Is the L2 Window Closed? Base Won While New Rollups Collapsed After Incentives Ended

General
1

I’ve been deep in research mode lately, trying to figure out which Layer 2 we should build on for our startup. The data I’m seeing is… honestly pretty alarming.

The market concentration is stark:

  • Base controls 46.58% of all L2 DeFi TVL ($4.63B)
  • Arbitrum holds 30.86%
  • Optimism adds another ~6%

That’s 83% of the entire L2 market controlled by just three networks. Base alone processes nearly 50% of all Layer 2 DEX volume and has over 1 million active addresses.

The Incentive Collapse

What really caught my attention is what happened to newer L2s when their incentive programs ended. We’re not talking about modest declines—we’re seeing 70-90% TVL collapses within weeks of tokens launching or rewards drying up.

The poster child for this is Blast: TVL crashed 97% from $2.2B in June 2024 to ~$55M by December 2025. Users left en masse to Base and Arbitrum the moment the airdrop disappointed and incentives ended.

21Shares is now predicting that most Ethereum L2s won’t survive past 2026. L2s without strong distribution or sustainable economics are becoming “zombie chains” or shutting down entirely.

The Founder’s Dilemma

Here’s my concern as someone building a company: if the top 3 L2s already process ~90% of all Layer 2 transactions, are we witnessing winner-take-most network effects locking in?

When I talk to other founders, everyone’s facing the same calculus:

  • Build on Base → get Coinbase’s massive onboarding funnel and 1M+ users
  • Build on Arbitrum → tap into mature ecosystem and established DeFi liquidity
  • Build on Optimism → leverage OP Stack and Superchain momentum
  • Build on NewChain™ → get… what exactly?

The brutal reality is that points-fueled TVL isn’t real demand. It’s rented attention that evaporates the moment rewards stop. If a new L2 needs to bribe users to show up, that’s not product-market fit—that’s a temporary marketing campaign.

Is There Still Room?

This is where I’m genuinely uncertain. Are we seeing:

Option A: True market consolidation where the window is closed for new general-purpose L2s, and anyone launching now is wasting capital?

Option B: A natural shakeout where bad projects fail, but legitimate innovation (specialized L2s for gaming, privacy, specific verticals) can still succeed?

Option C: Early innings of chain abstraction where the underlying L2 becomes invisible to users, opening opportunities for new infrastructure that solves specific problems?

I want to believe in Option B or C. But the data suggests we might be in Option A territory.

The VC Paradox

Here’s what confuses me: if the market is this saturated, why do VCs keep funding new L2 infrastructure instead of applications? Blockchain infrastructure startups raised $3.4 billion in 2025—that’s 23% of all blockchain funding going to build “the next Ethereum” when we already have 40+ L2s processing 100K+ TPS collectively.

That capital could fund thousands of applications on existing chains. Instead it’s fragmenting users across dozens of L2s that most people will never use.

What I’m Watching

For our startup, we’re probably building on Base. The Coinbase funnel is too powerful to ignore, and our investors are comfortable with that choice.

But I’m curious what others think:

  • If you’re building right now, which L2 are you choosing and why?
  • Do you see viable paths for new L2s to compete against the entrenched top 3?
  • Is this healthy market consolidation or problematic centralization?
  • Should new projects just accept that Base/Arbitrum/Optimism won and build there?

The pragmatist in me says the L2 wars are over and we should build where the users are. The idealist hopes there’s still room for innovation. Where’s the truth?

Sources:

2

Steve, the reality is actually worse than your analysis suggests—and I say that as someone who’s spent the last 6 years building L2 infrastructure.

The Infrastructure Economics Don’t Work

Running an L2 is expensive:

  • Validator/sequencer infrastructure costs
  • Security audits and ongoing maintenance
  • Developer relations and ecosystem funding
  • Bridge infrastructure and liquidity
  • 24/7 operational support

Most new L2s are burning through their token treasuries with no clear path to sustainable revenue. They’re hoping that transaction fees eventually cover costs, but with 50+ L2s competing on price, fee markets race to the bottom.

Base doesn’t have this problem—they have Coinbase’s balance sheet and strategic interest. Arbitrum and Optimism built early and captured the developer ecosystem before competition intensified.

The Developer Ecosystem Problem

When I talk to protocol developers, they’re making brutally pragmatic choices:

“Why build on NewChain when Base has 1M+ users I can tap into immediately? Why fragment my already-small user base across yet another chain?”

The apps are choosing established L2s because that’s where:

  • Users already have wallets set up
  • Liquidity already exists for token swaps
  • Infrastructure (indexers, RPCs, explorers) is mature
  • Developer tooling actually works

Starting from zero on a new L2 means you’re competing with established apps on established chains. Good luck with that.

The Technical Moat is Real

Base has something no new L2 can replicate: Coinbase’s onboarding funnel. They can turn 100M+ Coinbase users into Base users with UI updates. That’s not a moat, that’s a castle surrounded by an ocean.

Arbitrum has ecosystem maturity—the network effects of being first-mover to scale. GMX, Treasure DAO, Camelot DEX… these aren’t just apps, they’re economic ecosystems that took years to build.

New L2s are trying to compete against this with… airdrops? It’s like trying to build a city by bribing people to visit temporarily.

My Prediction: Zombie Chains by Q4 2026

Most L2s will become what I call “zombie chains”:

  • Technically operational but practically dead
  • Maybe 100-1000 daily transactions (all bots)
  • No organic developer activity
  • Founders moved on to “the next big thing”

Some will shut down entirely. I expect we’ll see:

  • Sequencer shutdowns announced as “planned maintenance”
  • Bridge withdrawals enabled for 6 months, then deprecated
  • Token treasuries quietly liquidated
  • GitHub repos archived

The ones that survive will be specialized L2s with clear value props:

  • Gaming L2s with actual games (not just infrastructure waiting for games)
  • Privacy-focused L2s solving real regulatory arbitrage use cases
  • Vertical-specific chains (finance, supply chain) with committed customers

The One Nuance: Application-Specific Rollups

Here’s where I disagree slightly with the doom thesis: the window might be closed for general-purpose L2s, but app-specific rollups could still work.

If you’re building a game that needs custom VM logic, ultra-low latency, and specific economic models—launching your own L3 on top of Base/Arbitrum might make sense.

But another general-purpose EVM rollup competing with Base for DeFi users? That ship has sailed, the harbor is closed, and the ocean is frozen over.

Where Should You Build?

Steve, you’re already making the right call choosing Base. But here’s the real calculus:

Build on Base if: You need maximum user accessibility and Coinbase integration
Build on Arbitrum if: You need mature DeFi ecosystem and established protocols
Build on Optimism if: You want OP Stack standardization and Superchain composability

Launch your own L2 if: You have $100M+ in funding, a 3-year runway, and a distribution strategy that doesn’t rely on incentives. (So basically, don’t.)

The L2 wars are over. The winners consolidated. Now we build applications.

3

Okay, maybe I’m going to have an unpopular take here, but… I actually think this consolidation is a good thing?

Hear me out.

The Developer Experience Was Awful

Six months ago, I was trying to figure out which L2 to deploy our protocol on. The options were overwhelming:

  • 50+ different L2s all claiming to be “the fastest” or “most decentralized”
  • Each one had different dev tools, different quirks, different communities
  • Documentation quality ranged from “excellent” to “404 page not found”
  • Some chains had active Discord servers, others were ghost towns

As a developer, this was paralyzing. I spent two weeks just evaluating options when I should have been building.

Consolidation = Better Infrastructure

Now that we’re seeing market concentration around Base, Arbitrum, and Optimism, the developer experience is so much better:

Documentation: These chains have comprehensive docs because they have the resources to maintain them

Tooling: Hardhat, Foundry, all the major dev tools have first-class support for the top 3 L2s

Community: Active Stack Overflow questions, Discord channels that actually get responses, tutorials that aren’t outdated

Infrastructure: Block explorers that work, reliable RPC endpoints, indexing services like The Graph

When you build on a smaller L2, you’re often the first person to encounter specific edge cases. On Base/Arbitrum, someone already hit that bug and documented the solution.

Liquidity Fragmentation Was Killing UX

From a user perspective, having 50 L2s was terrible:

  • Your tokens are stuck on Chain A but the app you want is on Chain B
  • Bridging costs money and takes time
  • Liquidity is split across dozens of chains, so slippage is high everywhere
  • Your wallet needs to track balances across 10 different networks

This is not better than just using Ethereum mainnet. It’s arguably worse.

Market consolidation means users can stay on one or two L2s and access most of the ecosystem. That’s a massive UX improvement.

Bad Projects Should Fail

Lisa mentioned “zombie chains” like it’s a bad thing, but… isn’t this just the market working correctly?

If an L2 only has users because they’re being bribed with points and airdrops, it doesn’t deserve to exist. That’s not product-market fit, that’s a marketing budget.

When incentives end and usage collapses 90%, the market is sending a clear signal: nobody actually wanted this product.

I know that sounds harsh, but it’s the same dynamic as any startup: most fail, a few succeed, and that’s healthy. We shouldn’t prop up bad L2s with endless incentives any more than we should prop up failing restaurants.

The Historical Parallel

This reminds me of the L1 wars in 2021:

  • Everyone was launching “Ethereum killers”
  • Solana! Avalanche! Fantom! BSC! Terra! Harmony!
  • VCs threw billions at infrastructure
  • Most of them still exist but have tiny market share

The winners (Ethereum + Solana) have 90%+ of developer activity now. And you know what? The ecosystem is healthier for it.

Developers know where to build. Users know where to go. Liquidity is concentrated. The whole system works better with 2-3 dominant L1s than with 20 mediocre ones.

Why Do VCs Keep Funding New L2s?

Steve asked this and I’ve wondered the same thing. My theory:

VCs are optimizing for different outcomes than users or developers. They need big exits, and infrastructure plays can return 100x if they work. But 90% will fail, and that’s okay from their portfolio perspective.

It’s rational from a VC standpoint even if it’s irrational from an ecosystem standpoint.

The misalignment is: VCs want shots on goal for 100x returns, while the ecosystem needs consolidation and quality over quantity.

Where I’m Building

For what it’s worth, we deployed on Base last month. The decision took about 15 minutes once we looked at the data:

  • 1M+ active users
  • Coinbase Smart Wallet integration
  • Best onboarding funnel for non-crypto-native users
  • Strong developer community

Could we have gotten “incentives” from a smaller L2 to deploy there? Probably. But why would we choose a chain with 5000 users over one with 1 million?

My Controversial Conclusion

The L2 window closing is actually good for everyone except L2 founders and their VCs.

Developers get better tools and clearer choices.
Users get better UX and less fragmentation.
The ecosystem gets sustainable infrastructure instead of bribe-based growth.

The only losers are:

  • Teams that raised $50M to build “yet another rollup”
  • VCs who funded 15 different L2s hoping one would win
  • Projects that relied on incentive tourism instead of genuine value

And honestly? I’m okay with that. Let’s build apps that people actually want to use, on infrastructure that actually works.

Is that too harsh? Maybe I’m being naive, but consolidation around quality feels like progress to me.

4

Emma, I love your optimism, but as someone who’s watched countless L2s try to buy users with yield, let me add some cold hard DeFi economics to this discussion.

Incentive Economics are Fundamentally Broken

I’ve run yield strategies on probably 30+ different L2s over the past two years. Here’s what I learned:

Rented attention is not organic growth.

Every new L2 follows the same playbook:

  1. Launch with 100M+ token allocation for “ecosystem incentives”
  2. Pay protocols to deploy (sometimes $500k-$1M per major DeFi protocol)
  3. Bribe liquidity providers with 50-200% APYs funded by token emissions
  4. Watch TVL skyrocket as yield farmers like me show up
  5. Tout “fastest growing L2” metrics to raise next round
  6. TGE happens, incentives dry up, everyone leaves

Rinse, repeat, collapse.

The Blast Case Study is Brutal

Steve mentioned Blast earlier. Let me give you the numbers from someone who farmed it:

June 2024 (peak): $2.2B TVL
December 2025 (post-airdrop): ~$55M TVL
Decline: 97.5%

But here’s what the TVL number doesn’t show: who was using it and why.

Before the airdrop:

  • 80%+ of capital was from professional yield farmers (hi!)
  • We were there for the points, not the product
  • Most “users” were farming multiple wallets for airdrop allocation
  • Nobody was actually building on Blast except a few incentivized protocols

After the airdrop disappointed:

  • Capital fled in less than 48 hours to Base and Arbitrum
  • Trading volume dropped 95%
  • Protocol deployments went to zero
  • The “community” evaporated overnight

This wasn’t a community. It was a temporary mercenary workforce.

Why Yield Farmers Don’t Care About Your L2

Here’s the uncomfortable truth from the yield farming side:

I don’t care about:

  • Your chain’s “vision”
  • Long-term sustainability
  • Decentralization claims
  • Technical innovation

I care about:

  • Risk-adjusted returns
  • Exit liquidity
  • Time to next incentive program ending (so I can exit before the rug)

When you pay me 150% APY to provide liquidity on your brand new L2, I’m doing the math:

  • How much can I extract before this collapses?
  • Where are the other yield farmers deploying next?
  • What’s the next incentive program starting?

We are locusts. We descend, extract value, and move to the next field. That’s the game.

The Capital Efficiency Disaster

Here’s what kills me about the $3.4B raised for L2 infrastructure that Steve mentioned:

Let’s say an average L2 raised $50M (probably conservative). They spend:

  • $15M on “ecosystem incentives” (bribing users)
  • $10M on “protocol partnerships” (bribing developers)
  • $10M on team/operations
  • $15M on marketing/“community building”

That’s $25M spent on bribes that generate temporary, fake demand.

Imagine if that $25M funded actual applications on existing L2s instead:

  • 50 teams with $500k each building real products
  • Or 25 teams with $1M building substantial protocols
  • With real users solving real problems

Instead it’s burned on incentive programs that create the illusion of demand for infrastructure nobody actually needed.

DeFi TVL is a Vanity Metric

TVL measures how much capital is deployed, not:

  • How many actual users you have
  • Whether anyone is actually using your protocols
  • If the capital is sticky or mercenary

Base has high TVL because people actually use Base. They’re swapping tokens, interacting with apps, onboarding from Coinbase.

Blast had high TVL because yield farmers parked capital to farm points. The moment farming ended, the capital left.

These are not the same thing.

Where DeFi is Actually Going

The reality is that DeFi liquidity is consolidating around 3-5 L2s with real volume:

Base: Retail users, Coinbase integration, actual non-crypto-native adoption
Arbitrum: Established DeFi ecosystem, GMX and Camelot doing real volume
Optimism: Superchain liquidity, Velodrome as legitimate DEX

Everything else is fighting for scraps.

If you’re a liquidity provider, you want to be where:

  • Volume is high (so you earn fees, not just incentives)
  • Liquidity is deep (so impermanent loss is manageable)
  • The chain won’t shut down in 6 months

New L2s offer none of this. They offer inflated APYs funded by dying token treasuries.

The Future of L2 Liquidity

My prediction for 2027:

3-5 major L2s will have 95%+ of all DeFi liquidity.

Why? Because:

  • Liquidity begets liquidity (deeper pools = lower slippage = more volume)
  • Users will consolidate where apps exist
  • Apps will deploy where users exist
  • Liquidity will flow where apps and users exist

This is a flywheel that new L2s cannot break with incentives.

What This Means for Founders

Steve, you’re making the right call on Base. But here’s my advice from the DeFi side:

Don’t chase incentives from small L2s. They will offer you money to deploy. They will promise “ecosystem support.” They will show you projections about “being early.”

It’s all bait.

Deploy where actual users and liquidity already exist. Your protocol will have:

  • Real volume instead of incentive-farming volume
  • Sticky users instead of mercenary capital
  • Long-term sustainability instead of boom-bust cycles

Emma’s Right, But For Wrong Reasons

Emma said consolidation is good, and she’s right—but not just for developer experience reasons.

Consolidation is necessary for DeFi to actually work.

You cannot have efficient markets with liquidity fragmented across 50 chains. You cannot have price discovery with volume split into tiny pieces. You cannot build sustainable yield when every chain is bribing users with unsustainable emissions.

The L2 wars created this mess. Consolidation is cleaning it up.

The only people who benefit from 50 L2s are:

  • VCs with portfolio diversification strategies
  • Founders who raised big rounds on “the next Base”
  • Yield farmers extracting incentive subsidies (guilty!)

Everyone else—actual users, legitimate protocols, long-term ecosystem health—benefits from consolidation.

The window is closed. Good. Now let’s build something real.

5

As someone who spends every day working on cross-chain infrastructure, let me add the systems perspective: the proliferation of L2s created a fragmentation nightmare that we’re still cleaning up.

The Interoperability Problem Nobody Talks About

Here’s what 50+ L2s actually means from an infrastructure standpoint:

For bridge operators:

  • 50 different chains to integrate and maintain
  • 50 different security models to audit
  • 50 different validator sets to monitor
  • 50 potential attack surfaces

For wallet developers:

  • 50 different RPC endpoints
  • 50 different chain IDs and network configs
  • 50 different gas token mechanics to explain to users
  • 50 different block explorers to link to

For dApp developers:

  • 50 different deployment targets
  • 50 different versions of “the same” asset (wrapped tokens everywhere)
  • 50 different liquidity pools to route through
  • 50 different places users might have funds stuck

This isn’t composability. This is chaos.

The Bridge Security Nightmare

Every new L2 creates new bridge risk. And bridges are the most attacked infrastructure in crypto:

  • Ronin Bridge: $624M (2022)
  • Wormhole: $326M (2022)
  • Nomad: $190M (2022)
  • Poly Network: $611M (2021)

Why do bridges keep getting hacked? Because each L2 adds complexity:

  • Different consensus mechanisms
  • Different finality assumptions
  • Different validator incentive structures
  • Different security guarantees

When you have 50 L2s, you have 50 different trust models. Bridges have to make security assumptions about all of them. This scales quadratically, not linearly.

If you have 5 major L2s, you need ~10 major bridge routes.
If you have 50 L2s, you need ~1,225 potential bridge routes.

Most won’t be built. Users get stuck. Capital gets fragmented.

User Experience is Broken

From a user perspective, multi-chain is a disaster:

Scenario: You want to use a new DeFi protocol.

  1. Protocol is on RandomChain L2 you’ve never heard of
  2. Your funds are on Base
  3. You need to:
    • Find a bridge that supports RandomChain (maybe none exist)
    • Or bridge Base → Ethereum → RandomChain (2 bridges, 2 fees, 2 risks)
    • Wait for finality on multiple chains
    • Hope the bridge has enough liquidity
    • Pay bridge fees + gas on 3 different chains
    • Spend 30-45 minutes and $20-50 in fees

Or you could just use a protocol on Base where your money already is.

This is why new L2s fail. It’s not about technology—it’s about accessible liquidity and users.

Fragmentation Fragments Security

Here’s a controversial take: more L2s = less security overall.

Why?

Security researchers are finite. When you have:

  • 3 major L2s → security researchers focus their attention, bugs get found quickly
  • 50 L2s → attention is fragmented, small L2s get almost no scrutiny

New L2s often fork existing codebases (OP Stack, Arbitrum Orbit, zkSync) and add “custom” modifications. These customizations:

  • Don’t get the same audit attention as the base code
  • Often introduce bugs
  • Create attack surface that researchers miss because they’re focused on bigger chains

Consolidation actually improves security because more eyes focus on fewer codebases.

The Chain Abstraction Dream vs Reality

A lot of people hope “chain abstraction” will solve this—make the underlying L2 invisible to users through intent-based systems.

I’m building this tech. Here’s the reality:

Chain abstraction works best when there are ~3-5 chains to abstract across. Why?

  • Liquidity needs to be deep on each chain for routing to work
  • Solver networks need to maintain capital on each chain
  • Gas costs and bridge fees eat into efficiency
  • More chains = more failure modes

Chain abstraction amplifies the consolidation effect—it makes the dominant L2s more dominant because that’s where solvers park capital and where routing is most efficient.

If you’re on a tiny L2 with thin liquidity, chain abstraction can’t magically find liquidity that doesn’t exist.

Why We Don’t Need 50 General-Purpose L2s

Here’s the uncomfortable truth: differentiation between general-purpose EVM L2s is marginal.

They all offer:

  • ~1-2 second block times
  • ~$0.01-0.10 transaction costs
  • EVM compatibility
  • Similar security models (optimistic or ZK proofs)

The differences are:

  • Marketing narratives (“we’re 10% faster!”)
  • Token distribution strategies
  • Which VCs backed them
  • Incentive programs

These are not meaningful technical moats. They’re go-to-market strategies.

Base won because Coinbase controls distribution.
Arbitrum won because they were first to scale with EVM.
Optimism won because they built OP Stack and created Superchain network effects.

Everyone else is competing on marginal technical improvements that users don’t care about.

Where Innovation Still Happens

Diana and Lisa are right that the window is closed for general-purpose L2s. But here’s where I see opportunity:

Application-specific rollups (L3s):

  • Gaming chains with custom VMs optimized for game state
  • Privacy L3s with built-in ZK circuits for specific use cases
  • Finance-specific chains with built-in oracles and settlement logic

Vertical-specific infrastructure:

  • Supply chain L2s with permissioned validators and compliance built-in
  • Social L2s optimized for high message throughput, low financial value
  • IoT chains with ultra-low latency and specific consensus models

These solve specific problems that general-purpose L2s can’t solve by making different tradeoffs.

But if you’re building “yet another fast, cheap, EVM-compatible L2”—you’re late by 3 years.

The Silver Lining: Interop Gets Easier

Here’s the good news: consolidation around 3-5 L2s makes my job easier.

Building bridges for:

  • Base, Arbitrum, Optimism, Polygon, zkSync

Is infinitely more manageable than building for 50 chains.

We can:

  • Focus security audits on high-value routes
  • Maintain deeper liquidity pools
  • Build better UX because we’re not spreading resources thin
  • Actually achieve meaningful cross-chain composability

The ecosystem is healthier with fewer, better L2s than many mediocre ones.

My Advice for Founders

Steve, you asked if the window is closed. From the infrastructure side:

The window is closed for general-purpose L2s. Don’t build one. Don’t deploy on one that isn’t in the top 5.

The window is open for:

  • Apps on established L2s (Base, Arbitrum, Optimism)
  • Specialized L3s/app-chains with clear differentiation
  • Infrastructure that makes the top L2s more interoperable

Build where the users and liquidity are. Every bridge, wallet, and infrastructure provider is prioritizing the top L2s. You want to be where the infrastructure actually works, not where you need to build everything from scratch.

Final Thought: Consolidation is Infrastructure Maturation

Emma said consolidation is good. Diana said it’s necessary. I’ll go further:

Consolidation is a sign that blockchain infrastructure is maturing.

Every technology goes through this:

  • Early days: explosion of experimentation
  • Middle phase: dozens of competitors
  • Maturity: consolidation around 2-5 dominant players

We’re watching the L2 market mature in real-time.

The winners emerged. The losers are shutting down. The market is optimizing for user experience, liquidity, and security rather than VC narratives and incentive bribes.

This is healthy. This is progress. And yes, the window is closed.

Build apps, not infrastructure. The infrastructure wars are over.