Author: YQ Source: X, @yq_acc Translation: Shan Ouba, Jinse Finance
I. From Philosophical Concepts to Infrastructure
Core Shift: In 2015, Ethereum regarded decentralization, censorship resistance, and trustlessness as its ultimate values (stemming from its own goals rooted in cypherpunk ideology); by 2025, these characteristics will become instrumental values (means to achieve practical goals such as eliminating counterparty risk, promoting global financial inclusion, and reducing operating costs). The difference between the two is crucial: idealism asks, "To what extent can we achieve decentralization?", while pragmatism focuses on "How much decentralization is needed to solve this problem?". The 2025 Devconnect conference in Argentina demonstrated that the Ethereum Foundation had clearly chosen the latter. When Ethereum launched in July 2015, its founders' vision was deeply rooted in the cypherpunk movement of the 1990s. Like 2,000 mathematicians, cryptographers, and software engineers who united to resist government attempts to ban encryption, the early Ethereum community held decentralization and censorship resistance as its ultimate goals. The project aimed to build a "world computer," allowing smart contracts to execute with mathematical determinism, free from corporate or governmental interference. This philosophical belief sustained the project through its early years, attracting a large number of developers drawn together by ideology rather than commercial interests. Ten years later, at the 2025 Devconnect conference in Buenos Aires, Argentina, the five-day technical conference showcased a drastically different Ethereum. Presentations from Ethereum Foundation researchers and application developers indicated that the project had decisively shifted towards pragmatic infrastructure optimization, focusing on enabling real-world applications rather than pursuing abstract protocol perfection. This shift is clearly visible in the conference structure itself: November 17th kicked off with "Ethereum Day," where Thomas Stanzak, Wang Xiaowei, Ansga Dietrich, Barnabe Mono, and Vitalik Buterin jointly introduced the Ethereum Foundation's strategic reorganization in April 2025, establishing three specific priorities: expanding the network by increasing the gas cap (Layer 1), expanding block availability by deploying PeerDataAS (PeerDataAvailability Sampling), and improving user experience through cross-chain interoperability. They then detailed the specific achievements for 2025: the Fusaka upgrade in December doubled the gas cap from 30 million to 60 million, the number of validators exceeded 1.1 million, the staked ETH reached $70 billion, and security was provided for the network. The following days of meetings focused on specific application areas rather than abstract protocol theory: Stanislav Kulechev highlighted decentralized lending markets, while Santiago Palladino elaborated on the Ethereum interoperability layer; the "Trustless Agent Day" on November 19th launched the ERC-8004 portable reputation protocol and the x402 micropayment protocol, laying the foundation for an autonomous AI agent economy; the Privacy Summit and Zero Knowledge Proof Day (ZK Day) showed that zero knowledge proofs have achieved an average block verification time of less than 10 seconds. Throughout the sessions, speakers consistently emphasized solving specific problems for real-world users: cross-chain fragmentation, DeFi privacy needs, institutional settlement needs, and autonomous agent payment infrastructure. We can clearly see Ethereum's evolution from idealism to pragmatism, a shift reflected in five key infrastructure areas—the ultimate philosophical goal has given way to instrumental engineering goals: scaling the Layer 1 network through incremental optimization rather than architectural revolution; solving Layer 2 (L2) network fragmentation through cross-layer interoperability; empowering a $300 trillion credit market with decentralized finance infrastructure components; building autonomous agent infrastructure for machine-native economies; and an institutional adoption framework that prioritizes privacy over transparency.

II. Layer 1 Network Scaling: Incremental Optimization Rather Than Revolution
Strategic Restructuring and a 3x Annual Growth Target
In their opening speeches at Ethereum Day, Ansgar Dietrichs and Barnabé Monnot detailed the Ethereum Foundation's strategic restructuring in April 2025—marking the project's departure from years of sharding research and a shift towards achievable throughput enhancement solutions, clarifying short-term timelines.
Client Performance Benchmarks and Engineering Constraints

Fusaka upgrade performance benchmarks are from the Sepolia testnet and mainnet shadow fork: The Geth client, which holds approximately 60% of the validator share, takes 3.0 seconds to process a 60 million Gas full block, achieving a throughput of 20 million Gas/second; Nethermind executes the fastest, requiring only 2.4 seconds (25 million Gas/second); Besu takes 3.3 seconds (18 million Gas/second); Erigon Block processing takes 2.7 seconds (22 million Gas/second). All client processing times are well below the critical 4-second threshold, ensuring that 90% of validators receive and process blocks within the first quarter of the 12-second slot, maintaining a consensus safety margin. Network propagation analysis shows that 90% of validators receive blocks within 0.7 to 1.0 seconds via the gossip protocol, but the remaining 10% may face delays of 2 to 3 seconds due to geographical differences. These engineering realities dictate that the Gas cap must be increased gradually, rather than suddenly and drastically, to avoid jeopardizing network stability. The current bottleneck has shifted from raw execution speed to state access patterns, disk I/O, and accumulated state growth: data shows that for complex transactions, accessing state accounts and storage slots has become the primary drain on execution time. With a gas limit of 60 million, the state grows by approximately 60GB annually. Without mitigation measures, increasing the gas limit to 300 million would result in an annual state growth of 300GB, reaching several terabytes within a few years. This reality initially spurred Verkle tree research, but advancements in quantum computing forced the project to abandon this approach in favor of a pragmatic short-term management strategy (active pruning, state rent economics), while simultaneously developing a quantum-resistant binary tree alternative for 2027-2030. The second strategic priority—expanding data block availability—directly addresses the needs of Layer 2 network Rollups. Current infrastructure supports 3 to 6 data blocks per block (128KB each), providing 384 to 768KB of data capacity per 12-second slot. The PeerDataAvailability Sampling (PeerDAS) deployed in the Fusaka upgrade uses erasure coding technology to scalable the number of data blocks to 16 in the short term and potentially 64 in the long term. Each data block is divided into multiple fragments using Reed-Solomon coding, and only 50% of the fragments are needed to reconstruct the complete data. Validators do not need to download the complete data block; they only need to randomly sample a portion of the fragments. The network collectively ensures data availability, eliminating the need for individual validators to store all the data. This sampling method increases the number of data blocks by more than 10 times while reducing the bandwidth requirement of a single validator from O(n) to O(log n). Deployment timeline: Development network testing in Q3 2025, entry into the test network in early 2026, and mainnet activation after security review in mid-2026. Once PeerDataAvailability Sampling is implemented, it will increase the data availability of Rollups by 10 times and reduce data block gas costs through scaling. ZK-EVM Verification: From Theory to Production Timeline Vitalik Buterin's talk on proxy infrastructure and Ansgar's protocol update both highlighted the groundbreaking progress in proof time for the Zero-Knowledge Ethereum Virtual Machine (ZK-EVM). In 2025, multiple ZK-EVM teams (PSE, Scroll, Polygon, Taiko, zkSync) reduced the average block proof time from 5 to 10 minutes in 2024 to less than 10 seconds, a key milestone towards real-time proofs (completed within 12 seconds to accommodate slot times). The deployment will be phased: Phase 1 will experimentally introduce proof of validity, allowing some blocks to be verified using zero-knowledge proofs, with verifiers choosing to verify the proofs instead of re-executing them; Phase 2 will implement a hybrid mode, requiring zero-knowledge proofs for critical blocks, while most blocks will still be executed normally; Phase 3 will transition to a mandatory proof mode, requiring all blocks to include zero-knowledge proofs; Phase 4 will implement a full zero-knowledge Ethereum Virtual Machine, enabling stateless clients to run without storing state and supporting mobile nodes and browser nodes with full security. The actual production deployment timeline is 2027-2030.
III. Cross-Layer Interoperability: Solving Layer 2 Network Fragmentation
Fragmentation Issue
In his keynote speech "Ethereum Is Everywhere," Santiago Palladino revealed a fundamental contradiction in Ethereum's Rollup-centric roadmap: despite the total throughput of over 50 Layer 2 networks exceeding 100,000 transactions per second, fragmentation has led to serious user experience issues and liquidity splits, potentially weakening the value proposition of Ethereum's unified ecosystem. When users hold assets on Arbitrum, if they want to purchase non-fungible tokens (NFTs) on zkSync, they must transfer them across chains via a Layer 1 network, waiting 7 days (the fraud proof window of optimism Rollup) and paying a $35 gas fee; liquidity is dispersed across chains, and the trading price of the same token differs across different Layer 2 networks; applications need to be deployed separately on each chain, diverting developer efforts and user base. Ethereum Interoperability Layer (EIL): Single Signature Enables Multi-Chain Operations The Ethereum Interoperability Layer (EIL), jointly developed by the Arbitrum, Optimism, Polygon, zkSync, and Base teams, is based on the ERC-4337 account abstraction and supports cross-chain operations via a single signature. Its core technological innovation is Merkle tree bulk authorization: users construct a Merkle tree containing operations across multiple chains, sign the Merkle root, and submit branches to various target chains; smart contract accounts on each chain achieve atomic multi-chain execution by verifying the Merkle proof corresponding to the signature root, eliminating the need for complex cross-chain messaging protocols. Palladino demonstrated the specific efficiency improvements of this mechanism in his presentation: A user holding 10,000 USDC on Arbitrum who wants to purchase an NFT on zkSync for 5,000 USDC only needs a single Merkle root signature for authorizing the Arbitrum deduction and the zkSync purchase operation; the cross-chain liquidity provider (XLP) completes instant settlement by providing 5,000 USDC in advance on zkSync, and then withdraws the user's funds from Arbitrum after the withdrawal delay period, incurring only a service fee of approximately $5 (0.1%). For the user, the entire transaction takes less than 1 minute, while traditional cross-chain transactions take more than 7 days and incur fees as high as $35.
As a foundational account abstraction
ERC-4337 provides the foundation for the Ethereum interoperability layer by replacing externally owned accounts (controlled by ECDSA private keys) with programmable smart contract accounts.
Traditional Ethereum addresses can only authorize one operation per signature, while smart contract accounts can implement arbitrary verification logic, including Merkle proofs that authorize multiple operations simultaneously. This functionality has existed in theory since Ethereum's launch, but ERC-4337 standardized the implementation and built the alternative mempool infrastructure needed for production deployment. The November 18th meeting revealed significant commitments from mainstream wallets: MetaMask, Argent, and Safe have all deployed smart account infrastructure, with Safe reporting over 100,000 active accounts as of November 2025. Improved user experience extends beyond cross-chain operations, including support for paying gas fees with ERC-20 tokens via payment brokers, social recovery mechanisms, and programmable spending limits. In his presentation on improving Ethereum's service quality as the "confirmation engine" for the entire ecosystem, Barnabé Monnot emphasized two key metrics: block confirmation time (currently averaging 12 seconds) and finality time (currently 13 minutes). Fast confirmation rules will be deployed in Q1 2026, providing 95% determinism within 1 to 2 blocks (12 to 24 seconds), eliminating the need to wait 13 minutes for economic finality. This weaker security assumption (probabilistic finality rather than economic finality) applies to various scenarios: Layer 2 networks can utilize confirmed Layer 1 network states more quickly (forward-looking Rollups will benefit), cross-chain bridge protocols can enable faster cross-chain messaging, and centralized exchanges can reduce deposit and withdrawal latency. Long-term plans include reducing slot time from 12 seconds to 6 seconds, doubling the network block generation rate. Current client performance (processing 60 million gas blocks in 2.4 to 3.3 seconds) indicates that with continued client optimization, a 6-second slot time will be feasible when the gas limit is increased to 100 million or more. A 6-second slot time, combined with fast confirmation rules, will achieve effective finality of 6 to 12 seconds, making it competitive with centralized payment networks.
IV. Decentralized Financial Infrastructure: A $300 Trillion Credit Opportunity
Stani Kulechov's Renaissance Financial Theory
In his keynote speech on "A New Architecture for Credit," Stani Kulechov compared Renaissance-era financial innovations with modern decentralized finance (DeFi) infrastructure components, proposing that decentralized credit markets have the potential to unlock $300 trillion in global capital flow opportunities.
In Florence in 1252, the florin became the first widely trusted base currency. Its standardized weight and predictable purity laid the foundation for credit expansion across Europe. Today, decentralized finance (DeFi) achieves similar functionality through stablecoins (USDC, DAI, USDT, etc., providing $150 billion in on-chain liquidity). The Venetian merchant intelligence network collected commercial information from Mediterranean ports, equivalent to the function of oracle infrastructure (Chainlink provides price feeds and off-chain data verification). Merchant networks like the Hanseatic League built liquidity layers connecting local markets, similar to automated market makers (Uniswap, Curve, etc., support instant token swaps across pools). The Comanda contract of the Renaissance allowed passive investors to fund merchants' maritime trade and agree on profit sharing, a precursor to smart contracts—automatically allocating capital based on programmed conditions. Kulechev's core argument is that the $300 trillion global credit market has yet to be reached by decentralized finance (DeFi) because traditional lending relies on local information (borrower reputation, law enforcement, collateral valuation), which cannot be directly mapped onto the blockchain. His proposed solution is the Aave Horizon protocol, which tokenizes local credit for global DeFi liquidity participation. Local credit analysts assess borrowers using traditional methods (credit history, cash flow analysis, collateral valuation) and then package loans into tokenized tiered products for on-chain trading. DeFi liquidity providers purchase these tiered products, earning returns from the local credit market, while the protocol handles compliance, collections, and default management. Kulechev specifically chose Argentina as the venue for his conference and as a case study: Argentina's credit market suffers from extreme inefficiency—credit card annual interest rates exceed 100%, mortgage loans are difficult to obtain despite real estate value, and capital controls hinder cross-border investment; local businesses with strong cash flow cannot access growth funding at reasonable interest rates, while institutional investors seek returns in emerging markets. The Aave Horizon Buenos Aires pilot project, announced during the Devconnect conference, addresses this gap by tokenizing accounts receivable from Argentine SMEs and offering them to global decentralized finance investors with annualized returns of 15% to 25% (attractive to investors and transformative for borrowers accustomed to financing costs exceeding 100%). Atomic Settlement and Programmable Composition: Danny Ryan's presentation emphasized how blockchain infrastructure revolutionizes operational efficiency through cryptographic settlement rather than legal enforcement. In traditional finance, stock transactions have a T+1 settlement cycle (one business day), corporate bonds T+2, and private equity transactions 90 to 180 days; each settlement involves multiple intermediaries (transfer agents, custodians, clearinghouses, payment processors), with corporate bond transactions requiring approximately 20 manual steps, and 5% to 10% of transactions failing due to reconciliation errors. Ethereum simplifies this process to atomic execution: smart contracts simultaneously receive assets from both parties, either completing the exchange instantly or rolling back the entire transaction. On Layer 2 networks, settlement can be completed within 12 seconds at a cost of less than $5, reducing both time and cost by 99.9%. More importantly, atomic composability makes financial products impossible in traditional systems a reality: Morpho's presentation demonstrated cross-collateralized lending—an institutional client deposits $100 million in tokenized U.S. Treasury bonds and can instantly borrow 90 million USDC, with loan terms automatically adjusted based on Treasury yields. If the collateral ratio falls below a safety threshold, programmatic liquidation is triggered. The entire process requires no legal contracts, credit checks, or settlement delays. Privacy Infrastructure and Limited Visibility The November 19th Privacy Summit clearly stated that privacy has replaced regulatory concerns as the primary obstacle to institutional adoption. Europe's Crypto Asset Market Regulation Act (MiCA) provides a clear regulatory framework, and the US approval of Bitcoin and Ethereum exchange-traded funds (ETFs) indicates that regulators have recognized cryptocurrencies as an asset class; however, privacy infrastructure development lags behind. Institutions need what the speaker calls "limited visibility": different stakeholders should see different subsets of data based on their roles and permissions—fund managers should see full holdings to make allocation decisions, regulators should verify compliance but not obtain strategic trading information, clients should only see their own positions and not those of others, and the public should only see aggregate metrics such as total assets under management. The complete transparency of public blockchains cannot meet these requirements, where all information is visible to everyone. The proposed technical solution integrates multiple cryptographic technologies: private layer-2 networks like Aztec encrypt state by default, with decryption keys distributed according to access policies defined in smart contracts; zero-knowledge proofs support selective disclosure—proving to regulators that customer verification (KYC) has been completed without revealing identity information, or proving that transactions comply with approved limits without exposing actual positions and counterparties; multi-party computation allows for collaborative analysis, and no participant can access the raw input data of others. As of November 2025, BlackRock's BUIDL fund had reached $500 million in assets. The fund operates on Ethereum but maintains privacy through authorized access and off-chain reporting. This fund structure reflects both the market's demand for blockchain settlement (atomic, programmable, and available 24/7) and the inadequacy of current privacy infrastructure. The presentation indicated that native privacy features will be available in production in 2026, potentially driving larger-scale institutional deployments, and that tokenized assets could reach $100 billion by 2027. V. Autonomous Agent Economy: ERC-8004 and x402 Portable Reputation Infrastructure The "Agent Day" on November 19th showcased a complete AI agent economy infrastructure, with the core assumption that agents will become major economic players within the next decade. This shift from a human-centric to an agent-native design is embodied in two complementary protocols: ERC-8004 for agent identity and reputation, and x402 for machine-native payments. ERC-8004 extends the ERC-721 nonfungible token standard, adding reputation tracking: each agent receives a unique token ID and accumulates performance metrics (number of tasks, basis point success rate, total transaction value, and Merkle roots of detailed performance proofs stored on IPFS or Arweave). On-chain reputation is portable across platforms, addressing the fragmentation problem of traditional service markets—where reputation is confined to a single platform (Upwork's reputation cannot be transferred to Fiverr). The technical specifications define on-chain and off-chain components to balance verifiability and storage costs: on-chain contracts store compact reputation vectors (number of tasks as uint256 type, success rate as basis point, total value as uint256 type, and Merkle roots of performance proofs); off-chain infrastructure indexes complete performance data. If an agent claims a 95% success rate across 1000 tasks, it must provide Merkel proof linked to verifiable task completion records (cryptographic signature of the task initiator, timestamp, and result description) to prevent exaggeration of reputation through false claims. x402: Payment Instead of Authentication The x402 protocol addresses the authentication problem for autonomous agents in unsupervised situations. Traditional API access requires developers to manually register accounts, complete OAuth processes, and manage API keys, all of which assume human interaction. Autonomous agents cannot complete CAPTCHA challenges or securely store long-term keys without centralized key management (which contradicts the principles of decentralization). x402's solution is payment instead of authentication: to access a resource, an agent simply pays a specified amount of cryptocurrency. The protocol flow is concise and clear: the proxy requests resources, the server returns a 402 (Payment Required) HTTP status code, including payment details (amount, token type usually USDC, receiving address, and a unique random number to prevent replay attacks); the proxy constructs a transaction, transfers the specified amount to the receiving address, submits it to the Ethereum Layer 2 network for fast confirmation (Arbitrum or Base provides sub-second finality), and then re-initiates the request with the transaction hash as proof of payment; after the server verifies the transaction on-chain (amount matches requirements, receiving address is correct, random number is not used), it provides the proxy with the required resources.
VI. Institutional Adoption: Counterparty Risk and Cryptoeconomic Security
Wall Street's Demand for Decentralization
Danny Ryan's November 17th speech on Ethereum institutional adoption overturned traditional understandings of the value proposition of blockchain.
VIII. Conclusion: Application-Ready Infrastructure
The 2025 Devconnect conference in Argentina demonstrated that the Ethereum Foundation has decisively shifted its focus to application infrastructure development, rather than protocol research itself. The five-day technical presentations focused on addressing specific issues through a clear timeline: Deploying the Ethereum interoperability layer by mid-2026 to address cross-chain liquidity fragmentation; launching zero-knowledge Rollup encrypted states in 2026-2027 to meet the privacy needs of decentralized finance; launching x402 and ERC-8004 in Q1 2026 to empower proxy payment infrastructure; meeting institutional settlement needs through privacy features and throughput scaling; deploying node data availability sampling in the "DeepSaka" upgrade on December 3, 2025, to expand data block availability; managing state growth through pragmatic optimizations (client pruning, witness compression), while developing quantum-resistant binary tree alternatives for the sustainability era of 2027-2030.