Ethereum is undergoing its most consequential architectural transformation since The Merge in September 2022. Over the course of twelve months — from Pectra's activation in May 2025 through the forthcoming Glamsterdam hard fork in H1 2026 — the network is executing a multi-vector scaling strategy...
"Now that ZKEVMs are at alpha stage — production-quality performance, remaining work is safety — and PeerDAS is live on mainnet, it's time to talk more about what this combination means for Ethereum. These are not minor improvements; they are shifting Ethereum into being a fundamentally different system." — Vitalik Buterin, February 2026
Ethereum L2 TVL: ~$47 billion[^1] | Base 2025 Revenue: $82.6 million[^2] | Glamsterdam Gas Limit Target: 200 million (up from 60M)[^3] | zkEVM Proving Time Reduction: 16 minutes → 16 seconds (60×)[^4] | Pectra Blob Capacity: 6 target / 9 max (doubled from Dencun)[^5] | L2 Smart Contracts Deployed (Q4 2025): 8.7 million (record)[^1] | Base DeFi TVL: $4.3 billion (~46.6% of all L2 DeFi TVL)[^2] | Ethereum DeFi Dominance: ~68% of total DeFi TVL[^6]
Ethereum is undergoing its most consequential architectural transformation since The Merge in September 2022. Over the course of twelve months — from Pectra's activation in May 2025 through the forthcoming Glamsterdam hard fork in H1 2026 — the network is executing a multi-vector scaling strategy that simultaneously increases Layer 1 throughput by 3.3×, integrates zero-knowledge proofs directly into the consensus layer, and doubles the data availability bandwidth for Layer 2 rollups. The cumulative effect is not incremental improvement. It is a structural reimagining of how the world's largest smart contract platform processes, verifies, and settles transactions.
The numbers substantiate the ambition. Pectra doubled Ethereum's blob capacity from 3 to 6 targets per block, with a January 2026 BPO update pushing that further to 14 blobs — laying the groundwork for the network's stated goal of exceeding 100,000 TPS across its Layer 2 ecosystem[^5]. Zero-knowledge EVMs have compressed proving times from 16 minutes to 16 seconds, a 60-fold improvement that makes real-time ZK verification operationally viable for the first time[^4]. And Glamsterdam — Ethereum's next major hard fork — will introduce parallel transaction execution via Block-level Access Lists (EIP-7928) and increase the gas limit from 60 million to 200 million, fundamentally altering the economic calculus for on-chain computation[^3].
Meanwhile, the Layer 2 landscape is consolidating with unprecedented speed. Base — Coinbase's OP Stack rollup — has emerged as the dominant L2, commanding 46.6% of all Layer 2 DeFi TVL, generating $82.6 million in 2025 revenue (a 30× year-over-year increase), and processing over 1.09 million daily active addresses[^2]. The broader L2 ecosystem hit $47 billion in total value locked and deployed a record 8.7 million smart contracts in Q4 2025 — a 45% surge from the previous quarterly peak set during the 2021 bull market[^1]. Yet beneath these headline figures, a stark concentration dynamic is emerging: most new Layer 2s see usage collapse after incentive cycles end, with meaningful activity consolidating around Base, Arbitrum, and Optimism.
This report provides a structural analysis of Ethereum's scaling renaissance — the technical upgrades driving it, the Layer 2 competitive dynamics reshaping it, and the strategic implications for developers, investors, and institutions positioning for the next phase of on-chain infrastructure.
Ethereum's development roadmap has accelerated into an unprecedented cadence of major protocol upgrades, each building on the last to create a compounding effect on network capability.
Pectra doubled Ethereum's blob target from 3 to 6 per block — and raised the maximum from 6 to 9 — effectively doubling the expected blob throughput available to Layer 2 rollups[^5]. This single change increased Ethereum's transaction capacity to approximately 420 TPS on the base layer, while dramatically reducing the cost of posting rollup data. The practical effect was immediate: L2 transaction fees dropped to fractions of a cent, removing the economic barrier that had constrained rollup adoption through 2024.
Beyond blobs, Pectra introduced smart account capabilities via EIP-7702, enabling externally owned accounts to temporarily function as smart contracts — a change that has already spawned new wallet designs, session key implementations, and gasless transaction flows[^5].
Fusaka activated PeerDAS (Peer Data Availability Sampling), a mechanism that allows validators to verify rollup blob data without downloading the full dataset[^7]. This is not a minor optimization. PeerDAS fundamentally changes the bandwidth requirements for Ethereum validators, enabling the network to support significantly higher blob counts without proportionally increasing hardware demands.
The January 2026 BPO update extended Fusaka's impact by raising the blob target from 6 to 14 per block — more than doubling capacity again and pushing Ethereum toward its stated ambition of supporting 100,000+ TPS across its L2 ecosystem[^7].
Glamsterdam represents Ethereum's most ambitious execution-layer upgrade since The Merge. Two headliner EIPs define the fork:
The combined effect targets a gas limit increase from 60 million to 200 million — a 3.3× expansion that will dramatically increase the computational capacity of Ethereum's base layer[^3].
Looking further ahead, the Hegota upgrade — targeted for late 2026 — will introduce Verkle Trees, a new cryptographic data structure designed to make Ethereum nodes "stateless"[^8]. Statelessness eliminates the need for validators to store the full state of the network locally, reducing hardware requirements and enabling a broader, more decentralized validator set. Combined with parallel execution, Verkle Trees represent the final architectural piece needed for Ethereum to scale its base layer without compromising decentralization.
At the heart of Glamsterdam sits a fundamental reimagining of how Ethereum processes transactions. Since launch, Ethereum has executed transactions sequentially — each transaction must complete before the next begins. This design choice prioritized simplicity and determinism but created an inherent throughput ceiling that no amount of hardware optimization could overcome.
EIP-7928 changes this by introducing Block-level Access Lists (BALs), which require transaction senders to declare which state slots their transactions will read and write[^3]. Armed with this information, execution clients can identify independent transaction groups — transactions that touch entirely different parts of the state — and process them in parallel across multiple CPU cores.
The implications are profound. Modern server hardware typically features 16-64 CPU cores, yet Ethereum validators currently utilize a single core for transaction execution. Parallel processing unlocks the remaining hardware capacity, enabling throughput increases that scale with hardware advancement rather than protocol-level changes.
The gas limit increase from 60 million to 200 million is the economic expression of this technical capability[^3]. A higher gas limit means more computation per block, but only if that computation can be processed within the 12-second slot time. Parallel execution makes this feasible by distributing the computational load across available hardware resources.
However, the transition introduces new complexity. Not all transactions are parallelizable — any transactions that read or write overlapping state must still be sequenced. The effectiveness of parallel execution depends heavily on the accuracy and completeness of access list declarations, creating new challenges for wallet developers, smart contract designers, and protocol researchers. Glamsterdam's devnet and testnet phases will be critical for validating these mechanics before mainnet activation[^3].
Ethereum's zero-knowledge roadmap has crossed a critical threshold. The L1-zkEVM initiative — detailed in a 2026 roadmap published by the Ethereum Foundation — centers on integrating ZK proofs directly into Ethereum's consensus layer, enabling validators to verify block correctness without re-executing every transaction[^4].
The performance improvements are staggering. In 2024, generating a zero-knowledge proof for a full Ethereum block required approximately 16 minutes of compute time — far too slow for real-time verification. By early 2026, optimized provers have compressed this to 16 seconds, a 60-fold improvement driven by advances in proof systems (Plonky3, Halo2), hardware acceleration (GPU and FPGA-based proving), and circuit optimization[^4].
The 2026 roadmap identifies six workstreams for L1-zkEVM integration[^4]:
The strategic significance is transformative. Once ZK verification is integrated into consensus, Ethereum validators no longer need to re-execute every transaction to confirm block validity. They simply verify a succinct proof — a computation that takes milliseconds regardless of block complexity. This decouples validator hardware requirements from network throughput, enabling Ethereum to increase execution capacity without concentrating validation among well-resourced operators.
The Layer 2 ecosystem has entered a consolidation phase that challenges the prevailing narrative of a diverse, multi-chain scaling future.
Base's 2025 performance was nothing short of extraordinary. Coinbase's OP Stack rollup generated $82.6 million in revenue — a 30× increase year-over-year — while commanding $4.3 billion in DeFi TVL and processing 1.09 million daily active addresses[^2]. Base accounted for approximately 46.6% of all Layer 2 DeFi TVL, making it the single most used Ethereum scaling solution by a wide margin[^2].
The source of Base's dominance is not technical superiority — it runs the same OP Stack as dozens of other rollups. It is distribution. Coinbase's 9.3 million monthly active trading users provide Base with a built-in onboarding funnel that no competing L2 can replicate[^2]. The compliance profile — operating under Coinbase's regulatory framework — further positions Base as the default choice for institutional users and TradFi applications seeking Ethereum-compatible infrastructure.
Arbitrum remains the most developer-rich L2 ecosystem, with 2,374 active developers and 189,957 GitHub commits. Optimism follows with 3,044 developers and 172,954 commits[^1]. Both networks maintain meaningful DeFi activity and serve as foundational infrastructure for Ethereum's rollup-centric roadmap. However, neither has matched Base's user growth trajectory, suggesting that developer activity alone is insufficient to win the L2 market — distribution and user acquisition matter more.
Perhaps the most significant finding from 2025's L2 data is the fate of the long tail. Q4 2025 saw a record 8.7 million smart contracts deployed across L2s — a 45% surge from 2021's previous peak[^1]. Yet this deployment surge masks a concentration dynamic: most new L2s see usage collapse after incentive cycles end. The data suggests that the Layer 2 market is converging toward an oligopoly structure where 3-5 dominant networks capture the vast majority of users, liquidity, and developer attention, while dozens of niche and application-specific rollups compete for the remainder.
Ethereum's data availability layer — the mechanism by which rollups post transaction data to Layer 1 — has undergone three major expansions in twelve months:
| Upgrade | Date | Blob Target / Block | Cumulative DA Capacity | |---------|------|---------------------|----------------------| | Dencun (EIP-4844) | March 2024 | 3 target / 6 max | Baseline | | Pectra | May 2025 | 6 target / 9 max | ~2× Dencun | | Fusaka + BPO | Dec 2025 – Jan 2026 | 14 target | ~4.7× Dencun |
This 4.7× expansion in data availability bandwidth — accomplished in under two years — directly translates to lower rollup posting costs and higher aggregate throughput[^5][^7]. With PeerDAS enabling validators to verify blob data without full downloads, the network can support further blob count increases without proportionally inflating hardware requirements for validators.
The 100,000 TPS target is not a theoretical maximum — it is the operational goal that Ethereum's roadmap is engineering toward[^7]. With 14 blobs per block, each blob containing ~128 KB of data, and assuming efficient rollup compression ratios, Ethereum can support approximately 50,000-100,000 L2 transactions per second at current throughput levels. Further blob increases planned for 2027 and beyond aim to push this beyond 1 million TPS.
The Ethereum-Solana competitive dynamic has matured from a simplistic "speed vs. security" narrative into a more nuanced structural comparison.
Ethereum's Position: $300 billion in total ecosystem TVL (mainnet + L2s), 4,000+ dApps, 68% of total DeFi TVL dominance, and the deepest institutional capital pool in crypto[^6][^9]. Transaction costs on L2s have fallen below $0.01, effectively neutralizing Solana's historic fee advantage. The Glamsterdam upgrade's parallel execution directly addresses Ethereum's throughput limitations at the base layer.
Solana's Position: 2,600+ TPS with sub-second finality, 62 million daily transactions, $9.2 billion DeFi TVL, and a rapidly growing developer community expanding at 300% annually[^9]. Solana's Alpenglow consensus upgrade — introducing components that can finalize blocks in 100-150 milliseconds — aims to extend its performance lead[^10]. However, Solana's DeFi liquidity remains nearly 50× smaller than Ethereum's combined ecosystem.
The emerging reality is that Ethereum and Solana are not competing for the same market. Ethereum — with its L2 ecosystem, institutional infrastructure (BlackRock's BUIDL, JPMorgan's Kinexys), and regulatory integration — is becoming the settlement and institutional layer. Solana — with its raw performance, lower developer friction, and retail-friendly UX — is capturing consumer applications, trading, and emerging market use cases. The two ecosystems are diverging rather than converging.
The February 2026 market correction — which saw Bitcoin retrace from $126K to approximately $60K — provided an unexpected stress test for Ethereum's DeFi infrastructure[^6]. The results were telling:
This resilience validates the thesis that DeFi has matured from a speculative playground into structural financial infrastructure with institutional backing that provides a liquidity floor during market stress.
Despite the momentum, several material risks warrant attention:
1. Parallel Execution Complexity: EIP-7928's access list mechanism introduces new attack surfaces and edge cases. Incorrect access list declarations could cause transaction failures, state conflicts, or MEV extraction opportunities that are difficult to predict before mainnet deployment.
2. L2 Centralization: Base's dominance raises legitimate decentralization concerns. A single L2 capturing 46.6% of DeFi TVL — operated by a publicly traded company (Coinbase) subject to regulatory and political pressure — creates a concentration risk that contradicts Ethereum's decentralization ethos.
3. zkEVM Security Assumptions: The 128-bit security target for ZK provers is ambitious. Any vulnerability in proof systems could undermine the integrity of ZK-verified blocks, with consequences that scale with adoption.
4. Value Accrual to ETH: The persistent concern that Layer 2 activity does not sufficiently accrue value to the ETH base layer remains unresolved. As L2s capture more transaction fees and MEV, the economic relationship between Ethereum L1 and its rollup ecosystem requires ongoing monitoring.
5. Solana's Execution Risk: Solana's Alpenglow upgrade, if successfully deployed, could widen the performance gap with Ethereum's base layer enough to attract latency-sensitive applications that would otherwise build on L2s.
Ethereum is executing a four-upgrade cascade (Pectra → Fusaka → Glamsterdam → Hegota) that will increase gas limits by 3.3×, enable parallel execution, integrate zkEVM proofs into consensus, and introduce stateless validation via Verkle Trees — all within 18 months.
zkEVM proving times have improved 60× (16 minutes to 16 seconds), making real-time ZK block verification operationally viable and setting the stage for validators to verify blocks without re-execution.
Base has won the Layer 2 war — for now. With $82.6M in 2025 revenue, 46.6% of L2 DeFi TVL, and Coinbase's 9.3M monthly active users as a distribution moat, Base is the de facto standard L2. The long tail of Layer 2s is collapsing.
Data availability bandwidth has expanded 4.7× in under two years (3 blobs → 14 blobs per block), with PeerDAS enabling further increases without proportional hardware cost growth.
DeFi demonstrated structural resilience during Bitcoin's 50%+ correction, with TVL declining only ~12% and 1.6M ETH being actively deployed into protocols during the drawdown — evidence of institutional capital providing a liquidity floor.
Ethereum and Solana are diverging, not converging. Ethereum is becoming the institutional settlement layer; Solana is capturing consumer and high-frequency use cases. The "flippening" narrative is being replaced by a "parallel economies" thesis.
The Ethereum scaling renaissance is not a single event but a compounding cascade of technical, economic, and competitive forces converging in 2026. Glamsterdam's parallel execution and 200M gas limit represent the most significant base-layer upgrade since The Merge. The zkEVM integration transforms the verification model from "re-execute everything" to "verify a proof" — a paradigm shift with profound implications for validator decentralization and network throughput. And the 4.7× expansion in data availability bandwidth is enabling Layer 2s to process transactions at fractions of a cent, removing the cost barrier that constrained adoption.
Yet the most significant development may be the Layer 2 consolidation. The data from 2025 makes clear that the "thousand rollups" thesis was overly optimistic. The market is converging toward an oligopoly dominated by Base, Arbitrum, and Optimism — with Base's distribution advantage via Coinbase making it the likely winner of the consumer L2 market. For developers, investors, and institutions, the strategic implication is clear: the question is no longer "which L2 should I build on?" but "how do I position within the Base-Arbitrum-Optimism triopoly before the window closes?"
Ethereum is not just scaling. It is industrializing. And the organizations that recognize this shift — from experimental blockchain to institutional-grade financial infrastructure — will capture the value that this transformation creates.
[^1]: The Block, "2026 Layer 2 Outlook" — https://www.theblock.co/post/383329/2026-layer-2-outlook [^2]: RootData / Bitget, "Base's 2025 Report Card: Revenue Grows 30 Times" — https://www.rootdata.com/news/480230 [^3]: GetBlock, "Ethereum Glamsterdam — What We Know So Far" — https://getblock.io/blog/ethereum-glamsterdam--what-we-know-so-far-about-the-hard-fork/ [^4]: Ethereum Magicians, "L1-zkEVM Roadmap 2026" — https://ethereum-magicians.org/t/l1-zkevm-roadmap-2026-integrating-zkevm-proofs-into-ethereums-core-protocol/27595 [^5]: Coinbase, "The Ultimate Guide to Ethereum's Pectra Upgrade" — https://www.coinbase.com/blog/the-ultimate-guide-to-ethereums-pectra-upgrade [^6]: CoinDesk, "DeFi's Quiet Strength: TVL Holds as Market Selloff Tests Traders" — https://www.coindesk.com/business/2026/02/03/defi-s-quiet-strength-tvl-holds-as-market-selloff-tests-traders [^7]: Fidelity Digital Assets, "The Fusaka Upgrade: Scaling Meets Value Accrual" — https://www.fidelitydigitalassets.com/research-and-insights/fusaka-upgrade-scaling-meets-value-accrual [^8]: CoinDesk, "Ethereum's 'Hegota' Upgrade Slated for Late 2026" — https://www.coindesk.com/tech/2025/12/28/ethereum-s-hegota-upgrade-slated-for-late-2026-as-devs-accelerate-roadmap [^9]: OSL, "2026 DeFi: Ethereum vs Solana" — https://www.osl.com/en/bits/article/ethereum-vs-solana-defi-2026-guide [^10]: Vitalik Buterin on X, February 2026 — https://x.com/VitalikButerin/status/2007559523528233041