every six months, gradually approaching long-term goals like single-slot finality, native privacy, and post-quantum security.
In other words, Glamsterdam is only the beginning of a stepwise leap in L1 capabilities.
The Data Reality of the L2 Ecosystem
As the mainnet accelerates its catch-up, the L2 ecosystem presents a contradictory data landscape.
L2 TVL scale and user activity. As of February 2026, total value locked (TVL) in Ethereum L2s ranged from $38 billion to $43 billion. Overall L2 network throughput exceeded 300 TPS. L2s still handle about 95% to 99% of all Ethereum ecosystem transactions. However, according to L2BEAT data, as of April 27, 2026, total L2 TVL had fallen to around $34.21 billion, a 1.61% drop over seven days, reflecting recent market volatility’s suppressive effect on TVL.
The "zombie chain" phenomenon. As of February 2026, Base, Arbitrum, and Optimism collectively accounted for about 90% of total L2 transaction volume. Base held roughly 46.58% of L2 DeFi TVL, Arbitrum about 30.86%, and Optimism around 6%, totaling over 83%. Among the roughly 130 L2s in the Ethereum ecosystem, more than 80% had daily active users below 1 UOPS.
Structural decline in mainnet fees. The bigger trend pressure comes from fees. By January 2026, average Ethereum transaction fees had dropped to about $0.01, more than 99% lower than the 2021 peak. After the Glamsterdam upgrade, fees are expected to compress further. When mainnet fees are no longer a barrier to entry, L2s whose core value proposition is simply "cheaper Ethereum" lose their reason to exist.
Price context. According to Gate market data, as of May 21, 2026, ETH was priced at $2,142.28, up 1.61% in 24 hours, down 6.19% over seven days, and down 5.70% over 30 days. Ethereum’s market cap was about $258.542 billion, with a staking rate above 31%. This price reflects a tug-of-war between short-term macro pressures and long-term fundamentals, rather than direct pricing of L2 ecosystem prospects. However, ETH price weakness has indeed suppressed L2 token market performance and fundraising to some extent.
The table below summarizes key data for the three L2 giants:
| Network | DeFi TVL Share | Daily Transaction Volume | Core Ecosystem Features | Decentralization Stage |
|---|---|---|---|---|
| Base | ~46.58% | Over 8.9 million daily transactions (March 2026), nearing 15 million in May | Consumer and social focus; Coinbase user onboarding | Stage 1 |
| Arbitrum | ~30.86% | Over 2 million daily transactions (post-Cancun upgrade) | Deep DeFi liquidity hub | Stage 1 (BoLD permissionless fraud proofs live) |
| Optimism | ~6% | About 800,000 daily transactions (estimate) | Superchain governance and cross-chain interoperability | Stage 1 |
Data sources: BlockEden, L2BEAT, covering Q1 2026 through May public data.
Competitive Strategies of the Three Giants
Facing shared structural pressure, Base, Arbitrum, and Optimism have pursued distinctly different survival strategies.
Base: Differentiated moat as a consumer gateway. Base’s core advantage isn’t unique technical performance, but its channel and user base. Leveraging Coinbase’s global user foundation, Base has consistently led all L2s in daily transaction volume—about 8.93 million per day in March 2026, approaching 15 million in May. Base’s daily active addresses have surpassed 1 million, far exceeding Arbitrum’s 250,000 to 300,000. Base’s development focus is squarely on consumer applications, social finance, and gaming, establishing a differentiated ecosystem compared to traditional DeFi hubs. For Base, the impact of lower mainnet fees is relatively limited—its users are mainly converted from Coinbase’s retail channel, not simply motivated by "escaping high mainnet costs."
Additionally, a key event warrants attention: On February 18, 2026, Base officially announced its departure from Optimism’s OP Stack and Superchain architecture, opting for a unified, self-operated tech stack. This means Base will retain all sequencer revenue, no longer sharing with Optimism.
Arbitrum: Deep DeFi and accumulation of security trust. Arbitrum’s core asset is its deep DeFi ecosystem, built up over years. Top protocols like Aave, Uniswap, and GMX have concentrated liquidity on Arbitrum, creating an almost irreplaceable network effect. On the security front, Arbitrum has launched BoLD permissionless fraud proofs, becoming one of the first major rollups to achieve fully permissionless fraud proofs. Any participant can challenge on-chain state, eliminating centralized validator bottlenecks. For institutional DeFi users, this security trust cannot be easily replaced by lower mainnet fees.
It’s worth noting that Arbitrum has recently faced some capital outflows—between April 18 and May 9, 2026, Arbitrum’s TVL dropped by about $449 million to around $1.57 billion, with over $1 billion in stablecoin net outflows since early May.
Optimism: Ecosystem alliance and infrastructure layer. Optimism’s approach is more distinctive. By early 2026, Optimism launched OP Enterprise, supporting fintech companies and financial institutions in deploying production-grade chains within 8 to 12 weeks. It offers three service models: fully managed, self-managed, and OP mainnet. Its Superchain framework already covers multiple chains, including Unichain, Celo, and other early partners.
Economically, Optimism passed a proposal in January 2026 to use 50% of Superchain sequencer net revenue for regular OP token buybacks, launching a 12-month pilot in February. However, Base’s departure in February 2026 has put pressure on Superchain’s revenue model. According to Gate market data, as of May 21, 2026, the OP token is down more than 80% year-over-year.
Narrative Review: Will L2 Be "Eaten" by the Mainnet?
The narrative that "L2 will be eaten by L1" gained widespread traction in 2026, but its accuracy requires closer examination.
Oversimplified aspects: Glamsterdam improves transaction execution speed but does not fundamentally solve the mainnet’s state bloat problem. The mainnet still needs to permanently store all historical data, while L2s use blob mechanisms to submit batch data to the mainnet at lower cost. Even if L1’s parallel execution efficiency improves dramatically, L2s retain absolute unit cost advantages, especially for high-frequency trading and micropayment scenarios.
Underestimated aspects: Glamsterdam’s impact on L2s is not "direct competition," but "repositioning." When mainnet fees are around $0.01, generic L2s whose sole selling point is "lower fees" lose their reason to exist. Vitalik has previously noted that L2 decentralization progress is far below expectations, with most networks stuck at Stage 0, relying on centralized security councils. He made it clear: "If you don’t want to go beyond Stage 1, you’re not really scaling Ethereum."
From substitution to complementary differentiation: A more accurate description is that the relationship between L1 and L2 is shifting from a "vertical division of labor" to "horizontal functional differentiation." The Ethereum Foundation’s Strawmap explicitly lists parallel targets for Gigagas L1 (10,000 TPS) and Teragas L2 (about 10 million TPS). L2’s future is not about competing with the mainnet on price, but about providing capabilities the mainnet cannot or will not offer—privacy protection, specialized virtual machines, ultra-low latency, compliance environments, and more.
Multi-Scenario Evolution Projections for the L2 Ecosystem
Based on the above analysis, the L2 ecosystem could evolve along five possible paths over the next 12 to 24 months:
Scenario 1: The giants remain strong, differentiated moats deepen. Base and Arbitrum continue to build their moats in consumer and DeFi directions, respectively. A few specialized L2s gain incremental users in verticals like privacy, AI, and gaming. The foundation for this scenario is that the developer ecosystems and user habits of leading L2s have extremely high migration costs, and lower mainnet fees are unlikely to disrupt them in the short term.
Scenario 2: Gentle reshuffling, accelerated exit of small and medium L2s. Many "copy-paste" L2s lacking differentiated value will face dual exhaustion of capital and users. Institutional analysis suggests generic L2 tokens lacking differentiation may face large-scale purges by the end of 2026. This is essentially a natural deflationary process for the market and may not negatively impact the overall Ethereum ecosystem.
Scenario 3: Significant mainnet reflow, L2 TVL growth slows. If Glamsterdam and Hegotá upgrades go live as scheduled and mainnet experience improves significantly, some protocols with higher security requirements but lower price sensitivity may opt to return directly to L1. Especially in scenarios like RWA tokenization and institutional settlement, L1’s "trusted neutrality" is irreplaceable. In this scenario, L2 TVL growth will slow, but core users and liquidity of leading networks are unlikely to see large-scale attrition.
Scenario 4: Breakthroughs in cross-L2 interoperability, ecosystem shifts from fragmentation to aggregation. The development of Ethereum’s interoperability layer and intent routing systems may greatly reduce friction across L2s, allowing users to perceive a unified ecosystem rather than fragmented networks. In this scenario, boundaries between individual L2s will blur, and competition will shift from single-chain metrics to ecosystem service capabilities.
Scenario 5: External competing chains capture L2 users. Mainnet scaling combined with ongoing evolution of high-performance external chains could create a more complex competitive landscape: L2s are no longer "Ethereum’s high-performance alternative," but are squeezed between an upgraded L1 and other high-performance standalone chains. Currently, external chains excel in metrics like active developer count and DEX trading volume. While this scenario is not highly probable, L2 builders should take it seriously.
Conclusion: L2s Won’t Disappear, But the "Cheap Ethereum" Era Is Over
Ethereum’s L2 ecosystem stands at a structural turning point in 2026. The challenge is not a binary "will the mainnet replace L2s," but a more complex dynamic game: As L1’s capability boundaries expand dramatically, L2s must transform from "Ethereum’s cheap scaling solution" to "an indispensable specialized capability layer."
Base leverages Coinbase’s distribution channel and consumer application focus, Arbitrum relies on years of accumulated DeFi depth and security trust, and Optimism builds on its Superchain infrastructure alliance. Each is constructing its own differentiated moat, but their paths and foundations vary—Base continues to lead in transaction volume, Arbitrum faces short-term capital outflows, and Optimism is restructuring its revenue after Base’s departure from Superchain.
For participants watching the Ethereum ecosystem, several key indicators deserve attention: actual mainnet fee levels after the Glamsterdam upgrade, changes in revenue structure for leading L2 networks, the pace of cross-L2 interoperability, and progress toward Stage 2 decentralization as highlighted by Vitalik. These threads will collectively outline whether L2s can break free from the "cheap Ethereum" narrative and move toward a new value proposition.
As Strawmap suggests: Ethereum’s future is not about choosing L1 or L2, but about enabling both to play irreplaceable roles. Which L2s will survive and thrive in this new landscape? The answer will gradually emerge over the coming months of 2026.
