Five years later, Vitalik overturned the future he had envisioned for Ethereum, fundamentally changing its development path and community direction.

Just like the disproof of Plasma back in the day.

Article by: Rhythm

On February 3, 2026, Vitalik Buterin posted a message on X.

This statement caused a shockwave in the Ethereum community, comparable to his push in 2020 for the “Rollup-centric” roadmap. In that post, Vitalik openly admitted: “The original vision of Layer2 as ‘Branded Sharding’ to solve Ethereum scalability is no longer valid.”

In one sentence, it almost signaled the end of the mainstream Ethereum narrative over the past five years. The Layer2 camp, once highly anticipated and seen as Ethereum’s lifeline, is now facing its greatest legitimacy crisis since inception. More direct criticism followed, with Vitalik bluntly writing in the post: “If you create an EVM that processes 10,000 transactions per second, but its connection to L1 is via a multi-sig bridge, then you are not actually scaling Ethereum.”

Why did the once-saving grace become a burden to be abandoned today? It’s not just a technical shift; it’s a brutal game of power, interests, and ideals. The story begins five years ago.

How did Layer2 become Ethereum’s lifeline?

The answer is simple: it’s not a technical choice, but a survival strategy. Rewind to 2021, when Ethereum was mired in the “Noble Chain” quagmire.

Data doesn’t lie: on May 10, 2021, the average transaction fee on Ethereum hit a record high of $53.16. During the NFT craze, Gas prices soared over 500 gwei. What does this mean? An ordinary ERC-20 token transfer could cost dozens of dollars, and a token swap on Uniswap could cost up to $150 or more.

The DeFi Summer of 2020 brought unprecedented prosperity to Ethereum, with total value locked (TVL) skyrocketing from $700 million at the start of the year to $15 billion by year-end, a growth of over 2100%. But this prosperity came at a cost: network congestion. By 2021, as the NFT wave swept in, blue-chip projects like Bored Ape Yacht Club caused further congestion, with individual NFT transactions costing hundreds of dollars in Gas fees. Some collectors in 2021 were willing to pay over 1000 ETH for a Bored Ape, but ultimately gave up due to high Gas costs and complex transaction processes.

Meanwhile, a challenger called Solana emerged. Its data was shocking: tens of thousands of transactions per second, with fees as low as $0.00025. The Solana community not only mocked Ethereum’s performance but also directly attacked its bloated and inefficient architecture. The narrative that “Ethereum is dead” spread widely, and the community was filled with anxiety.

Against this backdrop, in October 2020, Vitalik formally proposed in the “Ethereum Roadmap centered on Rollups” a concept: positioning Layer2 as Ethereum’s “Branded Sharding.” The core idea was that Layer2 would handle massive off-chain transactions, then compress and bundle the results back to the mainnet, theoretically achieving unlimited scalability while inheriting Ethereum’s security and censorship resistance.

At that time, the entire Ethereum ecosystem’s future was almost entirely dependent on the success of Layer2. From the March 2024 Dencun upgrade introducing EIP-4844 (Proto-Danksharding), which provided cheaper data availability for Layer2, to various core developer meetings, everything was paving the way for Layer2. After the Dencun upgrade, Layer2 data publishing costs dropped by at least 90%, with Arbitrum’s transaction fees plunging from about $0.37 to $0.012. Ethereum was gradually pushing L1 into the background, quietly becoming a “settlement layer.”

But why did this bet fail to pay off?

Those “centralized databases” valued at $1.2 billion

If Layer2 had truly achieved its original vision, they wouldn’t be losing favor today. But what exactly went wrong?

Vitalik pinpointed the fatal flaw in his article: decentralization progress is too slow. Most Layer2 solutions have yet to reach Stage 2 — having fully decentralized fraud or validity proof systems, allowing users to unilaterally withdraw assets in emergencies without permission. They are still controlled by centralized sequencers, which handle transaction bundling and ordering. Essentially, they are closer to centralized databases disguised as blockchains.

The conflict between business reality and technical ideals is laid bare here. Take Arbitrum as an example: its developer, Offchain Labs, raised $120 million in Series B funding in 2021, with a valuation of $1.2 billion, backed by top firms like Lightspeed Venture Partners. Yet, to this day, this giant with over $15 billion in TVL and holding about 41% of the Layer2 market share remains in Stage 1.

The story of Optimism is similarly intriguing. Led by Paradigm and Andreessen Horowitz (a16z), it completed a $150 million Series B in March 2022, with total funding reaching $268.5 million. In April 2024, a16z privately bought $90 million worth of OP tokens. Despite such strong capital backing, Optimism also only reached Stage 1.

The rise of Base reveals another dimension of the problem. Launched by Coinbase in August 2023, Base quickly became a market darling. By the end of 2025, its TVL hit $4.63 billion, accounting for 46% of the entire Layer2 market, surpassing Arbitrum to become the DeFi TVL leader. But Base’s decentralization is even lower, as it is fully controlled by Coinbase, making it more akin to a centralized sidechain in terms of architecture.

Starknet’s story is even more ironic. This Layer2, using ZK-Rollup technology and developed by Matter Labs, raised $458 million, including a $200 million Series C led by Blockchain Capital and Dragonfly in November 2022. Yet, its token STRK has shrunk 98% from its all-time high, with a market cap of about $283 million. On-chain data shows its daily protocol revenue is barely enough to cover a few servers’ operating costs, and its core nodes remain highly centralized. It only reached Stage 1 around mid-2025.

Some project teams even privately admit they may never fully decentralize. Vitalik cited a case in his post: a project claimed they would never further decentralize because “regulatory requirements from clients demand they retain ultimate control.” This infuriated Vitalik, who responded bluntly:

“This might be the right thing for your clients. But it’s clear that if you do that, you’re not ‘scaling Ethereum.’”

This comment almost sentenced all projects claiming to be Ethereum L2 but refusing to decentralize. Ethereum’s goal is to create a scalable, decentralized extension that can expand to broader spaces, not a bunch of entities cloaking themselves in Ethereum but acting as centralized fiefdoms.

Deeper still, a fundamental contradiction exists between decentralization and commercial interests. Centralized sequencers mean project teams can control MEV (Maximum Extractable Value), respond more flexibly to regulations, and iterate products faster. Fully decentralized systems mean relinquishing these controls, handing power to the community and validator networks. For projects backed by venture capital and under growth pressure, this is a tough choice.

If Layer2 truly achieved full decentralization, would it still fall out of favor? The answer might still be yes. Because Ethereum itself has changed.

When the mainnet becomes faster and cheaper than sidechains

Why does Ethereum no longer need Layer2 for scaling as much?

As early as February 14, 2025, Vitalik sent a key signal. He published an article titled “Even in a L2-heavy Ethereum, there are reasons to have higher L1 Gas limits,” explicitly stating “L1 is scaling.” At the time, this sounded more like reassurance for purists of the mainnet, but looking back now, it was actually a rallying cry for Ethereum’s mainnet to re-engage in competition with Layer2.

Over the past year, Ethereum L1’s expansion has far exceeded expectations. Breakthroughs include: EIP-4444 reducing historical data storage needs, stateless client tech making nodes lighter, and most importantly, continuous increases in Gas Limit. Early 2025, the Gas Limit was still 30 million; by mid-year, it had risen to 36 million, a 20% increase. This was Ethereum’s first significant Gas Limit hike since 2021.

But that’s just the beginning. According to Ethereum core developer plans, two major hard forks are scheduled for 2026. The Glamsterdam upgrade will introduce perfect parallel processing, with Gas Limit soaring from 60 million to 200 million — over three times higher. The Heze-Bogota fork will add FOCIL (Fork-Choice Enforced Inclusion Lists), further improving block construction efficiency and censorship resistance.

The Fusaka upgrade completed on December 3, 2025, already demonstrated the power of L1 expansion. Post-upgrade, Ethereum’s daily transaction volume increased by about 50%, active addresses rose by roughly 60%, and the 7-day moving average of daily transactions hit a record high of 1.87 million, surpassing the peak during the 2021 DeFi boom.

The result is astonishing: transaction fees on Ethereum’s mainnet have dropped to extremely low levels. In January 2026, the average transaction fee fell to $0.44, over 99% lower than the $53.16 peak in May 2021. During off-peak times, a single transaction often costs less than $0.10, sometimes only $0.01, with Gas prices as low as 0.119 gwei. This figure is approaching Solana’s level, and Layer2’s cost advantage is rapidly diminishing.

Vitalik calculated a detailed scenario in his February article. Assuming ETH price at $2,500, Gas price at 15 gwei (long-term average), and demand elasticity close to 1 (doubling Gas Limit halves the price). Under these assumptions:

Censorship-resistant demand: Currently, forcing a transaction reviewed by L2 on L1 costs about 120,000 gas, or $4.50. To reduce this to below $1, L1 needs to expand 4.5 times.

Cross-L2 asset transfers: Withdrawing from one L2 to L1 costs about 250,000 gas, and depositing into another L2 costs 120,000 gas, totaling $13.87. With optimized design, only 7,500 gas is needed, costing $0.28. To reach a $0.05 target, L1 needs to expand 5.5 times.

Large-scale exit scenarios: Take Sony’s Soneium as an example, with about 116 million monthly active users. Using an efficient exit protocol (7,500 gas per user), Ethereum can support the emergency exit of 121 million users within a week. But supporting multiple applications of this scale would require about 9 times expansion of L1.

And these expansion goals are gradually being realized in 2026. Technological advances are fundamentally changing the game. When L1 itself becomes faster and cheaper, why should users endure the cumbersome cross-chain bridges, complex interactions, and potential security risks of Layer2?

The security issues of cross-chain bridges are no joke. In 2022, bridges became prime targets for hackers. In February, the Wormhole bridge was hacked for $325 million; in March, the Ronin bridge suffered the largest DeFi attack, losing $540 million; others like Meter and Qubit were also compromised. According to Chainalysis, in 2022, the total stolen cryptocurrency from cross-chain bridges reached $2 billion, accounting for most DeFi attack losses that year.

Liquidity fragmentation is another pain point. As the number of Layer2 solutions explodes, DeFi protocol liquidity is dispersed across dozens of chains, increasing slippage, reducing capital efficiency, and degrading user experience. Users wanting to move assets between Layer2s face complex bridging processes, long confirmation times, and additional costs and risks.

This leads to the next, and perhaps most brutal, question: what should the Layer2 projects that raised huge funds and issued tokens do now?

Valuation bubbles and ghost cities

Where did all the money in Layer2 go?

In recent years, the Layer2 space has been more like a massive financial game than a technological revolution. Venture capitalists waved their checks, pushing the valuations of various L2 projects to astonishing heights. zkSync raised $458 million; Offchain Labs behind Arbitrum was valued at $1.2 billion; Optimism raised $268.5 million; Starknet raised $458 million. Behind these numbers are top VCs like Paradigm, a16z, Lightspeed, Blockchain Capital, and others.

Developers are keen to “stack” across different L2s, building complex DeFi Lego sets to attract more liquidity and airdrop hunters. But real users are worn down by repeated cumbersome cross-chain operations and hidden costs.

A harsh reality is that the market is becoming highly concentrated at the top. According to data from crypto research firm 21Shares, Base, Arbitrum, and Optimism now control nearly 90% of transaction volume. Base, leveraging Coinbase’s traffic and user base, experienced explosive growth in 2025, with TVL soaring from $1 billion at the start of the year to $4.63 billion at year-end, and quarterly trading volume reaching $59 billion, a 37% increase quarter-over-quarter. Arbitrum remains second with about $19 billion TVL, followed by Optimism.

But outside the top players, most L2 projects, deprived of airdrop expectations, saw their real user numbers plummet to near zero, turning into true “ghost cities.” Starknet is a prime example. Despite its token price dropping 98% from its high, its extremely low daily active users and fee income mean its P/E ratio remains in a bubble zone. This indicates a huge gap between market expectations and its current ability to generate real value.

Ironically, as L2 fees dropped sharply due to EIP-4844, the data availability fees they pay to L1 also fell, reducing Ethereum’s fee income. In January 2026, some analysts pointed out that the Dencun upgrade caused many transactions to shift from L1 to cheaper L2, which was a major reason for Ethereum’s network fees dropping to their lowest since 2017. While lowering their own costs, Layer2s are also draining the economic value of L1.

In its 2026 Layer2 outlook report, 21Shares predicts that most Ethereum Layer2s may not survive past 2026. The market will undergo a brutal consolidation, and only projects with high performance, true decentralization, and unique value propositions will succeed.

This is the real purpose behind Vitalik’s recent critique. He aims to burst this infrastructure hype bubble and pour cold water on this unhealthy market. If a Layer2 cannot offer more interesting and valuable features than L1, it will ultimately become just an expensive transitional artifact in Ethereum’s development history.

Ethereum is reclaiming its sovereignty

Vitalik’s latest advice points to a new path for Layer2: abandon scalability as the sole selling point, and instead explore features and added value that L1 cannot or will not provide in the short term. He specifically mentions several directions: privacy protection (via zero-knowledge proofs for on-chain private transactions), efficiency improvements for specific applications (like gaming, social networks, AI computing), ultra-fast transaction confirmation (milliseconds instead of seconds), and exploration of non-financial use cases.

In other words, Layer2’s role will shift from being Ethereum’s clone to becoming a suite of specialized plugins. They will no longer be the only solution for scaling but will serve as functional extension layers within the Ethereum ecosystem. This is a fundamental repositioning and a return of power — Ethereum’s core values and sovereignty will be re-anchored on L1.

Vitalik also proposed a new framework: viewing Layer2 as a spectrum rather than a binary classification. Different L2s can make different trade-offs in decentralization, security guarantees, and functionality, with the key being transparent communication to users about what guarantees they offer, rather than all claiming to “scale Ethereum.”

This reckoning has already begun. Those Layer2s that rely on high valuations but lack real daily active users face the final judgment. Projects that can find their unique value propositions and truly decentralize may survive in this new landscape. Base might continue to leverage Coinbase’s traffic and Web2 user base, but must face questions about decentralization. Arbitrum and Optimism need to accelerate their Stage 2 development to prove they are more than centralized databases. zkSync and Starknet, as ZK-Rollup projects, must demonstrate the unique value of zero-knowledge proofs while significantly improving user experience and ecosystem vitality.

Layer2 has not disappeared, but the era where it was Ethereum’s only hope is over. Five years ago, when competitors like Solana cornered Ethereum, it handed the scaling hope to Layer2 and restructured its entire roadmap. Now, it turns out the best scaling solution is to make itself stronger.

This is not betrayal, but growth. Projects that cannot adapt to this evolution will pay the price. When Gas Limit hits 200 million by the end of 2026, when Ethereum L1 transaction fees stabilize at a few cents or less, and users no longer need to endure complex cross-chain bridges and security risks, the market will vote with its feet. Projects that once boasted sky-high valuations but failed to create real value will be forgotten in the great washout.