Sources: EvolutionaryScale, founded by members of Meta's recently axed protein-folding team to build LLMs for biology, raised a $40M+ seed at a $200M valuation
Forbes : X: @_iainmartin and @kenrickcai X: Iain Martin / @_iainmartin : Meta shuttered it's AI project focused on protein-folding in April. Now that team have raised $40m to build new biological LLMs that could help develop everything from cancer-fighting programmable cells to organisms that clean up toxic waste https://www.forbes.com/... @kenrickcai Kenrick Cai / @kenrickcai : Scoop: Researchers from Meta AI's recently axed protein-folding team have started a new company to keep building a LLM for biology. EvolutionaryScale raised $40M+ from Lux Capital, per sources. From @_IainMartin and I, with details from a pitch doc: https://www.forbes.com/...
Context & Ripple Effects
Meta’s decision to disband its ESMFold protein-folding team created a clear organizational break between the research effort and its commercial priorities. This financing gives departing researchers an independent vehicle to continue work on language models for biology.
The story is an early marker of a field that later drew larger specialized-model efforts: EvolutionaryScale subsequently released ESM3 and raised a larger seed round, while Latent Labs also emerged to pursue programmable biology.
First-order effects
- EvolutionaryScale gains more than $40 million in seed funding to hire, train models and operate independently of Meta’s research organization.
- Meta loses direct control over a team and research direction it had wound down, while Lux Capital becomes an early backer of a biology-model company.
Second-order effects
- The raise validates a venture-backed route for researchers leaving large AI labs when internal priorities shift, increasing competition for specialized biology-AI talent.
- Other teams building scientific foundation models will face pressure to show that their models can translate from research capability into useful protein-engineering workflows.
Third-order effects
- If capital continues to fund spinouts from shuttered corporate research programs, frontier work in scientific AI may increasingly move into independent companies rather than remain inside general-purpose AI labs.
- The category’s durable winners will likely be determined less by model announcements than by whether specialized biological models can sustain costly research and prove practical utility.
The trend: Scientific foundation models are becoming a distinct venture category, with displaced corporate research teams forming focused startups around domain-specific AI.