Ethereum After the Mumbai Conference: The 2028 Post-Quantum Race, the Glamsterdam Delay, and the Hard Fork Delivery Dilemma

Author: YQ

Translated by: TechFlow

TechFlow Summary: At the recently concluded Devcon 8 in Mumbai, Ethereum Foundation researcher Justin Drake made a striking proposal, aiming to fully complete post-quantum and AI-resilient upgrades by the end of 2028. This schedule advances the official roadmap by an entire year, meaning the underlying network must deliver three major hard forks at high frequency within 24 months.

The cryptographic algorithms are ready, but real-world delivery capabilities have hit roadblocks. The Glamsterdam testnet was forced to delay until 2027 after its "Gas Reservoir" mechanism broke several contract invariants. Ethereum stands at the intersection of technological ideals and systems engineering complexity. How to smoothly switch tracks without disrupting the existing ecosystem has become the industry's most severe test.

Key Takeaways

Ethereum Foundation researcher Justin Drake set the milestone for Ethereum to achieve a "fully post-quantum and post-AI era" on December 2028 during the opening day of Devcon 8. This advances the Foundation's previously official planned deadline of December 2029 by a full year.

Achieving this vision by the end of 2028 means Ethereum must successfully deliver three major hard forks continuously over the next 24 months. However, looking at Ethereum's version history in recent years, the actual delivery interval between single hard forks has generally ranged from 7 to 14 months.

The next core upgrade, Glamsterdam, has stalled after deploying to the Sepolia testnet. A brand-new "Gas Reservoir" repricing rule unexpectedly broke eight core invariants relied upon by existing smart contracts, making it nearly certain that the mainnet upgrade will be delayed until 2027.

At the underlying cryptographic level, the post-quantum defense puzzle is essentially complete: hash-based signature algorithms, leanVM lightweight virtual machine aggregation verification, and sub-second real-time zero-knowledge proofs have all passed feasibility checks. The system's biggest bottleneck is no longer theoretical algorithms, but how to safely complete industrial-grade engineering deployment without disrupting trillions of dollars in existing DeFi contracts.

Mumbai Conference Data Overview: More Restrained Scale and Focused Values

The 8th Ethereum Developer Conference (Devcon 8) took place from November 3 to 6, 2026, at the Jio World Centre in Mumbai, India, marking the first time Devcon has been held on the South Asian subcontinent.

In terms of overall scale, the four-day conference featured 370 talks and 399 speakers, representing a roughly 43% reduction in agenda compared to Devcon 7 in Bangkok. This scaling down was a deliberate and thoughtful choice by the Ethereum Foundation, which recently streamlined 54 roles, accounting for approximately one-fifth of the team's total headcount.

Fig: Overview of Core Data and Topic Distribution for Devcon 8 Mumbai

This year's agenda was built entirely around the Foundation's core values. Two-thirds of the plenary sessions were explicitly tagged under the Ethereum Foundation's mission framework CROPS: Censorship Resistance, Open Source, user Privacy, and foundational Security.

Among all tags, foundational security (70 sessions) and privacy protection (65 sessions) topped the discussion charts, closely followed by artificial intelligence-related topics (57 sessions). In his opening keynote, Ethereum co-founder Vitalik Buterin pointed out that modern digital technology often demands excessive blind trust from users while providing too little capacity for public verification; Ethereum must build verifiable foundations of trust across everything from operating systems and chip architectures to communications and AI.

The next developer conference (Devcon IX) is officially scheduled for October 11–13, 2027, in Lisbon, Portugal.

Racing to 2028: An Aggressive Schedule for Post-Quantum Transition

According to the Strawmap roadmap drafted by the Ethereum Foundation's architecture team, hard forks following Glamsterdam will advance sequentially by alphabetical codes: H* corresponds to Hegotá, followed by versions I*, J*, K*, and L*.`

On this aggressive evolutionary path, the plan to build a post-quantum Ethereum is broken down into four key milestones:

Fig: Ethereum Post-Quantum Evolution Roadmap and Hard Fork Delivery Cycle Projections

The first step is Hegotá, launching in 2027. Although not purely a post-quantum fork itself, it will formally introduce a "Frame Transactions" architecture, allowing each on-chain account to freely bind independent signature schemes, laying the groundwork for eventually moving away from ECDSA elliptic curve algorithms.

The second step is the I* fork, introducing the ultra-lightweight virtual machine leanVM optimized for zero-knowledge proofs, alongside native support for post-quantum lightweight transactions. The third step is the J* fork, rolling out hash-based validator signatures and post-quantum data sampling, defined by the Foundation as the "Minimum Viable Post-Quantum" defense line. Finally, the K* and L* forks will enforce state execution proofs and complete the final loop of the Lean architecture.

Delivering three major hard forks within 24 months means successfully launching one every 8 months on average. However, the actual delivery intervals for Ethereum's last three upgrades were 11, 14, and 7 months, and the current Glamsterdam has already taken 11 months since the previous upgrade. Given the current engineering delivery pace, achieving all goals by the end of 2028 faces significant practical resistance.

The urgency driving core researchers to voluntarily accelerate stems from two external threats:

The first is the accelerating lowering of the threshold for quantum attacks. A recent joint paper from Google Quantum AI, UC Berkeley, the Ethereum Foundation, and Stanford University indicates that breaking Ethereum's current elliptic curve signatures might require fewer than 1,200 logical qubits (equivalent to fewer than 500,000 physical qubits). Such an attack could complete calculations in minutes, sufficient to directly hijack high-value transactions queued in the Mempool awaiting inclusion.

The second is the unexpected breakthrough in AI-assisted cryptographic cryptanalysis. In July, Anthropic reported that its cutting-edge large model discovered a novel attack vector, crashing the key recovery complexity for the lattice-based HAWK-256 signature scheme from 2^64 down to 2^38. Researcher Justin Drake stated that the industry must proactively enter a "bunker defense mode" to prevent AI-assisted mathematical cracking from breaching traditional cryptographic systems before physical quantum computers arrive.

Deconstructing the Glamsterdam Failure: What Did the Gas Reservoir Rule Break?

As a pivotal transitional node, the Glamsterdam hard fork has been live on the Sepolia testnet since October 6. Its core vision is to implement EIP-8037 and EIP-8038, attempting to recycle previously burned execution gas back into a so-called "Gas Reservoir" to reduce state storage costs.

However, on October 22, the core development team had to urgently withdraw the version from the second testnet. After an in-depth review, the security audit team discovered that this seemingly clever repricing rule broke eight core invariants that on-chain smart contracts have long relied upon.

Fig: Analysis of the Glamsterdam Hard Fork Gas Reservoir Mechanism and Broken Contract Invariants

In Ethereum's current architecture, developers widely adhere to a series of ironclad rules: for instance, the gas consumption of child contexts in external calls never exceeds the limit granted by the parent transaction; or reverted transactions absolutely cannot have a persistent resource impact on subsequent operations.

However, the new Gas Reservoir mechanism breaks this determinism: even a completely reverted malicious child transaction can accumulate unsettled refunds in the reservoir, causing abnormal gas increases when subsequent legitimate code executes. This anomaly not only allows attackers to brute-force CREATE2 contract address salts at near-zero cost, but also directly pierces the security validation logic of Paymaster proxy contracts in the ERC-4337 account abstraction system.

The rollout of the main testnet Hoodi, originally scheduled to launch on November 5, has been comprehensively delayed. Core developers are currently weighing whether to patch the reservoir algorithm or scrap it entirely. Due to this setback, the Glamsterdam mainnet upgrade will likely slip into Q1 2027.

The True Cost of Post-Quantum Consensus: From Hash Signatures to LeanVM Aggregation

If Ethereum abandoning elliptic curves in favor of hash-based signatures is an irreversible historical destiny, then the underlying network must pay a steep price in bandwidth and storage.

Currently, a single BLS aggregate signature used by Ethereum validators occupies only 96 bytes. In contrast, hash-based signature schemes with anti-quantum properties (such as SPHINCS+ or its lightweight variant leanSPHINCS) typically exceed 4 KB per entry, inflating by more than 40 times.

Fig: Comparison of Aggregation Overheads for Hash-Based Signatures and LeanVM in the Post-Quantum Consensus Layer

Currently, over 1.1 million active staking validators operate across the network. If millions of 4 KB signatures were indiscriminately broadcasted in every 12-second block slot, each slot would generate a terrifying data flood exceeding 4.4 GB, instantly suffocating and paralyzing any decentralized network.

The core solution to this dilemma is integrating the state-proof virtual machine, leanVM. By using STARK proof systems at the validator node level to recursively compress massive volumes of hash signatures, discrete validations from thousands of nodes can be bundled into a single, concise proof, squeezing broadcast bandwidth back down to a safe level of tens of kilobytes. Meanwhile, the EIP-8450 proposal is establishing a unified registry for post-quantum validator public keys to ensure a smooth transition of validation rights across forks.

EIP-8288 Proposal: Offloading Complex Cryptography to the Transaction Mempool

To achieve equivalent anti-quantum protection at the transaction execution layer without congesting Ethereum block space, core researchers introduced the ambitious yet ingeniously designed EIP-8288 proposal.

The core design philosophy of EIP-8288 is explicit: isolate heavy, complex cryptographic proofs outside the base blockchain, forcing distributed mempools and professional builders to shoulder the computational burden.

Fig: EIP-8288 Transaction Mempool Encryption and Block Recursive Proof Workflow

Under this mechanism, post-quantum transactions initiated by regular users enter the mempool carrying complete anti-quantum cryptographic credentials. However, at the moment block nodes package transactions onto the chain, block builders run off-chain aggregators to unify compress hundreds or thousands of heterogeneous signatures into a minimalist recursive STARK proof, ultimately writing only a proof summary to the block header.

This mechanism delivers three disruptive advantages: First, the actual gas cost for post-quantum signatures drops drastically from 150k–200k gas for direct on-chain verification to roughly 33k gas. Second, the on-chain verification cost for privacy transaction proofs crashes from millions of gas down to around 60k gas. Finally, the base network no longer needs to deploy dedicated precompiles for each emerging algorithm, allowing any future cryptographic breakthrough to seamlessly integrate into Ethereum directly as a proof.

Finality Determinism: Shrinking Confirmation from Minutes to Seconds

Besides anti-quantum defenses, another core breakthrough established by Ethereum's consensus layer at the Mumbai conference is Single Slot Finality (SSF).

On the current Ethereum mainnet, a transaction takes about 16 minutes on average to cross two full epochs from being included in a block until it achieves irreversible finality. This severely limits the settlement efficiency of cross-chain bridges, decentralized exchanges, and centralized institutional channels.

Fig: EIP-8453 Decoupled Consensus Mechanism and Second-Level Deterministic Confirmation Architecture

The draft EIP-8453 "Decoupled Consensus" proposal completely splits traditional consensus duties in two: a rotating fast committee of 512 validators handles instant block production and availability voting, while a post-confirmation engine composed of full-validator nodes specializes in granting immutable finality locks.

This mechanism initially reduces final confirmation time by a factor of over four. In long-term planning, Ethereum finality is expected to shrink first to 36 seconds, ultimately advancing toward the "North Star goal" of 1-second blocks and 2-second final confirmations, completely aligning decentralized network settlement experiences with centralized high-performance cloud platforms.

ZK Proof Adoption: Replacing Full Node Re-execution with Real-Time Validity Proofs

In the execution-layer scaling dimension, Ethereum is firmly undergoing a comprehensive paradigm shift from "global full-node re-execution" to "real-time zero-knowledge proof verification".

For a long time, the industry's biggest concern regarding zkEVM has been the compute consumption and time latency of proof generation. However, during live benchmark tests on November 8, three independent prover teams successfully met sub-second real-time proof standards running on environments of just 4 to 16 consumer-grade GPUs. The proof cost per single block has astonishingly plummeted to between $0.003 and $0.007.

Once the compute and cost bottlenecks are breached, formal security assurance becomes the new institutional hurdle. The Ethereum Foundation plans to establish 128-bit provable security technical standards by early December, ensuring complex recursive verifier code is absolutely free of logical flaws. As future K* forks mandate state proofs as a requirement, Ethereum will officially usher in a super-throughput era evolving toward hundreds of thousands of transactions per second (Gigagas).

Other Key Ecological Developments During Mumbai Week

During the week-long Mumbai tech week, other key tracks within the Ethereum ecosystem also demonstrated vibrant evolutionary trends:

Blob Data Availability: Since the adjustment in January this year, the Blob target capacity has increased to 14 (max 21). For the past few months, the actual average usage per block has been around 4.4, with prices well below 1 cent. Infrastructure data throughput capacity has significantly outpaced current ecosystem demand.

Layer 2 Ecosystem Landscape: L2BEAT, which has long tracked second-layer networks, has officially renamed itself to ETHBEAT. The vast majority of mainstream Rollup networks are now solidly operating in Stage 1 (with Base locking approximately $14 billion and Arbitrum around $9 billion).

Cross-Chain Interoperability: The Ethereum Economic Zone successfully demonstrated atomic-level cross-chain calls between L1 and L2 on its dev network. Gnosis Chain has explicitly announced it will officially integrate this standard in Spring 2027.

Security & Self-Custody Reflections: Regarding the recent severe security incident where Binance's hot wallet suffered a targeted breach resulting in $6.36 billion stolen, on-chain forensics firm TRM Labs noted that hackers exploited pre-signed EIP-7702 authorizations to drain funds en masse. While account abstraction amplifies the execution radius of single operations, the root cause remains compromised internal signing credential management, not inherent flaws in protocol design.

Future Observation Windows and Reflections

Looking across the entire Mumbai conference, Ethereum has displayed extremely clear technical resolve. From the cryptographic foundation and consensus mechanism restructuring to execution-layer proofs, Ethereum has completely clarified its technological blueprint for its third decade.

The key windows most worthy of attention in the coming months include: the mid-November ACDC and ACDE core developer telecons, where the patch path for the Glamsterdam vulnerability and timelines for new testnets will be finalized; the Foundation's formal review of the post-quantum timeline in January next year; and the delivery of the 128-bit zkEVM security milestone at year-end.

Ethereum knows how to construct an ultimate distributed world computer capable of withstanding quantum attacks and rapid technological shifts for decades to come. And what the Mumbai conference leaves to the entire industry as the sole, most profound suspense is whether, atop this increasingly complex leviathan, the core team can truly uphold the engineering miracle of delivering a hard fork every eight months while maintaining industrial-grade security.

`