He left machine learning to do the job nobody wanted: cleaning up blockchain data. Now Visa calls when the numbers have to be right.
Ethan Chan, co-founder & CEO of Allium. The man behind the curtain calls himself the plumber.
Ask most founders what they do and you get a manifesto. Ask Ethan Chan and you get a job title he made up to be funny: Chief Data Plumber. It is a joke that happens to be the most honest description of Allium in the building. Pipes carry things you never think about until they leak. Chan spends his days making sure the blockchain pipes do not.
Allium ingests raw events from more than 150 blockchains and 10,000 protocols, verifies them, and turns the mess into standardized, finance-ready datasets. The output is dull by design. A bank does not want a thrilling number. It wants the correct one, every time, reconciled across chains that fork, reorganize, and disagree with each other by the second. Chan built a company on the premise that the boring work is the valuable work, precisely because almost nobody wants to do it.
The customers tell the story better than any pitch deck. Visa uses Allium for its Onchain Analytics Dashboard, the one tracking stablecoin active users, volumes, and transaction sizes. Stripe is a customer. So are Coinbase, Uniswap, Phantom, and Grayscale. These are companies that could build the plumbing themselves. Blockchain data is public, after all. They pay Chan instead because doing it right is a full-time act of maintenance, not a one-time build.
Before crypto, Chan was deep in artificial intelligence. He took an MS in Computer Science at Stanford after a BS at the University of Illinois at Urbana-Champaign, did research at the Stanford Intelligent Systems Lab, and lectured machine learning to graduate students. Teaching a room full of Stanford students how models actually work is a strong filter for whether you understand the thing yourself.
At Primer he became Director of Engineering, where he founded and led the Primer Command product and re-architected the core ML pipelines that powered the company's analysis tools. The work was real-time machine learning for Fortune 500 companies, big tech, and governments handling mission-critical security and intelligence problems. The thread running through all of it was the same lesson, learned the hard way: a model is only as good as the data underneath it, and the data is almost always worse than anyone admits.
In 2021 Chan and co-founder Cheng Han Lee looked at finance moving on-chain and saw the same data problem in a new costume. Public blockchains were a terrible system of record. Information was fragmented, schemas were inconsistent, teams kept rebuilding brittle infrastructure to manage RPC connections, normalize data, and reconcile chain reorgs. Two engineers who had spent careers wrangling data systems at scale decided the fix should exist as a product, not as in-house pain repeated at every firm. Allium was born to become the system of record for onchain data.
In late 2025 Chan sat down with Ted Merz, who spent thirty years at Bloomberg, at Two Hands in Nolita. The comparison practically wrote itself, and Chan leaned into it. Bloomberg abstracted away the grueling work of collecting and cleaning financial data so traders never had to think about where a number came from. Allium does the same for the chain.
The logic is identical. Yes, a firm could assemble its own blockchain data team. Yes, the raw data is free for the taking. But the upkeep never ends, and the cost of a wrong number compounds. Outsourcing the plumbing is not laziness, it is arithmetic. Chan's bet is that as more money moves on-chain, the number of institutions willing to maintain that infrastructure themselves trends toward zero, and the number that need a trustworthy feed trends toward all of them.
Allium Explorer for query and visualization; Allium Developer for real-time ingestion into high-performance apps.
Dashboards, a query interface, flexible APIs, and custom data shares and streams. Data however you need it.
Reconciling chain reorgs, normalizing schemas, and keeping RPC connections alive across 150+ chains.
Standardized, verified, finance-ready data institutions can actually trust as a system of record.
Allium raised a seed round in 2022 and then a $16.5M Series A in July 2024 led by Theory Ventures, with Kleiner Perkins and Amplify Partners alongside. That brings the total to $21.5M. The plan for the money is unglamorous in the most on-brand way possible: infrastructure investment to improve the customer experience, and building launchpads for financial firms that want to engage with digital currencies and tokens without drowning in raw data first.
Seed figure approximate; total raised $21.5M. Led by Theory Ventures, with Kleiner Perkins & Amplify Partners.
When the companies most paranoid about data accuracy outsource their data, that is the endorsement that matters.
Allium runs on a culture Chan helped write, and it reads like a portrait of how he works. A pro-athlete mindset of dedication and discipline. Figure it out and extreme ownership. High agency. Leading from the front. Strong opinions, loosely held. Read together, they describe someone who would rather act than wait for permission, then change his mind fast when the evidence turns.
The company is split across New York, Singapore, and remote, more than fifty people keeping the feeds clean while the rest of crypto chases the next narrative. Chan's quirk is that he has chosen to be unfashionable on purpose. While founders pitch the future of finance in soaring language, he describes his own work as boring, and means it as the highest compliment. The plumbing either works or it does not. His does.
His self-styled title is Chief Data Plumber, and it stuck.
He lectured machine learning at Stanford before founding a crypto data company.
Allium powers the stablecoin metrics inside Visa's Onchain Analytics Dashboard.
His prior work at Primer served Fortune 500s, big tech, and governments on security and intelligence problems.