Recommended Articles
Fundraising

Albert Invent Raises $55M Series B to Give R&D Labs a Chemistry-Native AI

By Angeled · October 15, 2026 · Fundraising
Albert Invent Raises $55M Series B to Give R&D Labs a Chemistry-Native AI
Albert Invent Raises $55M Series B to Give R&D Labs a Chemistry-Native AI

Company Background

Albert Invent, based in Oakland, California, was founded in 2022 by Nick Talken, the former head of digital R&D at Henkel. The company's founding insight came from direct experience: enterprise AI built on general-purpose language models consistently fails inside chemistry and materials-science R&D, because scientific language is deeply specialized. Two formulations can read almost identically but behave very differently in practice — a subtlety that generic models miss.

Albert's answer is what it calls the first "chemistry-native AI operating system" for R&D. Rather than wrapping a general model with a retrieval layer, Albert built its intelligence from the ontology up: the data models, workflows, and reasoning structures are designed for chemistry from the start. Its specialized model, Valence, was trained on 15 million chemical structures and the chemical patent corpus — described as covering nearly the entire public information space around chemistry.

The customer list reads like a who's who of industrial chemistry: BASF, Henkel, Clorox, Kenvue, AkzoNobel, Avient, Chemours, and Nouryon. Traction figures are striking: project creation on the platform grew 26x year over year, the scientist user base rose 6x across 44 countries, and workflow creation grew 4x.

Round Details

On October 13, 2026, Albert announced a $55 million Series B led by TCV. New investors included Delta-v Capital, Socium Ventures, HighSage Ventures, Coral Capital, and Legion VC, joined by existing backers Coatue, J.P. Morgan Growth Equity Partners, Index Ventures, and Applied Ventures. Forbes reported a $340 million valuation attributed to one unnamed source — treat that figure as reported, not company-confirmed.

CEO Nick Talken has pointed to a simple, expensive problem the company solves: about 30% of all chemistry experiments are "duplicate" and wasteful, because most are still recorded in paper notebooks and spreadsheets that make institutional knowledge nearly impossible to track over time.

The Technology: FormBench and Valence

Two developments make this round technically interesting:

FormBench. Albert created a benchmark for formulation chemistry that sets a global standard for evaluating AI in the field. In an industry where vendor claims are hard to verify, publishing a benchmark is both a marketing move and a genuine contribution — it forces the conversation onto measurable ground and, not coincidentally, establishes the company as the arbiter of quality.

Valence's 45x claim. Albert says Valence demonstrated 45 times greater accuracy than general-purpose AI retrieval models in identifying correct answers to scientific queries under its own benchmark methodology. That figure comes from the company and should be read with the usual caution. But even discounted heavily, the direction is clear: domain-specialized models dramatically outperform general models in fields with specialized ontologies.

The enterprise features matter equally: zero data retention, fine-grained access controls, and reasoning audit trails. For R&D departments handling trade-secret formulations, these are not nice-to-haves — they are procurement requirements.

Why It Matters for Founders and Investors

  1. Vertical AI is the clearest funding pattern of 2026. Generalist AI platforms are being squeezed between frontier labs and open-source models. The money is flowing to companies that own a domain's ontology — chemistry, law, drug discovery, engineering. For founders: if your AI product works "for everyone," you may have a thinner moat than a competitor that owns one industry's language and data.
  1. Benchmarks are becoming moats. Publishing FormBench was a strategic masterstroke: it makes Albert the standard-setter in its field and forces competitors to compete on Albert's turf. Founders in vertical AI should consider whether a benchmark or evaluation framework could anchor their market position.
  1. Institutional knowledge is the next data frontier. The observation that 30% of experiments are wasteful duplicates is a specific instance of a general truth: most large organizations have enormous, unsearchable knowledge trapped in notebooks, inboxes, and spreadsheets. AI products that surface institutional knowledge — without compromising security — are attracting serious capital because the ROI math is straightforward: recover the waste.

Building vertical AI for a specialized industry? Angeled.com connects founders with investors.

Company press release via albertinvent.com, October 13, 2026; citybiz.co, October 13, 2026; FinSMEs, October 14, 2026; aiweekly.co (reporting Forbes valuation figures), October 14, 2026.

Read Next