Slicing Ethereum's 12-Second Block Into 100 Pieces

Slicing Ethereum's 12-Second Block Into 100 Pieces

October 31, 2026
7 min read
ethereumblockspacepreconfirmationsmevvalidatorsinstitutional-defiai-agents

Kevin Lepsoe started with a time zone correction: "literally good morning from Hong Kong." The founder of ETHGas was recording on a Friday that host Narb described as a whiplash day for crypto. The morning brought news involving Hyperliquid and the New York Stock Exchange. A few hours later came reports that Polymarket had been hacked, and by the time they spoke Binance was halting deposits and withdrawals.

Listen on your favorite platform

View full episode details

Lepsoe calls himself "a bit of an ETH maxi," and he saw the string of exploits as a question of which targets are easier to hit. As AI gets better at finding common exploits and spamming them everywhere, he argued, centralized institutions with hundreds or thousands of employees and systems built 5, 10, or 20 years ago offer more entry points than the Ethereum network itself. "I'm not going to say it's not hackable, but it's far stronger than many of these institutions," he said.

Vancouver, Lehman, and Hong Kong

Lepsoe was born and raised in Vancouver and spent 10 to 12 years in traditional finance. He started as an FX options and volatility trader, then moved into interest rate options, credit, and correlation products. He worked through the Lehman collapse in 2008, which he called a pleasure "or displeasure, depending how you look at it." After that he and his wife built and ran a tech company. Living in Hong Kong 10 to 15 years ago, when many early Bitcoin figures and large exchanges were based there, made crypto hard to ignore. Back then it still went by the name fintech.

Scale is what drew him in. In his telling, a tech product gets global distribution the moment it ships, and a new trading strategy can be rolled out across markets at once. Put the two together, he said, and they "amplify one another."## A Billion-Dollar Order and a Searcher

Narb asked how a transaction actually lands on Ethereum, and Lepsoe walked through the process. Validators, run by perhaps tens of thousands of operators, authorize a new block every 12 seconds. Users pay gas fees to a handful of block builders, maybe half a dozen, who compete to put together the most valuable candidate block. The validator picks the one that pays best and proposes it to the network. He compared this to energy, with oil pumped in the Middle East, off the coast of Norway, in Calgary, or in Texas and sent through a pipeline before it reaches anyone who needs it. ETHGas, he said, sits inside that pipeline.

That same structure is where maximal extractable value, or MEV, comes from. Lepsoe gave a hypothetical using BlackRock: an institution places a billion-dollar order on-chain, and a searcher in what he called "a very deep, dark corner" of the network spots it and front-runs it. In traditional markets, firms compete on latency by moving their infrastructure closer to the exchange or using lasers and microwave towers, and he considers that fair play. On-chain, he said, the gap between the institution and the searcher is too wide.

The second barrier is where the fees end up. A trade might cost 5 or 10 cents, but the institution still has to know who gets that money.

Is it going to Narb, this upstanding citizen here, or is it going to this dodgy person in some sanctioned country? In which case, like, I just can't do that, right?

ETHGas calls its response institutional blockspace. Lepsoe said it removes front-running and sandwiching by changing how blocks are built, and it can limit payments to American validators, for example. He also described a separate initiative in Japan where payments might go only to Japanese and American validators. ETHGas documentation lists sanctions screening before inclusion, ordering controls, and identifiable validator operators. These are the company's proposed mechanisms, not an independently verified guarantee of compliance. The alternative, as Lepsoe described it, is a bank starting its own chain with no liquidity, which "may as well be a database."

Cutting a Block Into 100 Pieces

Lepsoe's starting point was that validators set up their software, press a button, and "go get a coffee perpetually," even though they control the one thing Ethereum sells. "It's like Ethereum has one product, that's blockspace. It doesn't sell anything else," he said. If global financial rails end up running on Ethereum, he expects people to fight over that capacity. ETHGas began with the idea that blockspace should be tradable and hedgeable the same way oil and gold are.

The next step was to refine it "into like jet fuel." He described the offer to validators this way:

Why don't you give it to us? We will take your block, cut it into maybe 100 pieces, and we will sell off or execute trades every 100 milliseconds or 200 milliseconds.

Each slice comes with a preconfirmation, which is an early commitment that a transaction will be included. Lepsoe said this "synthetically" makes the network feel about 100 times faster, so a MetaMask trade is done as soon as the user lifts a finger. That makes confirmation feel faster to the user, but it does not change Ethereum's consensus finality, and no independent benchmark of the 100x figure turned up. The company's documentation describes a centralized limit order book that matches buyers and sellers, plus a noncustodial smart contract that holds validator collateral.

Lepsoe said the basic version is permissionless. A validator connects to the ETHGas relay and registers its keys. Ethereum announces the next 64 proposers about 12 to 13 minutes ahead, so ETHGas knows when a participating validator's slot is coming and can turn that slot into a low-latency block. Validators connect through MEV-Boost or a module for Commit-Boost. On the trading side, an RPC provider can connect to get Ethereum's synthetic state every 100 to 200 milliseconds instead of every 12 seconds.

Eighty Percent of the Network

The hard part, Lepsoe said, "was less about technical challenges and more a coordination effort." Base effectively runs a single sequencer and can change how it builds blocks whenever it wants. Ethereum has thousands of independent operators, and each one wants to know whether a validator earning $50 per block will now earn $51. ETHGas presented its sub-second approach at Devconnect in Buenos Aires about a year before the recording, citing 50 to 60 milliseconds. Lepsoe said it then took a year of meetings, first with validators holding 3 percent or 8 percent of the network and eventually with groups representing perhaps 80 percent, before the economics made sense to them. In April 2026 ETHGas reported a three-year agreement with ether.fi covering about 40 percent of ether.fi's ETH holdings, described as a $3 billion commitment.

His case to validators was about capital moving to L2s, Robinhood, or Canton Network. "You don't get fired for choosing Ethereum, really, right? But you get questioned over, well, it is very slow, we should also try something else," he said. He added that Ethereum's capacity doubled over the past year and could triple again with the Glamsterdam upgrade, which would push fees toward fractions of a cent.

The same reasoning is behind ETHGas's support for x402, which lets AI agents buy preconfirmations using signed HTTP requests and WETH. An agent working at machine speed should not have to wait 12 seconds, Lepsoe said, and it should trade "right downtown at the major exchange" rather than out in the suburbs.

Share this post

Share on X