Skip to main content

The Unified Verification Layer Wars: ZK Proof Aggregation Becomes Ethereum's Missing L2 Composability Primitive

· 14 min read
Dora Noda
Dora Noda
Software Engineer

Ethereum has a $40 billion problem hiding in plain sight. By Q3 2026, Layer 2 TVL is projected to surpass mainnet DeFi for the first time — roughly $150 billion on rollups versus $130 billion on L1. The catch: nearly $40 billion of that L2 value sits stranded across more than 60 disconnected networks, each with its own bridge, its own liquidity pool, its own proof system, and its own definition of finality. Ethereum scaled. It just scaled into a hall of mirrors.

The fix everyone now agrees on is some flavor of unified verification. The fight is over whose flavor wins. Polygon AggLayer, Risc Zero's Boundless, Succinct SP1, zkSync Boojum, and the newer ILITY Network are all converging on the same insight from different starting points: if rollups are going to behave like one chain, somebody has to verify all of their proofs in one place. That somebody is now a market — and the market is loud.

The Fragmentation Tax Nobody Wants to Pay​

Vitalik Buterin's February 2026 reframing of L2s as "independent platforms with legitimate economies of their own" was a polite way of admitting that the rollup-centric roadmap had shipped its scaling promise and its fragmentation problem in the same envelope. The Ethereum Foundation's March 23 blog post on L1/L2 division of labor doubled down: L1 should be a "permissionless and maximally resilient global hub for settlement," and L2s should compete on execution and product fit.

That sounds clean on a slide. In production, it means a user holding USDC on Arbitrum who wants to use a vault on Base has to bridge through a third-party messaging layer, accept a finality delay measured in minutes, and trust whatever security assumptions the bridge introduces. Multiply that friction across 60 chains and the experience converges on "worse than a centralized exchange."

The numbers behind the friction explain why: onchain SNARK verification on Ethereum L1 typically costs 250,000 to 500,000 gas per proof at 2026 prices. With each major rollup posting proofs independently, a back-of-envelope estimate from proof-aggregation researchers puts cumulative annual L1 verification spend in the tens of millions of dollars — and this is after L2s have already been optimized for cost. Aggregation, in theory, can compress those costs by an order of magnitude. The Block reported NEBRA's universal aggregator going live on Ethereum mainnet in part because the savings model was compelling enough to attract real volume.

What's emerged in 2026 is a five-way race to be the substrate where that compression happens.

Five Architectures, One Endgame​

Each of the major contenders is solving the same problem from a different architectural seam. Understanding the differences matters because they are not interchangeable — the choice of verification layer cascades through trust assumptions, latency, and composability.

Polygon AggLayer: The Pessimistic Proof Bet​

AggLayer, which moved from concept to mainnet-ready when pessimistic proofs went live in early 2025, takes a defensive approach. The core trick is the pessimistic proof — a ZKP that treats every connected chain as suspicious by default. If Chain A claims it has 100 POL deposited on the shared bridge, AggLayer mathematically forces Chain A to prove it before letting any withdrawal go through. No chain can risk another chain's deposits, even if it gets compromised.

Polygon's AggLayer v1.0 is anticipated in Q2 2026, and Succinct's SP1 — built on Polygon's Plonky3 prover — is already announced as the performance backbone for cross-chain interoperability. It's a tight vertical integration story: shared bridge, shared prover, pessimistic settlement. The trade-off is that the model works best for chains using Polygon CDK or compatible stacks. Pulling in a sovereign rollup with different proof assumptions means rebuilding parts of the trust model.

Risc Zero Boundless: The Open Proof Market​

Boundless launched on mainnet in September 2025 and Risc Zero terminated its hosted proof service that December — forcing all generation through the decentralized network. The architecture is closer to a commodities exchange than a settlement layer. GPU operators bid on proof jobs, generate the proofs, and earn rewards through a mechanism Risc Zero calls Proof of Verifiable Work (PoVW), where each proof carries cryptographic metadata attesting how much computation went into it.

The thesis: by turning ZK proof generation into an open market, Boundless can drive verifiable computation toward the cost of execution. Boundless also explicitly supports competing zkVMs on its roadmap — meaning developers can route to Risc Zero, SP1, or other systems without leaving the marketplace. That positions Boundless less as a single verification layer and more as the liquidity venue where proofs get produced and priced before they hit any specific chain.

Succinct SP1: The Real-Time Proving Wedge​

Succinct's SP1 is the zkVM that turned the benchmarking war into a public sport. The "SP1 Hypercube" announcement in early 2026 went further — claiming real-time Ethereum proving throughput, the holy grail that lets a zkVM prove blocks roughly as fast as Ethereum produces them. If real-time proving becomes table stakes, the bottleneck shifts from prover speed to aggregation and verification — exactly the choke point that AggLayer is trying to monopolize and that Boundless is trying to commoditize.

SP1's role inside AggLayer is the most interesting tell. Polygon picked Succinct as the proving engine, which means the two ecosystems are interlocked at the moment they appear competitive. The deeper question is whether SP1 can stay neutral enough to be the prover-of-record across multiple aggregation layers, or whether it gets absorbed into the AggLayer flag.

zkSync Boojum: The Vertically Integrated Stack​

zkSync's Boojum proving system uses 15 recursive circuits to aggregate proofs internally, with the recently shipped Atlas upgrade letting ZK Stack chains tap directly into Ethereum L1 liquidity. Vitalik publicly endorsed zkSync's work in late 2025 as "underrated and valuable," and BoojumOS's 2026 roadmap targets 30,000 TPS while preserving an Ethereum L1-equivalent dev experience.

Where AggLayer sells horizontal aggregation across heterogeneous chains, Boojum sells vertical aggregation within a single technology family. Pick the ZK Stack, and you get proof aggregation, shared liquidity, and a unified UX out of the box. The cost is sovereignty: the price of joining is committing to Matter Labs' technology decisions.

ILITY Network: The Privacy-First Verification Layer​

ILITY went live with its Alpha Mainnet in January 2026, and the architectural twist is that it's designed for privacy-preserving cross-chain data verification — proving asset ownership, holding history, and onchain behavior across mainnets without revealing wallet addresses. ILITY's pitch is that the "transparency trap" of public Web3 is itself a fragmentation problem: users can't move state across chains without exposing their entire identity to every observer along the path.

The technology stack is a sovereign Layer-1 blockchain with a ZK engine optimized for cross-chain data retrieval. ILITY raised $2 million in seed funding to build out the verification layer, and the strategic positioning is distinctly different from AggLayer's bridge-security model or Boundless's open prover market — ILITY is competing on the privacy axis, where MiCA-driven compliance pressure and institutional treasury demand are converging on confidential cross-chain primitives.

The zkVerify Wildcard​

Outside the five-way frame, zkVerify launched its mainnet on September 30, 2025 as the first L1 purpose-built for zero-knowledge proof verification. The cost reduction claims are dramatic: zkVerify quotes verification costs at less than 1/100 of Ethereum's L1 cost, with millisecond processing times and native support for Groth16, UltraPlonk, RiscZero, ultrahonk, Space and Time, and SP1 proofs.

zkVerify's existence raises the awkward question that none of the major players want to answer: do we even need verification on Ethereum L1? If a dedicated chain can verify proofs 100 times cheaper, the rollup-Ethereum economic relationship gets re-priced. Proofs settle on zkVerify, and Ethereum becomes the security anchor rather than the verification venue. That's a much smaller fee surface for ETH stakers.

The Ethereum Foundation's stated priority — "ETH re-becomes the settlement hub and trust anchor" — is partially a defensive response to exactly this scenario. The Foundation isn't wrong that whoever owns settlement captures the value; the open question is whether Ethereum's social layer can hold settlement gravity if the economics drift to dedicated verification chains.

What the Fragmentation Solvers Are Actually Solving​

It's worth being precise about which fragmentation problem each layer addresses, because the marketing tends to blur them:

  • Bridge fragmentation (AggLayer's wheelhouse): Multiple bridges with incompatible security models. Solved by a shared, ZK-secured bridge.
  • Proof generation fragmentation (Boundless and SP1's wheelhouse): Each rollup runs its own prover, duplicating GPU effort. Solved by a shared proving market.
  • Verification cost fragmentation (zkVerify's wheelhouse): Each rollup pays L1 verification gas independently. Solved by amortized aggregation.
  • Identity and data fragmentation (ILITY's wheelhouse): Cross-chain state queries leak user data. Solved by privacy-preserving verification.
  • Liquidity fragmentation (zkSync Atlas, Ethereum Economic Zone): Capital trapped per rollup. Solved by shared liquidity primitives or unified settlement.

The Ethereum Foundation's Ethereum Economic Zone (EEZ) initiative — launched at EthCC by Gnosis, Zisk, and the EF — is essentially an attempt to bundle several of these solutions under one ideological roof. EEZ promises that smart contracts on connected rollups can call contracts on mainnet or other EEZ chains within a single transaction, eliminating the bridge dependency that has dominated cross-rollup interaction for three years.

The PeerDAS Shift Changes the Stakes​

PeerDAS, shipping as part of the Fusaka upgrade, increases blob capacity from 6 to 48 per block — roughly an order of magnitude more cheap data availability for L2s. Vitalik has framed PeerDAS plus alpha-stage zkEVMs as moving Ethereum into "a fundamentally new and more powerful kind of decentralized network." The transaction-throughput implication is real: PeerDAS-fueled rollups could collectively handle up to 12,000 TPS by year-end.

That throughput surge doesn't solve fragmentation — it makes it worse. More blob space means more rollups can afford to exist, which means more verification surface, more bridge relationships, more proof systems to reconcile. PeerDAS is the supply-side scaling lever; aggregation is the demand-side coordination lever. Both have to work for the L2 ecosystem to feel like a single chain to users.

This is also where Celestia and EigenDA's roles get more sharply defined. Celestia, with roughly 50% DA market share and Matcha doubling block sizes to 128MB in Q1 2026, is a pure data availability play — it doesn't try to verify proofs. EigenDA V2's 100MB/sec throughput sits inside the EigenLayer restaking economy, where Ethereum validators reuse staked ETH to attest to data availability. Neither competes with proof aggregation. Both make aggregation more important by making rollup proliferation cheaper.

Why This Matters for Infrastructure Providers​

For RPC and indexing infrastructure, the unified verification layer race is a forcing function. Today, an indexer querying state across Arbitrum, Base, zkSync, and Optimism has to maintain four independent ingest pipelines, four sets of proof-handling logic, and four trust models. If AggLayer or Boundless or zkVerify wins enough share, the abstraction surface changes: a single proof-source endpoint can verify state across many chains, and the indexer's job becomes routing rather than reconciliation.

The harder bet is that no single layer wins outright in 2026. Polygon's CDK chains will route through AggLayer. zkSync ecosystem chains will route through Boojum. Sovereign rollups using Risc Zero zkVM will route through Boundless. Privacy-conscious institutional flows may route through ILITY. The likely 2026 endgame is a meta-aggregation layer — a router-of-routers that abstracts proof-source heterogeneity behind a single query interface — and that's a product surface neither legacy RPC providers nor specialty indexers currently offer at scale.

For developers building cross-chain applications, the practical implication is to assume verification heterogeneity and design for it. Lock yourself to one aggregation layer and you inherit its trust model and its potential decline. Build toward verification-agnostic state queries and you keep optionality as the layer wars resolve.

The 2026 Resolution Question​

The honest assessment is that none of these layers has decisively won. AggLayer has the strongest vertical integration story but is gated by Polygon-stack adoption. Boundless has the cleanest economic model but depends on enough demand to bootstrap a competitive prover market. SP1 has the best zkVM benchmarks but no neutral position in the aggregation wars. Boojum has the most complete ZK Stack but trades sovereignty for convenience. ILITY has the most differentiated architecture but the smallest ecosystem footprint. zkVerify has the cleanest cost story but threatens Ethereum's settlement value capture.

What changes in 2026 is that the question stops being whether unified verification matters and starts being who controls the layer where it happens. That's a settlement-layer fight, and history says settlement layers tend to consolidate to one or two winners. The next 18 months will tell whether the consolidation pattern repeats, or whether ZK proof aggregation stays plural the way data availability has — fragmented across Celestia, EigenDA, Avail, and PeerDAS even three years into the modular era.

The L2 sprawl problem isn't going away. Somebody is going to charge a toll for stitching it back together. The only question left is whose toll booth.

BlockEden.xyz provides production-grade RPC and indexing infrastructure across the major Layer 1 and Layer 2 networks shaping this race. As verification layers consolidate and cross-rollup composability becomes table stakes, our API marketplace and multi-chain endpoints help developers build applications that stay portable across whichever aggregation layer wins.

Sources​

Share on Twitter