Original Title: Why Scaling Ethereum is Bullish for L2s
Original Author: Etherealize
Original Compilation: Ken, ChainCatcher
On February 3rd, Vitalik Buterin made a post on X that garnered over 6 million views. "The original vision of L2s and their role in ETH no longer tenable," he wrote. "We need a new path."
Stakeholders of competing blockchains were quick to portray this as an admission of failure. Cryptocurrency news media called it a "major reversal." The narrative that formed was that ETH had finally admitted defeat—the rollup-centric roadmap wasn't working, and the monolithic scaling approach adopted by blockchains like Solana had proven correct.
This narrative is wrong. If you base your investment decisions on it, you risk being on the wrong side and missing the most important infrastructure transformation currently happening in the cryptocurrency space.
What Vitalik Actually Said
If you read the full text instead of just the headline, the message is clear. ETH is not abandoning Layer 2 (L2). It is shifting from a "rollup-centric" scaling approach (which expected L2s to be replicas of the base layer) to a model where L1 (Layer 1) itself scales radically. L2s are still important, but the reason has changed: customization.
The original vision saw L2 rollups as replicas—simple copies of the Ethereum Virtual Machine, free from the burden of base layer consensus. The idea was that these rollups would eventually decentralize to "Stage 2," inheriting ETH's full security guarantees while providing cheaper transactions. In exchange, they would contribute to ETH's liquidity network effects and security budget.
But this didn't happen. As Vitalik acknowledged: "The progression of L2s towards Stage 2 (and subsequently, progress in interoperability) has been much slower and more difficult than initially expected." Many chains calling themselves L2s are, in practice, centralized blockchains with ETH bridges. They can unilaterally change rules, censor transactions, migrate entirely, while contributing little to ETH's network effects.
Two things then happened that made the original vision obsolete. Both are positive developments.
The Base Layer is Scaling Rapidly
After the London hard fork in August 2021, ETH's gas limit was set at 30 million gas per block. This level was maintained for over three years. The ETH community has been cautious about increasing throughput because there is a real trade-off at the core of blockchain design: pushing too much computation on-chain raises the hardware requirements for validators, leading to network concentration in the hands of a few and weakening the decentralization that gives the system its value.
Largely, this is the trade-off that ETH's competitors chose to ignore. For example, a Solana validator today requires enterprise-grade hardware: 24+ physical CPU cores, 256 GB of RAM, multiple enterprise NVMe SSDs, and a 10Gbps network connection. The monthly hosting cost for a competitive validator can exceed $1,000. In contrast, an ETH validator can run on a $1,100 mini-computer under your desk. This is not a trivial difference. It is why ETH can have about 1 million active validators while maintaining a level of decentralization that other smart contract platforms struggle to match. As of early 2026, the Solana network had roughly 800 active validators.
But blockchains do need to scale. High-performance competitors have proven there is massive demand for cheap, fast L1 transactions. ETH's response has been a broader cultural shift—from "long-term research" to "short-term execution"—and the results are showing.
In 2025, through coordinated validator action, the gas limit doubled from 30 million to 60 million, while the Pectra and Fusaka upgrades expanded blob capacity and added other protocol improvements. The Ethereum Foundation has also committed to an aggressive roadmap aiming to roughly triple L1 throughput annually for the foreseeable future.
<极简雅黑span style="font-size: inherit; font-family: PingFang SC,Helvetica Neue,Helvetica,Arial,Hiragino Sans GB,Heiti SC,Microsoft YaHei,WenQuanYi Micro Hei,sans-serif;">By the end of 2026, the goal is to push the gas limit above 100 million. In 2027, block time is expected to halve from 12 seconds to 6 seconds (potentially even 4 seconds), effectively doubling throughput again without changing block size. That same year, block-level access lists will allow nodes to process transactions in parallel, removing a major computational bottleneck. In 2028, migrating to a binary tree state structure will allow for higher gas limits by eliminating the need for validators to store the entire state on disk. By 2029, the network will begin transitioning to a native zero-knowledge architecture—a fundamental architectural change that will revolutionize scaling mathematics.
The key breakthrough enabling this long-term vision is zkEVM. Currently, every node in an L1 blockchain must re-execute every transaction to verify the state. zkEVM compresses the verification process into a constant-sized cryptographic proof that can be checked with minimal computational resources. When combined with Ethereum's data availability sampling—which will allow validators to verify data existence without downloading it all—it creates a path to throughput rivaling high-performance chains while preserving the decentralization that makes ETH block space uniquely valuable.
This is about five years ahead of most observers' expected timelines. It is "spectacular" enough that Ben Edgington, a leader during ETH's transition to proof-of-stake, announced he was ending his retirement to rejoin the project.
Ethereum Foundation researcher Justin Drake outlined the technical North Star goals: a "fast L1" with second-level finality; a "giga-gas L1" with real-time zkEVM proofs achieving 10,000 transactions per second; and a "tera-gas L2" with data availability sampling achieving 10 million transactions per second. The roadmap also prioritizes post-quantum cryptography and native privacy features at the base layer.
The New Value Proposition for Layer 2
So, if L1 is scaling, what is the point of L2?
L2s have found their product-market fit: serving institutions that want both the security of ETH and the liquidity of the ETH ecosystem, but also need the ability to customize their chain to better serve their customers and comply with regulator requirements.
This ultimately leads to the "Stage 2" problem. For an L2 to achieve Stage 2 decentralization, it must give up the ability to unilaterally upgrade its governing bridge and proof system contracts—including the ability to respond quickly to regulatory demands or patch critical vulnerabilities. For institutions onboarding millions of users to the ETH ecosystem, this is a real operational constraint.
This is the tension at the core of today's L2 ecosystem. Users can still withdraw their assets back to ETH's L1, which is the most important security guarantee a rollup provides. But without achieving Stage 2, operators can still upgrade bridge contracts, censor transactions, or change rules. And, due to a lack of interoperability, each L2 fragments liquidity and competes with other L2s in ways not fundamentally different from alternative L1s.
Vitalik's article resolves this tension by acknowledging reality: L2s exist on a spectrum, and that's okay. Some L2s will pursue full Stage 2 decentralization and function as true extensions of ETH block space. Others will maintain more centralized control in exchange for customization, which is also a valid use case, as long as the trade-offs are honestly communicated in marketing.
The demand from institutions for the second type of L2 is huge and growing, with Robinhood's decision to build an ETH L2 being the clearest illustration.
In June 2025, Robinhood announced at EthCC (Ethereum Community Conference) that it would build its own Ethereum Layer 2 using the Arbitrum tech stack, rather than launching a new L1 blockchain. This surprised many in the crypto industry. Robinhood is one of the world's largest retail brokers. It has the resources and user base to launch its own chain. It had actively discussed doing so. But it ultimately chose not to.
The reasoning articulated by Robinhood Crypto head Johann Kerbrat gets to the heart of why L2s matter: "Ensuring the security of a truly and highly decentralized chain is extremely difficult, and we can basically get that for free from Ethereum. When you look at newly created L1s, they are not really decentralized or secure, so ultimately, what you have is a fancy database that might be slower than a real database."
The second factor is liquidity. Robinhood's goal is to tokenize all assets—starting with public stocks and expanding to private equity, real estate, and other real-world assets. This requires access to Ethereum's existing liquidity network. As Kerbrat said: "We need this liquidity... If you are alone on your private island, no one can come and go freely. I believe we can get customers because Robinhood is a large platform, but we want to rebuild the entire financial system on-chain, and we need everyone to be able to come to our island."
Robinhood CEO Vlad Tenev compared the customizability of L2s to building on alternative L1s like Solana, framing it as a trade-off between short-term and long-term value: "In the long run, control is more important; it allows us to build better products. Plus, the technology behind these rollups has become so excellent that you're not really missing out on much." As an L2, Robinhood retains full control over sequencer revenue, gas fees, regulatory customization, and product roadmap—while inheriting ETH's security and settlement guarantees. It can call it 'Robinhood Chain' while letting Ethereum handle the hardest parts.
Robinhood is not alone. Coinbase (Base), Kraken (Ink), and OKX (X Layer) have all launched their own ETH L2s. But the more telling signal is who is choosing to build with them. Just this month, Nasdaq partnered with Kraken to build a tokenized stock gateway, and the Intercontinental Exchange (ICE), parent company of the New York Stock Exchange, invested $200 million in OKX with plans to put NYSE-listed stocks on-chain.
These institutions need the security of ETH and the liquidity of its ecosystem. But they also need regulatory compliance, privacy controls, custom fee structures, and operational control. A permissionless, fully transparent base layer cannot meet all these needs for them. But a Layer 2 built on top of it can.
As Vitalik wrote a few days later in a follow-up clarification, L2s should "do something that actually brings something new" (e.g., privacy, app-specific efficiency, ultra-low latency, institutional compliance, etc.). Most critically: "The vibe should match the substance." The degree to which an L2 is connected to ETH in its public image should match the degree of connection in reality. A sidechain with a bridge is not the same as a Stage 2 rollup that cannot survive离开 ETH. Calling oneself an "ETH L2" should imply specific things about its security guarantees.
This is about protecting the integrity of the ETH brand, which in turn protects the trust that is beginning to be placed in ETH by institutions.
Layer 2 remains the best business model in crypto. You don't need to spend millions per year on validator infrastructure or inflate a token to pay for security. You inherit ETH's security and pay for it as you use block space.
The Flywheel Effect: Why Scaling L1 Makes L2s More Useful
This is the part those touting "ETH is abandoning L2s" completely miss: scaling the base layer doesn't compete with L2s. It makes them vastly more useful.
To understand why, you need to understand what ETH is at the protocol level. It operates as a globally replicated ledger. Every full node independently verifies every transaction to ensure the ledger is correct. Protocol parameters like the gas limit and block time must be kept conservative enough for ordinary machines to keep up, otherwise you end up requiring data-center-grade hardware to participate, recreating the centralized infrastructure you were trying to escape.
This means raw L1 throughput is inherently scarce, which is what makes ETH block space valuable. This is precisely why a transaction settled on ETH has stronger guarantees than one settled on a chain with only a few hundred validators running in three data centers.
Rollups solve this limitation through a clever division of labor. They move most user transactions off-chain to the L2, where it's fast and cheap, while primarily using ETH for two things: data availability (publishing compressed transaction data from which anyone can reconstruct the L2's state) and final settlement (anchoring the L2's state transitions to L1 consensus). By batching many off-chain transactions together, rollups allow many users to share the cost of a single L1 transaction.
When ETH scales its L1, it directly reduces the cost of these two functions. More gas per block means cheaper settlement costs. More blob capacity means more L2s can publish data simultaneously without competing for scarce data availability. Faster block times mean L2 withdrawals and cross-chain operations become faster. Faster finality means L2s can confirm transactions with higher certainty in less time.
The result is a system where each does what it does best: L1 handles what it's best at (low-risk DeFi, high-value settlement, and serving as the canonical data source), while L2s compete on specialized use cases. This competitive dynamic is much healthier than the current situation—where the primary reason for L2s' existence is simply that L1 is too slow and expensive for everyday transactions.
The Unresolved Issue: Liquidity Fragmentation
Layer 2 does not solve everything. With current technology, every new L2 is a separate island of assets and users. Without seamless interoperability, the ETH ecosystem functions less like a complete network and more like a dozen competing networks. This is the most valid criticism of the ETH L2 ecosystem.
The original rollup-centric roadmap assumed L2s would converge on interoperability standards, and liquidity would flow freely throughout the ecosystem. This didn't happen. Instead, liquidity became fragmented, and for most users, the experience of bridging assets between different L2s remains slow, expensive, and risky.
The Ethereum Foundation has made this a top priority for 2026. Central to the plan is an "open intent framework" where users simply state what they want to do—swap, bridge, pay—and the system automatically routes the optimal path across different L2s. Behind the scenes, a new Ethereum interoperability layer aims to make transactions across L2s feel indistinguishable from transactions on a single chain. Vitalik has also pushed for the development of native rollup precompiles, which would verify zkEVM proofs directly on L1, improving trustless composability between the base layer and rollups.
This is the next problem to solve. If ETH can get it right, making moving assets between different L2s feel like using one chain, then every new L2 strengthens the entire network instead of fragmenting it.
What This Means
As of this writing, ETH's market capitalization is approximately $240 billion. It is the world's most valuable blockchain after Bitcoin, and by a significant margin. The "ETH is dying" narrative simply doesn't match what the market is actually telling you.
Robinhood is tokenizing thousands of stocks on an ETH L2. The gas limit has doubled and has a credible roadmap to increase 10x from current levels within four years. Institutional adoption of ETH-based L2s is accelerating, not slowing down. And the engineering community's excitement is at its highest in years—not just about the roadmap itself, but about the quality of talent it is attracting back into the fold of active contributors.
What is happening is a maturation of strategy. The original rollup-centric roadmap was a pragmatic response to an emergency: in 2020, ETH could not quickly scale its L1 without sacrificing decentralization, and competitors were gaining market share. That emergency is over. But the engineering talent and infrastructure ETH invested in during that period—blobs, data availability sampling, zkEVM research, rollup frameworks—was not wasted. It laid the foundation for the next phase: a radically scaling L1 surrounded by an ecosystem of customizable L2s serving institutions and specific needs that a general-purpose blockchain could never meet.
The correct read of Vitalik's article is not that L2s failed. It's that the initial framework, which positioned L2s as brand shards bearing the full social responsibility of scaling ETH, was the wrong framing. The new framework is simpler and more honest: L2s exist on a spectrum of decentralization, serving different customer needs. Those closest to ETH inherit its security and contribute to its network effects. Those further away serve legitimate purposes but should not pretend to be what they are not. And the ETH L1 that gives all of this value is about to become much more powerful.
ETH is not abandoning L2s. It is just giving them a more lasting reason to exist than "L1 is too slow." And that should make you more bullish on ETH, not less.
