Ethereum Enters a New Era of 10,000 TPS—Powered by Zero-Knowledge Proofs
Ethereum is marching toward a new scaling milestone: 10,000 transactions per second (TPS). At the heart of this leap is zero-knowledge (ZK) technology.

On July 30, 2025—Ethereum’s 10th birthday—the community received more than nostalgic tweets. After a decade of experiments, the network’s scaling vision has a clear new direction.

Yes, ETH’s recent price surge helps morale, but the real excitement is technical: after years of relying on Layer-2 work-arounds, Ethereum’s Layer-1 finally has a credible path to extreme throughput without sacrificing decentralization.

In short, the gas limit and TPS will rise several-fold every year. Validators will stop re-executing every transaction and instead verify a single ZK-proof that attests the entire batch is correct. The base layer is on track for 10,000 TPS.

Layer-2s will scale in tandem, handling hundreds of thousands—perhaps millions—of TPS. A new flavor of L2 called “native rollups” will act like programmable shards, inheriting the full security of Layer-1.

None of these proposals are final, but they build on ideas Vitalik Buterin first sketched in 2017 and are now championed by Ethereum Foundation researcher Justin Drake. Speaking at EthCC in July, Drake said:

“We’re at an inflection point for Ethereum scaling. I’m convinced we’re entering the L1 gigagas era—roughly 10,000 TPS. The key unlock is zkEVMs and real-time proving.”

Drake’s endgame: 10 million TPS for the entire Ethereum ecosystem within ten years. No single chain can hit that alone. The future is a “network of networks”—a constellation of Layer-2s, each optimized for different use-cases, trade-offs, and user needs.

01. Why Has Ethereum Layer-1 Struggled to Scale?
While other blockchains crank throughput by throwing bigger servers at the problem, Ethereum has clung—some say stubbornly—to decentralization.

To ETH maxis, “datacenter chains” like Solana present million-dollar attack surfaces: one subpoena, one raid, and censorship becomes trivial. Even lower-spec chains such as Sui impose bandwidth and hardware costs that price out home stakers.

Ethereum, by contrast, can run on a Raspberry Pi. That low bar keeps 15,000–16,000 public nodes and a million validators in the game. Censoring transactions is nearly impossible, and the network remains resilient. (After 50–70 % of validators complied with OFAC sanctions on Tornado Cash, the “little guys” kept processing those transactions anyway.)

The trade-off is speed: Ethereum currently processes 18–20 TPS; Solana does ~1,500.

Blockchains are intentionally inefficient—think of a Google Sheet that forces every copy-holder to recalculate the entire workbook before you can edit a single cell. This design keeps the barrier to entry low but caps raw throughput. As Succinct Labs co-founder Uma Roy puts it:

“Ethereum wants anyone to be able to re-execute the chain. That overhead prevents us from cranking the TPS dial indefinitely.”

Unable to scale L1 without compromising decentralization, Ethereum pivoted to the (often-criticized) Layer-2 roadmap in 2020.

02. How ZK Breaks the Blockchain “Impossible Triangle”
Vitalik coined the term “blockchain trilemma” to describe the tension between security, scalability, and decentralization. Most scaling solutions pick two and sacrifice the third—until now.

Zero-knowledge proofs, the “moon math” Drake references, make it possible to prove that a bundle of complex transactions was executed correctly—without revealing the transactions themselves.

Generating the proof is computationally heavy, but verifying it is lightning-fast and lightweight.

So instead of asking a swarm of Raspberry Pis to replay every opcode, the network only needs to check a tiny ZK-proof. Roy summarizes:

“Don’t re-execute—just hand validators a proof that the work was done right. Anyone can verify the proof without redoing the computation.”

Drake jokes that even a $7 Raspberry Pi Pico—one-tenth the power of its bigger sibling—will suffice to validate proofs.

03. The zkEVM: A Roadmap to 10,000 TPS
Ethereum Foundation researcher Sophia Gold recently set the community abuzz: a ZK-powered Ethereum Virtual Machine (zkEVM) could land on Layer-1 within a year.

Much of the practical groundwork has come from Layer-2 pioneers. Linea—spun out of ConsenSys, co-founded by Joe Lubin—is a 100 % EVM-equivalent ZK-rollup. Any dApp that runs on Ethereum runs unchanged on Linea, which sees itself as an extension of the mainnet. Linea even burns 20 % of its ETH fees to funnel value back to L1.

Declan Fox, head of Linea, explains the trilemma fix:

“ZK lets us raise the L1 gas limit by orders of magnitude. Computation scales while verification stays trivial—eventually on a smartwatch.”

But temper expectations: even if zkEVM ships within a year, 10,000 TPS won’t appear overnight.

04. Slow, Steady—Then All at Once
Ethereum runs on five major client implementations. If one crashes, the network keeps ticking (unlike certain monolithic chains).

The upgrade path looks like this:

1. Two or three clients release ZK-ready versions.

2. A small cohort of validators opts in, checking proofs instead of re-executing blocks.

3. Bugs get squashed, confidence grows, and more validators migrate.

Ladislaus from the Foundation’s protocol-coordination team calls it “a gradual transition to a snarkified EVM.” Users will first notice higher gas limits and cheaper L1 activity.

Last week the gas limit already bumped 22 % to 45 million. Dankrad Feist has floated an EIP that would auto-raise the limit three times a year; four years of that cadence yields ~2,000 TPS. Drake pushes the timeline two extra years to reach 1 gigagas—roughly 10,000 TPS—by 2031.

What Comes Next
Real-time proving—generating a proof for every 12-second block—remains the next big milestone. Succinct’s SP1 HyperCube zkVM already proves 93 % of mainnet blocks in real time on a 200-GPU cluster, and the team expects 99 % coverage by year-end. Hardware targets for provers sit under $100 k and 10 kW—garage-scale rigs, not datacenters.

Meanwhile, the Beam Chain upgrade will overhaul Ethereum’s consensus layer to be “ZK-friendly from day one,” Drake says. If native rollups follow, validators won’t just verify the L1—they’ll also certify Layer-2 state transitions, making a rollup as safe as Ethereum itself.

The Merge rewrote Ethereum’s consensus engine. Snarkifying the EVM could be just as transformative—and the gears are already turning.

Ethereum Enters a New Era of 10,000 TPS—Powered by Zero-Knowledge Proofs
Ethereum is marching toward a new scaling milestone: 10,000 transactions per second (TPS). At the heart of this leap is zero-knowledge (ZK) technology.

On July 30, 2025—Ethereum’s 10th birthday—the community received more than nostalgic tweets. After a decade of experiments, the network’s scaling vision has a clear new direction.

Yes, ETH’s recent price surge helps morale, but the real excitement is technical: after years of relying on Layer-2 work-arounds, Ethereum’s Layer-1 finally has a credible path to extreme throughput without sacrificing decentralization.

In short, the gas limit and TPS will rise several-fold every year. Validators will stop re-executing every transaction and instead verify a single ZK-proof that attests the entire batch is correct. The base layer is on track for 10,000 TPS.

Layer-2s will scale in tandem, handling hundreds of thousands—perhaps millions—of TPS. A new flavor of L2 called “native rollups” will act like programmable shards, inheriting the full security of Layer-1.

None of these proposals are final, but they build on ideas Vitalik Buterin first sketched in 2017 and are now championed by Ethereum Foundation researcher Justin Drake. Speaking at EthCC in July, Drake said:

“We’re at an inflection point for Ethereum scaling. I’m convinced we’re entering the L1 gigagas era—roughly 10,000 TPS. The key unlock is zkEVMs and real-time proving.”

Drake’s endgame: 10 million TPS for the entire Ethereum ecosystem within ten years. No single chain can hit that alone. The future is a “network of networks”—a constellation of Layer-2s, each optimized for different use-cases, trade-offs, and user needs.

01. Why Has Ethereum Layer-1 Struggled to Scale?
While other blockchains crank throughput by throwing bigger servers at the problem, Ethereum has clung—some say stubbornly—to decentralization.

To ETH maxis, “datacenter chains” like Solana present million-dollar attack surfaces: one subpoena, one raid, and censorship becomes trivial. Even lower-spec chains such as Sui impose bandwidth and hardware costs that price out home stakers.

Ethereum, by contrast, can run on a Raspberry Pi. That low bar keeps 15,000–16,000 public nodes and a million validators in the game. Censoring transactions is nearly impossible, and the network remains resilient. (After 50–70 % of validators complied with OFAC sanctions on Tornado Cash, the “little guys” kept processing those transactions anyway.)

The trade-off is speed: Ethereum currently processes 18–20 TPS; Solana does ~1,500.

Blockchains are intentionally inefficient—think of a Google Sheet that forces every copy-holder to recalculate the entire workbook before you can edit a single cell. This design keeps the barrier to entry low but caps raw throughput. As Succinct Labs co-founder Uma Roy puts it:

“Ethereum wants anyone to be able to re-execute the chain. That overhead prevents us from cranking the TPS dial indefinitely.”

Unable to scale L1 without compromising decentralization, Ethereum pivoted to the (often-criticized) Layer-2 roadmap in 2020.

02. How ZK Breaks the Blockchain “Impossible Triangle”
Vitalik coined the term “blockchain trilemma” to describe the tension between security, scalability, and decentralization. Most scaling solutions pick two and sacrifice the third—until now.

Zero-knowledge proofs, the “moon math” Drake references, make it possible to prove that a bundle of complex transactions was executed correctly—without revealing the transactions themselves.

Generating the proof is computationally heavy, but verifying it is lightning-fast and lightweight.

So instead of asking a swarm of Raspberry Pis to replay every opcode, the network only needs to check a tiny ZK-proof. Roy summarizes:

“Don’t re-execute—just hand validators a proof that the work was done right. Anyone can verify the proof without redoing the computation.”

Drake jokes that even a $7 Raspberry Pi Pico—one-tenth the power of its bigger sibling—will suffice to validate proofs.

03. The zkEVM: A Roadmap to 10,000 TPS
Ethereum Foundation researcher Sophia Gold recently set the community abuzz: a ZK-powered Ethereum Virtual Machine (zkEVM) could land on Layer-1 within a year.

Much of the practical groundwork has come from Layer-2 pioneers. Linea—spun out of ConsenSys, co-founded by Joe Lubin—is a 100 % EVM-equivalent ZK-rollup. Any dApp that runs on Ethereum runs unchanged on Linea, which sees itself as an extension of the mainnet. Linea even burns 20 % of its ETH fees to funnel value back to L1.

Declan Fox, head of Linea, explains the trilemma fix:

“ZK lets us raise the L1 gas limit by orders of magnitude. Computation scales while verification stays trivial—eventually on a smartwatch.”

But temper expectations: even if zkEVM ships within a year, 10,000 TPS won’t appear overnight.

04. Slow, Steady—Then All at Once
Ethereum runs on five major client implementations. If one crashes, the network keeps ticking (unlike certain monolithic chains).

The upgrade path looks like this:

1. Two or three clients release ZK-ready versions.

2. A small cohort of validators opts in, checking proofs instead of re-executing blocks.

3. Bugs get squashed, confidence grows, and more validators migrate.

Ladislaus from the Foundation’s protocol-coordination team calls it “a gradual transition to a snarkified EVM.” Users will first notice higher gas limits and cheaper L1 activity.

Last week the gas limit already bumped 22 % to 45 million. Dankrad Feist has floated an EIP that would auto-raise the limit three times a year; four years of that cadence yields ~2,000 TPS. Drake pushes the timeline two extra years to reach 1 gigagas—roughly 10,000 TPS—by 2031.

What Comes Next
Real-time proving—generating a proof for every 12-second block—remains the next big milestone. Succinct’s SP1 HyperCube zkVM already proves 93 % of mainnet blocks in real time on a 200-GPU cluster, and the team expects 99 % coverage by year-end. Hardware targets for provers sit under $100 k and 10 kW—garage-scale rigs, not datacenters.

Meanwhile, the Beam Chain upgrade will overhaul Ethereum’s consensus layer to be “ZK-friendly from day one,” Drake says. If native rollups follow, validators won’t just verify the L1—they’ll also certify Layer-2 state transitions, making a rollup as safe as Ethereum itself.

The Merge rewrote Ethereum’s consensus engine. Snarkifying the EVM could be just as transformative—and the gears are already turning.