Albert Invent raises $55M Series B to build chemistry-native AI for industrial R&D

Albert Invent raises $55M Series B for chemistry-native AI

Albert Invent, the chemistry-focused AI company founded by Nick Talken, has closed a $55M Series B led by TCV. The round comes alongside the launch of a new frontier retrieval model that Talken claims is 45x better than OpenAI and Voyage on chemistry tasks.

The core pitch is that general-purpose models trained on the internet are good at broad domain classification but fall apart at the granular distinctions that industrial chemistry requires — differentiating, say, a nanoparticle encapsulant in a paint coating from an entirely different compound class. That gap is what Albert is building into.

“At Albert, we make chemistry native AI. We're helping scientists in the lab with the white coats invent faster. As part of this announcement, we're announcing a new frontier retrieval model that we trained — it's 45x better than OpenAI and Voyage. Customers have more tokens of chemistry information in their single enterprise than the entire public corpus of chemical information.”

The data moat

The structural advantage Talken describes is in proprietary failure data. Published science captures successes; failures stay in paper notebooks. Albert's customers, he says, hold more tokens of chemistry information in a single enterprise than the entire public corpus of chemical data available online. That asymmetry is what makes the 45x retrieval improvement defensible, at least as long as the publicly available training data for foundation models remains skewed toward documented successes.

The lab-in-the-loop problem

Chemistry AI faces a constraint that software and math do not: you can't close the feedback loop on a computer. Every hypothesis still has to go through a physical experiment, which slows the reinforcement cycle that lets coding and math models self-improve rapidly. Talken frames the shift as chemists moving from doing lab work to architecting experimental systems, with AI executing the underlying work. Whether that framing survives contact with the most senior researchers at hundred-year-old chemical companies — which make up much of Albert's customer base — is an open question.

The market

Talken resists a single holy grail, pointing instead to the breadth of open problems: ablative materials for deep-space vehicles, longer-lasting batteries, biodegradable plastics, PFAS remediation. The pitch is that accelerating the pace of chemistry R&D across all of those domains is the leverage point, and that without better tools, those problems won't be solved fast enough.