After the Ethereum Pectra upgrade: the underlying architecture is solid, PeerDAS is poised to take off, where are developers headed?

Author: OKX Ventures

Key points

• Pectra (Prague + Electra) is the largest hard fork bundled upgrade since The Merge, containing 11 Ethereum Improvement Proposals (EIPs) aimed at synchronously reducing data costs, simplifying smart account operations, and optimizing validator operations.

• EIP-7691 doubles the Blob target capacity to 6 and raises the hard cap to 9. This reduces the Gas fees for L1 data availability (DA) to approximately below 1 Gwei, and the exchange fees on major L2s briefly fell to about below 0.02 USD.

• Now, each externally owned account (EOA) can be converted into a smart account through a single transaction (EIP-7702), enabling gas fee sponsorship, stablecoin fee payments, and a batch user experience without the need for contract migration.

• The validator cap has been raised to 2,048 ETH (EIP-7251), significantly reducing DevOps costs for large operators without significantly impacting decentralization.

• PeerDAS will be the next step in the Fusaka upgrade, and its importance to Roll-up far exceeds that of any single EIP in Pectra.

OKX Ventures’ Perspective: Pectra itself is not the “turning point for modular data availability (DA)”; it is the last key piece that allows L2 to attract mainstream market users while Ethereum maintains its settlement dominance. The future winners will be those who (a) externalize all user experience friction through 7702-style accounts, (b) leverage the benefits of Blobs to drive new data-intensive verticals (on-chain AI, order book DEX, content storage), and © double down on proof compression technology stacks ahead of PeerDAS.

OKX Ventures held a Twitter Space event on May 13, inviting guests who were directly involved in code writing or infrastructure operations after the fork to discuss these changes in depth.

In this event, OKX Ventures closely examined these changes with those who must deliver code or run hardware after the fork goes live.

• Derek Lee – Core Protocol Product Manager at Offchain Labs

• Leonardo Lerer – Core Product Manager at StarkWare

• Qi Zhou – Co-founder of EthStorage

• Pok Kopp – Co-founder of Ether.fi

• Esme Zheng – Investor at OKX Ventures

1. In the first week, who changed direction and felt the shockwave?

Arbitrum – Derek does not need to rewrite any code, but each Arbitrum instance must upgrade its embedded Geth client. The only EIP he really cares about is 7691: Doubling the blob target capacity means that DEX and gaming users will have more low-fee space.

Pectra Update:

• BoLD’s permissionless fraud proof entered the testnet in April and plans to go live on the mainnet “before the fourth quarter.”

• Stylus compiles hundreds of Rust/C contracts daily — llama.cpp inference and real-time chess engine are examples demonstrated by Derek.

• TimeBoost has replaced the “first-price auction” mechanism in the Arbitrum sequencer mempool, enhancing the fairness of transaction inclusion.

Starknet – LeoPectra mainly adjusted the Blob cost model of Starknet; the fee schedule for Cairo 1.x remains unchanged. A bigger shift is in StarkWare’s own roadmap: the reduction of L1 data availability (DA) costs has lessened the urgency for “voluntary choice” (Volition, i.e., selective off-chain data) and redirected attention back to stateless client research.

Decentralized Milestone:

• Staking v2 (based on block reward proof) will be released this quarter.

• Sorter as a Service (v0.14) - a 3f + 1 Byzantine Fault Tolerance (BFT) cluster - will be launched in 2025; cheaper Blobs may allow them to shorten the proof generation period without increasing costs.

EthStorage – Qi Zhou’s higher Blob activity is a boon – his storage layer now plans to provide continuous incentives when data is discarded on Ethereum in two weeks. Pectra also enforced a full fleet Geth upgrade; an operator group using version v1.13.8 was frozen in the middle of the epoch.

Shockwave:

• Node operators: Within 48 hours, 732 validators will increase their staked amount to over 32 ETH.

• Roll-up users: As Blob Gas fees drop to around 1 Gwei, exchange fees on major L2s have fallen below two cents, despite fluctuations in the memory pool.

• Infrastructure developers: The quick point release of Go-Ethereum keeps all developers running archive nodes on high alert.

OKX Ventures Opinion: Cheap Blobs, disposable accounts, and larger validators are not headline features; they are the foundational “pipeline” that allows the next headline feature to be smoothly implemented. PeerDAS will soon validate this assertion.

2. In-depth Analysis of EIP

Scalable Smart Account ( EIP-7702)

EIP-7702 allows any externally owned account (EOA) to behave as a smart account in a single transaction, inheriting batch calling, gas sponsorship, and stablecoin payment functionalities without the need for contract migration.

Starknet has been in the AA (Account Abstraction) world since the genesis block. Leo illustrates that the productivity application FocusTree silently deploys an account for every mobile user and starts minting achievement NFTs supported by sponsorship fees. Users are not even aware that they are operating on-chain.

EthStorage will use pay-masters to fund the first ten transactions for users—such as deploying a personal website with one click via Blob.

Arbitrum has integrated the 7702 process through third-party AA providers with GMX, Camelot, and Plays. Derek expects the first direct indicators of improvement to be the transaction success rate and failed exchange refunds, followed by a second wave of native Web2 user onboarding channels.

OKX Ventures Opinion: 7702 eliminates the final user experience barrier for self-custody. Investable areas are those payers engaging in gas arbitrage across different tokens and chains, as well as secure middleware enforcing spending limits and fraud rules at the AA layer.

Cheaper data ( EIP-7691)

Arbitrum: Each Roll-up competes for the same Blob pool; doubling the capacity is just “providing breathing room before congestion pricing takes effect.”

Starknet only released state differences; the cheaper Blob did not unlock new features, but reduced the total cost per transaction and helped avoid a shift to “voluntary selection” (selective off-chain data).

EthStorage estimates that the new capacity (approximately 3 TB every 12 days) finally makes it possible to fully store static websites smaller than 100 MB—including vitalik.ca—entirely in the Blob. It is important to note that the Gas fees for on-chain transactions are now a bottleneck, often exceeding long-term storage costs. Qi is pushing for block-level access lists and higher Gas limits to alleviate this constraint.

OKX Ventures Perspective: The focus of data is shifting back to on-chain; the recent potential market (TAM) is non-financial Blob native content (AI inference weights, game assets, social graphs), which can tolerate Blob expiration if there are retrieval incentives. Long-term tail retrieval and proof markets will become key infrastructure.

Validator cap 2,048 ETH (EIP-7251)

Data point: So far, only 732 out of 1,000,000 validators have increased their staking amount - no centralization panic has occurred.

Arbitrum & Starknet: Pure Operational Victory - Lower DevOps Costs with Minimal Impact on Users. In theory, the penalty risk for each validator has increased, but Leo considers this an “interesting academic question” rather than a practical barrier.

EthStorage: Allows running hot/cold validator replicas without increasing infrastructure costs; makes large-scale Blob proofs more reliable. Qi emphasized the operational advantages of staking services: fewer devices, the same yield, and running hot backup validators without a double hardware budget.

OKX Ventures Insight: Cost savings from hardware will flow into the re-staking and shared security market, rather than directly to end users; protocols with an EigenLayer style will absorb the released liquidity. More idle ETH → More security capacity.

3. Pectra Post-Roadmap Signal

PeerDAS (Fusaka hard fork, EIP-7623)

Every speaker used the same adjective: existential. Derek called it “critical and non-negotiable” — Offchain Labs already has Prism engineers involved in the specification discussions, as the fraud proof throughput of Arbitrum is ultimately limited by DA bandwidth. Leo sees PeerDAS as “the foundation of the entire L2 roadmap”; once sampling goes live, he can increase Starknet’s proof frequency without worrying about Blob costs. Qi has already written the economic foundation for that world — EthStorage will pay incentives to peers to maintain their availability after the protocol forgets to sample the Blob.

Verkle tree and historical data expiration

Qi sees the two paths as complementary. The Verkle tree reduces the witness data size and enables stateless clients; Historical data expiration reduces the disk space of a full node by half by discarding old blocks. He estimates that about 50% of the storage space can be saved, but this is limited to the existence of a retrieval market – Portal + EthStorage – that pays someone to keep the cold data. Leo cares less about saving and more about what the Verkle tree unlocks: a phone that can verify Ethereum without synchronizing. He wanted to see how the Beacon client handled stateless patterns before porting that idea to Starknet.

SSZ Object Trading ( EIP-6404 Series )

Qi is leading the first public testnet. Its commitment is: smaller witness data, faster decoding, and perfect alignment with the Verkle tree object hash model. Derek currently holds an agnostic attitude (“Roll-ups can absorb any format”), while Starknet is just monitoring - Cairo has already serialized with its own field element layout.

Contract size: 128 KB (EIP-7907)

If you’ve ever split a monolithic application into 24 KB pieces, you know the pain. Qi and Curve’s development teams are leading the patch, but the obstacle is not consensus – it’s the need for a DDoS-proof gas meter for very large deployment transactions.

Block-level access list

The benchmark tests of Qi show that parallel preloading can reduce Geth’s IO wait time by about 70%. This, in turn, proves the reasonableness of increasing the Gas limit, which directly lowers the cost of publishing Blobs. He is collecting mainnet tracking data to validate this trade-off.

OKX Ventures Insight: Pectra is a “comfortable” release; the next twelve months will focus on data availability, stateless verification, and ultimately provide developers with the space to deploy large, complex contracts without workarounds.

IV. Macro Picture - How Will All This Expand the Moat?

The core metric worth paying attention to is settled value. Leo expects that the compound growth curve of L2 throughput will surpass all other key performance indicators (KPIs), and he believes that “each Roll-up will bring its own exponential growth.” If this assertion holds true, then all the settlement layers anchored by these L2s—Ethereum—will capture the entire flywheel effect.

The developer retention rate now seems to be finally trending towards health. The litmus test for Qi lies in the trial use of the AA wallet and subsidy-based Blob by Web2 teams—both of which did not exist in a credible form six months ago.

Security budgets are about to be repriced. Derek noted that validator integration, coupled with a recollateralization mechanism like EigenLayer, will shift yield from idle staking to active validating services. Ether, once idle in a personal 32 ETH hot wallet, is turning into remote proof of revenue for oracles, bridges, and DSPs.

Is Pectra a “pivot”? This statement is biased. Team members unanimously believe that even if the Blob utilization rate is only 40%, large Roll-ups will not turn to Celestia or EigenDA. Ethereum’s data availability (DA) remains the most cost-effective trust premium in the crypto space. Continuously expanding the base layer, the delegated market will solve the remaining issues on its own.

Derek’s core point is: “It’s not about a complete shift, but rather an increased investment, committed to enhancing the speed of Ethereum and ensuring the comfort of Roll-ups.”

Qi added: The killer-level user experience upgrade will first undergo prototyping on L1 - 7702 is already the case, and PeerDAS sampling will also follow this path - and then cascade down to L2.

OKX Ventures Perspective: The synergy between the highly modular Roll-up stack and the continuously expanding L1 capacity curve creates a flywheel that no single chain can match. Pectra did not “save Ethereum”; it merely cleared the last user experience gap before PeerDAS increased DA by 25 times.

V. Hot Topics: Is Pectra “Enough”?

Choice of DA for Roll-ups: Consensus is No - Arbitrum, Starknet, and EthStorage do not plan to shift to external DA, even if the Blob utilization rate remains below 60%.

Non-technical leverage: Derek believes that attention should continue to be paid to L1 expansion and Roll-up support; there is no need for issuance game. Qi stated that L1 expansion experiments (block-level access lists, larger Gas limits) will originate from Ethereum and be reverse propagated to L2.

6. OKX Ventures Future Investment Strategy

The smart account infrastructure will become the default entry point for consumers. EIP-7702, combined with third-party payer mechanisms, will compress the two-year gap between the design of crypto-native contracts and large-scale user onboarding. OKX Ventures prioritizes investments in modular payer liquidity networks, intent relays, and AA risk scoring engines, enabling any Web2 application to offer “stablecoin payment gas fees” and “one-click registration” features from day one.

Blob native content is the next blank frontier. The storage incentive layer that ensures Blob retrieval after protocol expiration is highly promising. Cheap Blobs combined with the storage incentive layer will create a medium where games, AI models, and social media can be entirely built on the security of Ethereum.

Re-staking will absorb the funds saved by validators. EIP-7251 releases the hardware budget and activates idle Ether. Protocols that can convert these collateral into measurable security—such as oracle proofs, bridging validation, and shared sequencer sets—will achieve excess returns. OKX Ventures anticipates that “investment-grade” re-staking targets will become scarce and is actively funding teams with the clearest risk-adjusted accounting.

PeerDAS is the real turning point. When sampling DA is implemented, Roll-up can dilute the marginal DA cost by 10 times without giving up Ethereum trust. Teams currently building data sampling clients, proof compression circuits, or DA market will master the core tooling layer after the activation of Fusaka. OKX Ventures is actively looking for those who can take the lead in laying out tools, proof compression, and data availability market projects for Fusaka.