Ethereum Will Start Scaling Exponentially With ZK in 2026

2026 is a pivotal year for Ethereum. The first Ethereum validators will process tiny zero-knowledge (ZK) proofs instead of reexecuting transactions. This unlocks immediate scaling benefits for the layer 1 and sets it on the path toward 10,000 transactions per second (TPS).

Researcher Justin Drake demonstrated that validating proofs on an old laptop is already possible at EthProofs Day at Devconnect in November. One in 10 validators are expected to make the switch to ZK before the end of the year.

It’s a complete overhaul of the fundamental way the blockchain works: comparable in scale to the Merge in 2022, when Ethereum successfully switched from proof-of-work to proof-of-stake.

At present, every validator reexecutes every single transaction, and while the introduction of perfect parallel processing in Glamsterdam will make that process more efficient, it’s only a step along the road.

The plan is to generate a ZK-proof for each block — magical math that proves the block executed properly — and then validators simply check if the proof is legit.

This neatly solves the blockchain trilemma because it’s so easy to validate a ZK-proof that you can theoretically do it on your smartphone or even on a smartwatch. That ensures the network can remain highly decentralized while being unburdened by the lowest-spec devices in the network. Ethereum can manage about 30 TPS at present (it currently does fewer), but the requirements for home validators are already at gaming laptop level.

“It’s a way to scale the network and scale the traffic with just fewer resources having to work harder,” said Gary Schulte, senior staff blockchain protocol engineer on the Besu client. He explained the gas limit can’t be raised much further without increasing the minimum specs for the validator hardware. But under the new system, most of the difficult work that requires beefy equipment is carried out by block builders and ZK provers.

“If we have a small handful of machines that are building these blocks, executing and proving these blocks, and all of our downstream validator network is doing very light work …. it allows us to scale,” said Schulte.

Drake expects around 10% of the network to switch over to validating ZK-proofs this year as part of “phase 1” of Lean Execution. Given that the validators most likely to switch over are the lowest spec home validators, this will enable the gas limit to be increased, as the remaining validators still reexecuting transactions will be running more capable machines.

”The sophisticated operators with large infrastructure will continue onwards, and this means that gas limits can be increased, as you don’t have to worry about solo stakers,” Drake told Bankless.

That said, switchover won’t happen until mid-year, when Ethereum stops penalizing validators for delayed execution. This currently disincentivizes the validation of ZK-proofs because these take longer to generate and propagate through the network.

Schulte says that only validators “willing to pay the penalties” will validate ZK-proofs initially, but the ePBS upgrade in the Glamsterdam upgrade will fix the issue.

“Basically, instead of having to attest immediately when the block arrives, you have more time. You think of a whole slot basically to attest,” said Drake.

How Ethereum’s ZK rollout is expected to work

Back in July, Sophia Gold wrote a blog for the Ethereum Foundation setting an ambitious goal to ship a zkEVM within the space of a year.

Several teams have already demonstrated the ability to generate proofs in real time, which in reality means about two seconds faster than Ethereum’s 12-second block times, given the need to propagate the proofs throughout the network.

Because all the current systems have some bugs, and they sometimes fail, one proposed solution is to use five different proving systems to offer up ZK-proofs, and as long as a validator receives three matching proofs, the block is good to go.

The aim is to eventually have one proof (enshrined proofs), but that will require the generation system to be formally verified to have zero bugs. There’s a team working on that, but it’s not expected to be complete until 2030.

Proving doesn’t need to be anywhere near as decentralized as validating (a proof is either correct or not), but the target specs are still set at the level of equipment a home enthusiast could conceivably afford to run in their garage: costing under $100,000 and requiring a similar amount of power to a Tesla Powerwall.

Given the astonishing rate of progress in the field, these specs might be higher than required: In May, it took SP1 Hypercube’s 160 GPUs to prove blocks in under 12 seconds. The ZisK team can now prove blocks in 7.4 seconds using 24 GPUs, while the ZKsync Airbender team can prove Ethereum blocks in under 50 seconds using a single GPU, albeit with lower security than required.

Ethereum’s switch to ZK: Phases 0, 1 and 2

We are currently in Phase Zero of the switchover, where only enthusiasts like Drake are willing to take the hit to validate proofs. Phase One is expected in 2026, with up to 10% of validators switching over. Drake hopes Phase Two will happen in 2027.

“Phase Two is where a lot of the magic really happens, which is the mandatory proofs where we require the block producer to generate the proofs and everyone is expected to be running on ZK EVMs,” he explained.

Across the year, expect to see more debate over whether the Ethereum Virtual Machine (EVM) should switch to RISC-V (reduced instruction set five) to make generating ZK-proofs easier.

Related: Vitalik Buterin pitches ZK-proofs to audit X algorithm and rankings

“The momentum is with RISC-V architecture because it’s a really small, tight instruction set architecture that is really appealing for reasons of licensing and royalties and things like that being an open standard,” explained Schulte.

But at Devconnect, Schulte noted a split developing “amongst core devs about the viability and risks of a ZKVM scaling approach.

“The concern I have heard is that the teams and software bases that are the most battle tested are the ones that have significant challenges targeting a zkEVM environment. Only relative newcomers are able to compile to RISC-V natively, and most ZKVM teams are running their own modified version of RETH,” he said.

Alex Gluchowski, co-founder and CEO of Matter Labs, told Cointelegraph that the embrace of ZK-proofs will also enhance interoperability in the short term among chains using proving standards, and in the long term, “they make interoperability more native.”

ETH in 2026: Ethereum Interoperability Layer

Another big improvement for the Ethereum ecosystem in 2026 is the Ethereum Interoperability Layer (EIL). It’s a trustless messaging system enabling different layer 2s to communicate and transact seamlessly, explained Taiko chief operating officer Joaquin Mendes.

“Ethereum now has 55+ L2 rollups that successfully scaled the network but created siloed ecosystems with fragmented liquidity. The EIL unifies these into what feels like a single chain,” he told Cointelegraph.

Once it’s operational, a user can send 100 USDC (USDC) from their Arbitrum wallet directly to another user’s wallet on Base, who will receive them in seconds. Wallets could even collect some Ether (ETH) from a Scroll address and some more ETH from an Arbitrum address to pay for a non-fungible token on Linea.

Built using ERC-4337 account abstraction, the EIL is an attempt to improve on the solver networks that underpin solutions like Near Intents, where users express what they want to happen and then a network of solvers works out how to make it happen for a fee.

Mendes says current solver networks are not decentralized enough and require too much trust. “When a few dominant solvers control the market, competition drops and extractive behavior rises. Without accountability frameworks and decentralized governance, solvers create censorship vectors that undermine Web3 principles,” he said.

The EIL is specifically designed to enable wallets to perform cross-chain actions without relying on intermediaries who can steal or freeze funds or launch a sandwich attack. Instead of carrying out transactions, liquidity providers are incentivized to rebalance funds in cross-chain pools.

“EIL is account-based interop: the user’s own account directly performs every call on every chain. Liquidity providers only supply gas and assets – they never submit transactions and never see the call targets,” explained EIL developer Yoav Weiss. “This removes the ‘mid-state’ trust dependency that exists in intents and bridges, where a 3rd party solver/relayer transacts on the user’s behalf.”

The Ethereum Foundation’s three-phase rollout was targeting completion by 2026, but it still appears to be in the research and development phase. The system won’t work as well as it should until fast L1 confirmations arrive in a future hard fork. However, the production contracts for the Open Intents Framework are already live.

Taiko is a based rollup. It uses Ethereum’s validators for sequencing, which offers the possibility of synchronous composability: real-time interoperability with other based rollups. It “is still a future milestone but closer than ever,” said Mendes.

“The EIL also benefits based rollups: while based rollups can more easily compose with each other through shared L1 sequencing, the EIL provides a standardized messaging layer to interoperate seamlessly with non-based rollups, improving connectivity across the broader ecosystem.”

ZKsync’s Atlas upgrade

Ethereum’s L2s are also taking advantage of ZK-proofs. Appchain Lighter is using custom ZK circuits to scale up to tens of thousands of TPS, and ZKsync is using the tech to provide instantaneous interoperability between the Ethereum L1 and ZKsync chains using its Atlas upgrade and new Gateway architecture. The first Atlas chain in production, the UAE’s institutional chain ADI, has just gone live.

Gluchowski from Matter Labs said that the upgrade enables funds to stay on mainnet but be traded in the fast execution environment of chains in ZKsync’s Elastic Network.

He explained that funds can stay on the Ethereum L1 but be used instantly on the L2s. At a high level, he said, “Assets are custodied on Ethereum L1, and ZK-proofs are used to safely reflect their movement across ZK Chains.”

“From a user or application perspective, you can treat L1-origin assets on ZKsync chains as effectively real-time while still inheriting Ethereum’s security and finality. For institutional and RWA flows that already wait for Ethereum finality, Atlas makes interop latency no longer the bottleneck.”

Atlas effectively means the L2s can directly tap Ethereum’s enormous TVL, the majority of which has been reluctant to bridge funds to L2s to date.

“Historically, every L2 had to bootstrap its own liquidity, spin up separate bridges, and rebuild the same DeFi stack, which is capital-inefficient and risky. With Atlas and Gateway, ZKsync chains can tap Ethereum’s liquidity directly in near real time, instead of fragmenting it across isolated pools. L2↔L2 transactions are ~1 second, and L1↔L2 transfers now finalize faster than a single Ethereum block.”

Magazine: Big questions: Would Bitcoin survive a 10-year power outage?