Is Ethereum Becoming Infrastructure for Its Own Competitors? L2 TVL About to Flip L1

General
1

As someone who’s been contributing to Ethereum’s core development for years, I need to ask the uncomfortable question: Is Ethereum succeeding as infrastructure but failing as an application platform?

The numbers are stark. By Q3 2026, Layer 2 TVL is projected to exceed Ethereum L1 DeFi TVL—$150 billion on L2s versus $130 billion on L1, according to recent industry analysis. The L2 share of total Ethereum TVL is climbing from 27% to 55%. Layer 2 networks are already processing more daily transactions than mainnet (1.9M+ on L2s daily).

The Rollup-Centric Bet

When the Ethereum Foundation pivoted to a rollup-centric roadmap, the vision was clear: L1 provides security and settlement, L2s provide execution and scale. Users and developers would migrate to rollups, which inherit Ethereum’s security while offering 10-100x cheaper transactions.

It worked. Arbitrum, Optimism, Base, zkSync—they’re thriving. But here’s what troubles me:

  1. L2s capture fees and MEV, not L1. All meaningful economic activity happens on L2s. The base layer gets settlement fees, but that’s pennies compared to what L2 sequencers earn.

  2. Liquidity is fragmenting. Every L2 has its own DeFi ecosystem. Bridging between them is clunky. Are we creating 50 isolated economies instead of one unified network?

  3. L2s are Ethereum’s competitors. They compete for developers, users, and liquidity. Some L2s (like Base) are run by centralized companies that could theoretically fork away from Ethereum.

The Uncomfortable Question

Did we abandon L1 scalability too early? Solana processes more transactions on a single chain than Ethereum + all its L2s combined. Aptos, Sui—they’re scaling monolithically without rollups.

If L2s are the answer, why does it feel like we’re building infrastructure for our own competitors?

By late 2026, launching a rollup will be as simple as deploying a smart contract (via Arbitrum Orbit, OP Stack, zkSync Hyperchains). That’s amazing for permissionless innovation. But if we have 500 L2s, each with 0.2% of Ethereum’s liquidity, did we win or lose?

I still believe in the rollup-centric roadmap. But as L2 TVL flips L1 TVL this year, I wonder: Is Ethereum becoming like TCP/IP—essential infrastructure that end users never directly interact with? And if so, is that success or an admission that Ethereum L1 can’t compete with monolithic chains for user-facing applications?

Would love to hear perspectives from the community. Especially those building on L2s—do you see yourselves as extending Ethereum or competing with it?

2

Brian, you’re touching on something I experience every single day as a developer building DeFi interfaces. The UX nightmare is real.

I spent last weekend trying to help a friend (total crypto newcomer) bridge USDC from Ethereum to Arbitrum, then to Optimism, then to Base. Here’s what we went through:

  1. Three different bridge UIs, each with different terminology and approval flows
  2. Waited 7 days for the Optimism → Ethereum withdrawal to finalize
  3. Lost track of which funds were on which chain by the end
  4. Gas fees were confusing—sometimes paid in ETH, sometimes deducted from bridged amount

My friend’s response: “This is ridiculous. Why can’t it just work like Venmo?”

The Liquidity Fragmentation Problem

From a developer perspective, here’s what scares me: I’m building a DeFi aggregator right now. On Ethereum L1, I can integrate with Uniswap, Curve, Balancer and get deep liquidity for most pairs.

But if I want to support L2s, I need to integrate with:

  • Arbitrum’s Uniswap V3 (different contract addresses)
  • Optimism’s Velodrome (OP-native DEX)
  • Base’s Aerodrome (Base-native DEX)
  • zkSync’s SyncSwap (zkSync-native DEX)

And the liquidity on each chain is 5-10x less than L1. So even if I do all this integration work, users often get worse prices than staying on L1.

Are We Really Succeeding?

You asked if Ethereum is succeeding as infrastructure but failing as an application platform. From a user perspective, I think we’re failing at both right now.

Users don’t care about “settlement layers” or “security inheritance.” They want:

  • Fast transactions :white_check_mark: (L2s deliver)
  • Cheap transactions :white_check_mark: (L2s deliver)
  • Simple UX :cross_mark: (we’re making it worse)
  • Unified liquidity :cross_mark: (totally fragmented)

Solana has problems, but at least it’s simple. One chain, one wallet, everything just works. I hate to admit this, but I’ve been using Solana more for my personal DeFi stuff lately just because the UX is better.

Maybe shared sequencers and cross-L2 messaging will solve this. But right now, in March 2026, the L2 future feels like a step backward for users even as it’s a step forward for throughput.

Am I being too pessimistic? How do other developers feel about this?

3

Emma’s liquidity fragmentation point hits home for me. As someone running yield optimization strategies, here’s what I’m seeing in Q1 2026:

Following the Yield (and the Liquidity)

Where DeFi capital is actually flowing:

  • Ethereum L1: $130B TVL but declining. Highest fees, slowest speeds. Only protocols with strong network effects (Aave, Maker) keeping liquidity here.

  • Arbitrum: $45B TVL. Winning the L2 DeFi race. GMX, Radiant, Pendle have built sticky ecosystems. Incentives are keeping liquidity here.

  • Optimism: $22B TVL. Velodrome and OP incentives drive liquidity, but it’s mostly mercenary capital that rotates quickly.

  • Base: $18B TVL, growing fast. Coinbase’s distribution is real—retail is onboarding directly to Base, skipping L1 entirely.

The problem: If you’re a DeFi protocol today, where do you deploy?

  • Deploy on L1 only → Miss 55% of Ethereum TVL
  • Deploy on every L2 → Fragment your own liquidity across 5+ chains
  • Pick one L2 → Bet your company on which L2 wins

The MEV and Fee Capture Problem Brian Mentioned

This is critical. L2 sequencers are capturing massive value:

  • Base sequencer revenue: Estimated $50M+ annually (all goes to Coinbase)
  • Arbitrum sequencer revenue: Estimated $30M+ annually (centralized sequencer)
  • Optimism sequencer revenue: Similar numbers

Meanwhile, Ethereum L1 blob fees are tiny because L2s batch transactions so efficiently. The base layer is getting pennies while L2s earn millions.

From a pure economic standpoint, L2s are eating Ethereum’s lunch.

What Fixes This?

Brian, you asked if we abandoned L1 scaling too early. Here’s my hot take: It doesn’t matter anymore. The ship has sailed. Capital and users are on L2s now.

The only path forward is making L2s feel like one unified chain:

  1. Shared sequencers (Based Rollups) so MEV doesn’t flow to centralized sequencers
  2. Native cross-L2 DEX aggregators that route orders across chains invisibly
  3. Instant, free bridging between L2s (without 7-day withdrawal periods)

If we don’t solve this, Solana and other monolithic chains will keep winning DeFi users who just want simplicity.

The yield tells the story: More farmers are moving to Solana than staying in the fragmented L2 ecosystem. That should scare everyone who believes in Ethereum’s future.

4

Coming from the startup/product side, I want to add a slightly different angle: Does this L2 fragmentation hurt or help the Ethereum ecosystem?

The Mobile OS Analogy

Remember when Android was fragmenting into dozens of manufacturer-specific versions? Samsung had TouchWiz, HTC had Sense, etc. Everyone worried about fragmentation.

What actually happened? The market consolidated. Most people use Samsung or Google Pixel now. The long tail died.

I predict the same with L2s:

  • Arbitrum will win general-purpose DeFi (first mover, best tech)
  • Base will win retail/payments (Coinbase distribution)
  • zkSync or Starknet will win if zero-knowledge really matters
  • Maybe 2-3 app-specific rollups for gaming/NFTs survive

The other 95% of L2s launching via Rollup-as-a-Service? Dead within 2 years. No users, no liquidity, no reason to exist.

The Business Reality Emma and Diana Are Missing

From a pure product perspective, users don’t care about architecture—they care about outcomes:

  • “Can I send money fast?” → Yes (any L2)
  • “Are fees low?” → Yes (any L2)
  • “Is it secure?” → They assume yes (Ethereum brand)
  • “Is it simple?” → No (fragmentation problem)

Here’s the thing though: The market will solve the simplicity problem. Not through tech, but through consolidation.

Coinbase is putting all its retail users on Base by default. That’s tens of millions of users who will never even know other L2s exist. They’ll just use “Coinbase crypto” which happens to be Base under the hood.

Does Ethereum Win in This Scenario?

Brian asked if Ethereum is becoming infrastructure for its own competitors. My answer: Yes, but that’s fine.

Think about cloud infrastructure:

  • AWS provides infrastructure
  • Shopify, Stripe, Databricks build on top
  • AWS still captures enormous value even though those companies “compete” with AWS services

Ethereum = AWS. L2s = Shopify. Ethereum provides the security layer, L2s provide the app layer. ETH still accrues value through:

  1. L2s needing ETH for settlement
  2. ETH as gas token for cross-L2 transactions
  3. ETH as collateral in DeFi (works on any L2)

The question isn’t “did we fail by building L2s?” The question is: “Will Ethereum capture sufficient value from L2s to justify its position as the #2 crypto?”

That’s TBD, but I’m optimistic. The brand matters. Ethereum = “the secure blockchain.” That’s worth something.

5

Everyone’s focused on UX and liquidity, but I need to inject the security perspective into this discussion, because the proliferation of L2s is creating significant new attack surfaces.

The Bridge Security Problem is Getting Worse, Not Better

Since 2022, cross-chain bridges have lost $2.8 billion to hacks. That’s more than all DeFi protocol hacks combined.

Now we’re multiplying the problem:

  • More L2s = more bridges
  • More bridges = more attack surface
  • Each bridge has different security models (optimistic vs ZK, trust assumptions, etc.)

When you have 50+ L2s, you need 50+ bridge implementations. Even if each bridge is 99% secure, the compounding risk is enormous.

Recent examples from Q1 2026:

  • IoTeX bridge: $4.4M lost (private key compromise)
  • CrossCurve bridge: $3M lost (zero gateway verification)

The Centralized Sequencer Problem

Diana mentioned L2 sequencers capturing fees. Here’s what worries me more: Most L2 sequencers are currently centralized:

  • Arbitrum: Single sequencer controlled by Offchain Labs
  • Optimism: Single sequencer controlled by OP Labs
  • Base: Single sequencer controlled by Coinbase

Security implications:

  1. Censorship risk: Centralized sequencer can censor transactions
  2. MEV extraction: Sequencer has perfect information for frontrunning
  3. Single point of failure: Sequencer goes down → chain stops

Yes, there are “training wheels” and fraud proofs. But if we’re being honest, most L2s in 2026 are still more centralized than we want to admit.

The Rollup-as-a-Service Security Disaster Waiting to Happen

Brian mentioned launching a rollup will be as easy as deploying a contract by late 2026. From a security perspective, this terrifies me.

Deploying a smart contract is already hard to secure (hence the $905M in OWASP Top 10 losses in 2025). Deploying an entire blockchain with:

  • Sequencer infrastructure
  • Bridge contracts
  • Fraud proof systems
  • Upgrade mechanisms

…is 100x more complex. Yet we’re telling developers “click this button and launch your L2!”

What could go wrong:

  1. Misconfigured bridge contracts (happened with CrossCurve)
  2. Weak key management for sequencers (happened with IoTeX)
  3. Untested fraud proof systems
  4. Upgrade keys held by inexperienced teams

My Take on “Infrastructure for Competitors”

Brian, to answer your question directly: Yes, Ethereum is becoming infrastructure, and yes, that increases security risks.

The more complex the system, the more attack surface. A monolithic chain like Solana has one security model to audit and verify. Ethereum + 50 L2s + 100 bridges has 150+ security models, each with different assumptions.

The only way this works long-term:

  1. Shared security standards for L2s (not happening yet)
  2. Formal verification of all bridge contracts (too expensive/slow)
  3. Decentralized sequencers with slashing (years away)
  4. Market consolidation to 3-5 battle-tested L2s (Steve’s point—I agree)

Until then, every new L2 that launches is a potential $100M+ hack waiting to happen. And when that hack occurs, users will blame “Ethereum” not “some obscure L2.” The brand risk is real.