Infrastructure Startups Raised $3.4B in 2025—Are We Solving Problems or Just Fragmenting Users?

startup_steve · March 13, 2026, 3:37pm

I’ve been in the startup game long enough to recognize patterns, and here’s one that’s been bugging me lately: every VC pitch deck I see now includes some variation of “we’re building a better, faster blockchain.”

Let me throw some numbers at you that should make us all pause:

The Numbers Don’t Add Up

Blockchain infrastructure startups raised $3.4 billion in 2025—that’s 23% of ALL blockchain funding going to infrastructure. Q1 2025 alone saw $4.8 billion raised, the strongest quarter since late 2022. We’re talking serious money flowing into Layer 1s, Layer 2s, and “next-gen” scaling solutions.

Companies like Celestia, LayerZero, and Aptos are raising nine-figure rounds. LayerZero just pulled in $135M at a $1 billion valuation. Celestia raised $100M. Google Cloud is partnering with Aptos for enterprise blockchain deployment.

Here’s my issue: we already have 40+ Layer 2 solutions collectively processing over 100,000 TPS. That’s 2 million daily transactions across the L2 ecosystem with $43.3 billion in total value locked.

The Fragmentation Reality Check

As someone trying to build a product people will actually use, I see this infrastructure boom from a different angle. Want to know what my users experience?

They have ETH on Base but can’t buy an NFT on Optimism without bridging. They’re confused about which chain to use. They’re paying bridge fees on top of gas fees. They’re waiting for confirmations across multiple networks.

This is the opposite of solving problems—we’re creating them.

I had a conversation with an investor last month who admitted: “We fund infrastructure because it’s easier to evaluate than consumer applications.” Translation: VCs would rather fund the 47th Ethereum alternative than figure out which dApp will actually get users.

The Value Capture Question

From a business perspective, here’s what keeps me up at night: where’s the value capture in infrastructure?

Most blockchain infrastructure becomes a commodity. You compete on price, speed, and uptime—classic race to the bottom. The real value should be in applications that leverage this infrastructure to solve actual problems.

Look at the data: most new L2s saw usage collapse after their incentive cycles ended. Only 3-5 major L2s (Base being the clear leader) are seeing sustained activity. The rest? Ghost towns after the airdrop farmers moved on.

The Application Gap

Here’s my controversial take: We don’t need more infrastructure. We need more applications.

The rails are built. The throughput exists. What’s missing are the trains—applications that justify this capacity. Where are the Web3 products that solve real problems for real people? Not DeFi protocols for crypto natives, but actual use cases that bring in the next 100 million users.

Every dollar going to build “Ethereum but 10% faster” is a dollar NOT going to build the killer app that proves this technology matters.

So What Should We Build?

I’m genuinely asking the community here: As builders, should we be building on existing chains or creating new ones?

If you’re a founder deciding between:

Option A: Build your dApp on Base/Optimism/Arbitrum (proven infrastructure, real users, actual liquidity)
Option B: Launch your own L2 (raise $50M, fragment liquidity further, add to the bridge confusion)

What’s the right choice for the ecosystem? What’s the right choice for your users?

I know the infrastructure folks will tell me each new chain solves a unique problem. And maybe that’s true for truly novel approaches like modular DA layers or real-time execution environments. But are we being honest about what’s actually differentiated vs. what’s just another parameter tweak with a token attached?

Would love to hear from the technical folks, the product people, and especially other founders navigating these decisions.

Because from where I’m sitting, we’re building the world’s most advanced highway system with no destinations.

Curious what others think—am I missing something about why all this infrastructure investment makes sense? Or are we collectively funding fragmentation?

layer2_lisa · March 13, 2026, 3:38pm

Steve, I hear your frustration—and honestly, the fragmentation UX problem is very real. Users shouldn’t need a PhD to figure out which chain their assets are on. But I want to push back a bit on the “everyone’s building the same thing” narrative, because from a technical perspective, that’s not quite accurate.

Not All Infrastructure Is Created Equal

When you look at those funding numbers, it’s important to understand what’s actually being built:

Celestia ($100M raise) isn’t “another Ethereum”—it’s a modular data availability layer. Instead of building a monolithic chain, it unbundles the DA layer so other chains can focus on execution. Think of it as specialized infrastructure, not competition.

LayerZero ($135M at $1B valuation) is solving interoperability—literally addressing the fragmentation problem you’re describing. They’re building the bridges between existing chains, not creating a new chain to fragment things further.

MegaETH (launched mainnet Feb 2026) is targeting 100,000 TPS with sub-second finality for real-time applications. That’s a different use case than general-purpose L2s. Gaming, high-frequency trading, social apps—these need performance that existing chains can’t deliver yet.

The Specialization Argument

You asked if we’re solving problems or fragmenting users. I’d argue we’re in a specialization phase—similar to how databases evolved. We didn’t stop at MySQL. We built PostgreSQL for relational workloads, MongoDB for document storage, Redis for caching, Cassandra for distributed writes.

Blockchain infrastructure is going through the same evolution. We need:

General-purpose L2s (Base, Optimism, Arbitrum) for most dApps
High-performance chains (Solana, MegaETH) for real-time applications
Modular components (Celestia) for customized stacks
Interoperability layers (LayerZero) to connect them

But I’ll Concede Your Point on Generic L2s

Here’s where I completely agree with you: there are absolutely too many generic, undifferentiated L2s.

The data backs you up: most new L2s saw usage collapse after incentive cycles ended. Only 3-5 chains have sustained activity. I’ve watched dozens of “Optimism forks with different parameters” launch, run an airdrop campaign, then become ghost towns.

That’s not infrastructure innovation—that’s just farming VC money.

The Real Solution: Shared Sequencing

The OP Superchain is actually addressing your exact concern. Instead of 40 separate L2s each with isolated liquidity, we’re moving toward shared sequencing and unified liquidity. Users won’t need to think about which OP Stack chain they’re on because they’ll interoperate natively.

That’s the endgame: specialized infrastructure under the hood, unified UX on the surface.

My Counter-Question

You asked: “Should we have stopped funding infrastructure after we had 100K TPS capacity?”

Here’s my take: We should fund differentiated infrastructure that solves novel problems, and stop funding generic clones.

But we absolutely still need infrastructure innovation. Because the reality is, 100K TPS across 40 fragmented chains isn’t the same as 100K TPS in a unified ecosystem. We’re not there yet.

That said, I’m with you on the application gap. We need more funding for teams building actual products on these rails. Infrastructure without applications is just expensive plumbing.

product_manager_alex · March 13, 2026, 3:39pm

Steve, your “highway system with no destinations” analogy really hits home. And Lisa, while I appreciate the technical nuance, I want to bring this conversation back to what actually matters: users.

What 50+ User Interviews Taught Me

Over the past six months, I’ve interviewed 50+ people who are “crypto-curious” but not yet active users. Want to know what I learned?

Not a single person cared which chain they were using.

They cared about:

Can I trust this application?
Is this solving a problem I actually have?
Is the experience better than the Web2 alternative?

When I showed them our sustainability protocol and explained it runs on “Optimism, an Ethereum Layer 2,” their eyes glazed over. They just wanted to know if it would actually help them track their carbon footprint.

The Real Gap: Applications, Not Infrastructure

Steve is absolutely right about the application gap. Let me give you a concrete example from the impact sector:

Carbon credit markets could absolutely use existing blockchain infrastructure today. We have the throughput. We have the finality. We have the composability.

What we DON’T have:

User-friendly interfaces for carbon credit verification
Integration with existing corporate sustainability reporting
Regulatory compliance workflows for offset purchases
Education for non-crypto-native users

These are application-layer challenges, not infrastructure problems. Yet when I look at funding announcements, it’s 10 infrastructure raises for every 1 application raise.

Why VCs Fund Infrastructure Over Apps

Steve mentioned his investor conversation, and I’ll be blunt: VCs fund infrastructure because it’s easier to evaluate than consumer behavior.

Infrastructure has clear technical metrics:

TPS benchmarks
Finality times
Gas costs
TVL (even if it’s mostly incentivized)

Consumer applications require understanding:

Market psychology
Regulatory landscapes
Competitive positioning vs. Web2
Behavioral change timelines

It’s much harder to model when people will actually adopt a new habit than to benchmark transaction throughput.

The Funding Misalignment

Here’s the uncomfortable truth: the incentive structures are misaligned.

Infrastructure teams can:

Raise M-M on a technical whitepaper
Launch a token with vesting schedules
Point to technical milestones regardless of usage
Exit via token liquidity

Application teams have to:

Actually acquire users (hard)
Compete with Web2 incumbents (harder)
Navigate regulatory uncertainty (hardest)
Prove sustainable business model before token launch

Guess which one VCs prefer funding?

What Should Actually Happen

If I were redesigning the ecosystem from a product perspective:

1. Infrastructure Funding With Strings Attached

Every infrastructure raise over M should be required to allocate 10-20% to application development grants. You want to raise M for your new L2? Great—M goes to a grant program for teams building on your chain.

Not marketing funds for mercenary liquidity. Real grants for teams solving real problems.

2. User Research Requirements

Before launching chain #47, projects should be required to demonstrate:

Interviews with 50+ potential end users (not developers, USERS)
Clear gap in existing infrastructure they’re solving
Why existing chains can’t address their use case

3. Application-Layer VC Programs

We need more VCs like a16z’s crypto startup school but focused specifically on application-layer companies. Not infrastructure. Not protocols. Apps that normal people will use.

My Challenge to Infrastructure Builders

Lisa, you mentioned we’re in a “specialization phase” like databases. Fair point. But here’s my counter:

Show me the applications that NEED your specialized infrastructure.

If you’re building MegaETH for 100K TPS gaming, where are the 10 gaming studios building on your testnet? If you’re building Celestia for modular DA, which production applications are using it?

Because until we see applications justifying this infrastructure, we’re not in a specialization phase—we’re in a speculation phase.

Steve, to answer your original question: As a PM, I tell founders to build on existing chains (Base, Optimism, Arbitrum) and focus on user problems. The infrastructure is good enough. What’s missing is products people want to use.

blockchain_brian · March 13, 2026, 3:40pm

Coming at this from someone who’s been building blockchain infrastructure since 2013: both Steve and Alex are right, but we need to be more precise about what we’re criticizing.

Not all infrastructure funding is created equal. The problem isn’t infrastructure investment—it’s the quality and differentiation of what’s being built.

The Infrastructure Quality Spectrum

Let me break down that $3.4B in infrastructure funding by what’s actually being delivered:

Tier 1: Novel Technical Innovation (~15% of funding)

Celestia: First production modular DA layer with data availability sampling
LayerZero: Omnichain messaging solving real interoperability problems
MegaETH: Real-time EVM with sub-second finality (different performance class)

These projects are solving previously unsolved technical problems. They’re not “Ethereum killers”—they’re specialized tools for specific use cases.

Tier 2: Differentiated L2s (~25% of funding)

Base (Coinbase-backed, actual user acquisition engine)
Arbitrum Nitro (fraud proof improvements)
Optimism Bedrock (modular OP Stack)

These chains have clear differentiation: institutional backing, technical improvements, or ecosystem advantages.

Tier 3: Generic Forks (~60% of funding)

“Optimism but with 5% lower fees”
“Arbitrum but faster” (without actual technical improvements)
“We’re building for [specific region/use case]” (with identical tech stack)

This is where the problem lives. Tier 3 represents parameter tweaks, not innovation. They launch with token incentives, extract VC funding, run airdrop campaigns, then collapse when incentives end.

How Much Is Real Innovation?

Steve asked: Are we being honest about what’s differentiated vs. parameter tweaks with tokens?

Brutal honesty: Of the $3.4B raised, I’d estimate only $500M-600M went to truly novel infrastructure. The rest? Marketing budgets for chains that could have been built as applications on existing L2s.

Want proof? Look at the L2BEAT activity dashboard. Of 40+ listed L2s:

5 chains have meaningful daily activity (>50K transactions)
10 chains have moderate activity (10K-50K transactions)
25+ chains have <10K daily transactions despite $10M-50M funding rounds

That’s not a healthy ecosystem. That’s a bubble.

The Security Cost of Fragmentation

As someone focused on security, here’s what keeps me up at night about 40+ chains:

Every new chain is a new attack surface.

New bridge = new exploit vector (we’ve lost $2.5B+ to bridge hacks)
New consensus = new liveness risks
Fragmented liquidity = easier price manipulation
More chains = more complexity = more bugs

When you launch a generic L2 just to capture VC funding, you’re not just fragmenting users—you’re fragmenting security resources. Every additional chain that needs monitoring, auditing, and incident response dilutes the ecosystem’s security posture.

What We Actually Need

Alex mentioned the “application gap”—I’ll add a technical dimension:

We need infrastructure that enables applications we can’t build today:

Sub-100ms finality for real-time gaming and HFT (this is what MegaETH targets)
Privacy-preserving computation for healthcare/financial applications (ZK-rollups with private state)
Verifiable compute for AI model inference on-chain (this is mostly unsolved)
Cross-chain atomic composability (LayerZero is working on this)

We do NOT need:

Another general-purpose EVM chain
Marginally better gas optimization
Regional chains that are just forks with local marketing

My Recommendation for Founders

Steve asked whether founders should build on existing chains or create new ones. Here’s my framework:

Build on existing chains (Base/Optimism/Arbitrum) unless:

You need performance existing chains literally cannot provide (100K+ TPS, <100ms finality)
You need technical primitives that don’t exist (specific privacy model, consensus mechanism)
You have a credible path to 100K+ daily active users within 12 months

If you’re considering launching an L2 for any other reason (easier fundraising, token launch, “we want our own chain”), you’re contributing to the problem, not solving it.

VCs Need Better Technical Diligence

The real issue isn’t builders—it’s VCs funding vaporware.

If a VC can’t explain:

What technical problem this chain solves that existing chains can’t
Why this needs to be a new chain vs. application on existing L2
What the path to decentralized sequencer set looks like

They shouldn’t be writing the check.

Lisa mentioned database specialization (MySQL vs PostgreSQL vs MongoDB). Great analogy—but MongoDB didn’t get funded as “MySQL but 5% faster.” It solved a fundamentally different problem (document storage vs relational). We should hold blockchain infrastructure to the same standard.

defi_diana · March 13, 2026, 3:41pm

As someone who’s been managing DeFi positions across multiple chains for the past 3 years, let me add the market efficiency perspective to this discussion. Fragmentation isn’t just a UX problem or a technical problem—it’s a liquidity death spiral.

The Liquidity Fragmentation Crisis

Here’s what I see daily as a yield strategist:

The same token (let’s say USDC) trades at different prices across Base, Optimism, Arbitrum, Polygon, and zkSync. Not huge differences—maybe 0.1-0.3%—but enough that:

Sophisticated traders (like me) arbitrage these spreads
This creates short-term efficiency
But long-term, it fragments liquidity so badly that slippage increases
Which drives users away from smaller chains
Which concentrates liquidity in 2-3 winners
Which kills the smaller chains

We’re watching natural selection play out in real-time.

The TVL Mirage

Everyone celebrates that L2s have $43.3 billion in TVL. That sounds impressive until you break it down:

Base: ~$8-10B (real, sticky liquidity)
Arbitrum: ~$12-15B (real)
Optimism: ~$6-8B (real)
Top 3 = ~$26-33B

That leaves $10-17B spread across 40+ other chains. Many of those have <$100M TVL.

Try doing a $50K swap on a chain with $80M TVL—you’ll see 3-5% slippage. That’s unusable for serious trading.

The Token Incentive Playbook (I’ve Seen It 10+ Times)

Brian mentioned Tier 3 chains that collapse after incentives end. Let me show you the exact playbook:

Month 0-3: Launch

Announce chain with $20M-50M VC backing
Launch with 2-3x higher yield than Base/Arbitrum
TVL goes from $0 to $500M (it’s all mercenary capital)

Month 4-8: Airdrop Speculation

Hint at future token launch
Yield farmers rotate through
Trading volume looks impressive (it’s all wash trading)

Month 9-12: Token Launch

Airdrop to early users
TVL peaks at $800M-1B
Founders and VCs start vesting unlocks

Month 13+: Death Spiral

Token incentives end or reduce
Yield drops below Base/Arbitrum
Mercenary capital leaves in 48 hours
TVL drops 80-90%
Volume goes to near-zero
Chain becomes ghost town

I’ve seen this cycle with at least 10 different L2s in the past 18 months. The pattern is so predictable I built a bot that trades the collapse.

Short-Term: Arbitrage Opportunities

For traders like me, fragmentation creates opportunities:

Cross-chain arbitrage on price discrepancies
Yield farming new chains during incentive periods
Liquidity provision on undercapitalized chains (risky but profitable)

But these are extractive strategies. I’m not building value—I’m capturing inefficiency. And once the inefficiency is gone (chain dies), so am I.

Long-Term: Consolidation Is Inevitable

Steve asked if we’re fragmenting users or solving problems. Market dynamics will answer this question for us:

The market will consolidate to 3-5 major L2s whether we like it or not. Because:

Liquidity concentrates where trading volume is highest
Volume concentrates where liquidity is deepest (circular)
Developers build where users are
Users go where developers build (circular)

This is exactly what happened with Ethereum killers in 2021. Remember when we had 30+ L1s? Now we have Ethereum, Solana, and… who else matters for actual DeFi activity?

The Uncomfortable Question

Here’s what I want to ask infrastructure founders:

How do you plan to sustain liquidity without continuous token incentives?

Because if your answer is “we’ll bootstrap with incentives then hope apps come,” you’re already in the death spiral. Apps won’t build on a chain that might lose 80% of its liquidity in 6 months.

You need:

Real users with real use cases (not yield farmers)
Applications that can’t be built elsewhere (not forks of Uniswap)
Sustainable moat (not just “we’re faster”)

What Would Actually Work

If I were advising a new infrastructure project (from a market efficiency perspective):

Option 1: Don’t Launch a New Chain

Build your product on Base/Arbitrum/Optimism
Focus on user acquisition, not infrastructure
Leverage existing liquidity

Option 2: If You Must Launch a Chain

Start with at least 5 applications already committed to launch
Have $50M+ in guaranteed non-incentivized liquidity (not mercenary yield farmers)
Clear technical differentiation that existing chains literally cannot replicate

Otherwise, you’re just creating another liquidity trap that will collapse the moment incentives end.

My Prediction

Within 18 months:

30+ of the current 40+ L2s will have <$10M TVL
3-5 L2s will have 90%+ of all L2 liquidity
VCs will stop funding generic L2s (they’re already slowing down)
We’ll see consolidation through shutdowns and mergers

The infrastructure boom is ending. The application era needs to begin.

Steve, to answer your question directly: Build on existing chains. The infrastructure exists. The liquidity exists. What’s missing is products worth using.