EvolutionaryScale
Frontier biological foundation-model lab — the ESM3 protein sequence-to-structure-to-function model and successor releases — founded in July 2023 by the ex-Meta FAIR ESM research team; emerged from stealth June 2024 with ESM3 and the $142M seed and acquired by the Chan Zuckerberg Biohub Network in November 2025 on a philanthropic-research footing while continuing to ship commercial API access and a partial-open-research release model.
The Business
EvolutionaryScale is a frontier biological foundation-model lab — the maker of the ESM3 protein sequence-to-structure-to-function model and successor releases, distributed through three main surfaces: a hosted EvolutionaryScale API (developer and pharma R&D access), a partial-open-research release model on the ESM3 model weights (the canonical biology-foundation-model open-research distribution surface, established with the June 2024 bioRxiv preprint and the subsequent Science peer-reviewed publication), and platform-partner distribution through NVIDIA BioNeMo (the biological-foundation-model platform that ships ESM3 across the NVIDIA-customer biotech and pharma footprint) and Amazon Web Services. The company was founded in July 2023 by Alex Rives, Tom Sercu and Salvatore Candido; emerged publicly in June 2024 with the ESM3 release and $142M seed announcement — the ex-Meta FAIR ESM research team and authors of the most-cited biological-language-model research lineage in the industry. EvolutionaryScale was acquired by the Chan Zuckerberg Biohub Network in November 2025, bringing the ESM technology and approximately 50 employees in-house on a philanthropic-research footing while continuing to ship ESM3 and successor releases under the partial-open-research distribution model. The product family covers protein sequence-to-structure-to-function modeling (ESM3 and successor releases) as the canonical biological-foundation-model reference platform for in silico protein engineering.
Customers and Distribution
EvolutionaryScale’s customer footprint is structurally pre-commercial relative to horizontal frontier model APIs: the primary distribution channels are the hosted EvolutionaryScale API (developer and pharma R&D direct), the partial-open-research release of the ESM3 weights (which routes into academic research labs and applied-protein-engineering platforms across the biotech research footprint), and the platform-partner stack across NVIDIA BioNeMo (ESM3 integrated into the BioNeMo biological-foundation-model platform across the NVIDIA-customer biotech and pharma footprint) and Amazon Web Services (cloud and strategic-investment alignment from the June 2024 seed round). The November 2025 Chan Zuckerberg Biohub Network acquisition reshapes the commercial-GTM frame entirely — EvolutionaryScale now distributes into the CZ Biohub Network research footprint on a philanthropic-research footing, with the pharma R&D partnership cadence the closest published indicator of commercial-traction signal. EvolutionaryScale does not publish a revenue figure, and the IM tracker labels the headline as “pre-commercial (biology-FM platform stage)” rather than an ARR number; we decline-to-publish a precise stand-alone revenue figure pending a primary disclosure. Headcount is press-triangulated to a 30–60 employee range pre-acquisition, with approximately 50 employees referenced at the CZ Biohub absorption.
Model Strategy
EvolutionaryScale is a Frontier-first biological-foundation-model play with a partial-open-research / hosted-API hybrid distribution model: the strategic bet is that frontier biological foundation-model capability — protein sequence-to-structure-to-function on ESM3 and successor releases — shipped in part as open research on the model weights and in part as a hosted API alongside platform-partner distribution through NVIDIA BioNeMo and AWS, holds the category-defining research-and-distribution position against horizontal frontier model labs offering biology-adjacent capability through general-purpose APIs. The infrastructure backstop is the philanthropic-research footing established by the November 2025 Chan Zuckerberg Biohub Network acquisition — effectively unbounded runway at the model-research layer — alongside the NVIDIA strategic alignment from the original June 2024 seed and AWS cloud and strategic-investment alignment from the same round. The partial-open-research release design is intentional and lineage-driven: Rives, Sercu and Candido carried the ESM research lineage out of Meta FAIR with an open-research publication tradition, and EvolutionaryScale was founded explicitly to extend that open-research footprint into a commercially-funded — and now philanthropically-funded — frontier biology lab. The trade-off is structural customer portability — pharma R&D customers and biotech research teams can self-host the open-research-released ESM3 weights at low switching cost, and the strategic bet relies on continued capability cadence and the biology-specific moat versus horizontal frontier model substitution rather than lock-in.
At A Glance
The Numbers
Pharma partnerships available
Headcount (FTE)
Leadership Team
EvolutionaryScale is unusually founder-and-research-concentrated for a frontier model lab — the Rives / Sercu / Candido trio carries the ESM research lineage out of Meta FAIR into the EvolutionaryScale founding cohort, and the company has so far chosen to scale headcount slowly (press-triangulated 30–60 employees range pre-acquisition) rather than build out a conventional executive bench. The November 2025 Chan Zuckerberg Biohub Network acquisition brought approximately 50 employees and the ESM technology in-house on a philanthropic-research footing. CFO, CRO and CTO roles have not been publicly disclosed as separate appointments at the time of writing. Nat Friedman and Daniel Gross (who led the $142M seed) were the closest pre-acquisition board-level commercial voices in public coverage.
IM Framework Scoring
IM’s structured assessment of EvolutionaryScale’s competitive position. The summary below is the headline; expand “Show the full analyst-grade analysis” near the bottom for the per-dimension reasoning and evidence. Methodology →
Funding History
| Date | Round | Raised | Post-money | Lead investor(s) |
|---|---|---|---|---|
| Nov 2025 | Acquisition by CZ Biohub Network | — | — | Chan Zuckerberg Biohub Network (philanthropic-research absorption) |
| Jun 2024 | Seed | $142M | — | Nat Friedman & Daniel Gross co-led; Lux Capital, Amazon, NVentures (NVIDIA) participating |
Cumulative external equity is $142M through the June 2024 seed round — reported across named press as one of the largest seed rounds for an AI research lab at founding — co-led by Nat Friedman and Daniel Gross alongside Lux Capital, Amazon and NVentures (NVIDIA’s strategic-investment arm). The round was widely covered by named-author tech and biotech press (TechCrunch, Endpoints News, SynBioBeta) at announcement. The November 2025 Chan Zuckerberg Biohub Network acquisition reshaped the funding profile entirely — EvolutionaryScale is now backed on a philanthropic-research footing through CZ Biohub, with effectively unbounded runway at the model-research layer and a reorientation of the commercial-GTM frame toward research distribution. Acquisition financial terms have not been publicly disclosed.
Competitive Landscape
EvolutionaryScale’s competitive set sits in three concentric rings: frontier biological-foundation-model research lineages (AlphaFold / Isomorphic Labs as the category-defining research-and-commercial play; Profluent and Generate Biomedicines as the closest pure-play biology-foundation-model competitors), applied-protein-engineering platforms (Cradle and adjacent biotech-R&D SaaS plays), and the horizontal frontier model labs whose general-purpose APIs increasingly cover biology-adjacent capability through tool-use. EvolutionaryScale is unusual in the set because the November 2025 Chan Zuckerberg Biohub Network acquisition shifts the commercial-vs-research footing decisively — the company now sits closer to a philanthropic-research platform on biological foundation models than to a venture-funded commercial API, while continuing to ship ESM3 and successor releases under a partial-open-research distribution model.
| Competitor | Positioning | Distribution edge | Threat profile |
|---|---|---|---|
| AlphaFold / Isomorphic Labs (Alphabet (NASDAQ: GOOGL)) |
The category-defining protein-structure-prediction research lineage (AlphaFold 2 / 3, Nobel Prize-recognised research) combined with the Isomorphic Labs drug-discovery commercial arm. The brand most directly paired against ESM3 in biological-foundation-model procurement and academic discussion. | AlphaFold open release plus the AlphaFold Server hosted-API surface; Isomorphic Labs partnerships with named pharma companies (Eli Lilly, Novartis) converting research capability into drug-discovery contracts. | High and asymmetric — the most direct research-and-distribution competitor on biological foundation models; structurally larger compute base through Google DeepMind and structurally larger commercial-distribution base through Isomorphic Labs’ pharma contracts. |
| Profluent | Frontier protein-design foundation-model lab with a CRISPR / gene-editing research focus; founded by the ex-Salesforce ProGen research team. The closest pure-play competitor to EvolutionaryScale on the biological-foundation-model-as-platform thesis. | Direct research-and-commercial distribution; named pharma partnerships and a hosted-model API. | Medium-high — narrower (protein design and gene editing) but the most direct head-to-head on the biological-foundation-model-as-platform framing. |
| Generate Biomedicines | Generative-biology platform combining protein-design models with a vertically-integrated therapeutic-development pipeline; founded out of Flagship Pioneering with substantial capital base. Competitor on the biological-foundation-model layer and on the biotech commercial frame. | Direct therapeutic-development pipeline plus named pharma partnerships (Amgen, Novartis). | Medium-high — vertically integrated through to therapeutic candidates rather than purely a foundation-model API; structurally further down the commercial-revenue curve than EvolutionaryScale. |
| Cradle | Protein-engineering AI platform aimed at biotech R&D teams — an applied-protein-design SaaS layer on top of foundation models including open-research releases. The closest competitor on the applied-protein-engineering commercial surface. | Direct SaaS subscription to biotech R&D teams plus integrations with named lab-automation partners. | Medium — narrower than EvolutionaryScale (applied protein engineering rather than foundation-model research) but stickier inside its lane and a credible flanking risk on the applied-biology customer segment. |
| GPT-5 / Claude Opus / Gemini biology-adjacent (OpenAI / Anthropic / Google (NASDAQ: GOOGL)) |
Horizontal frontier model labs whose general-purpose APIs increasingly cover biology-adjacent capability (literature synthesis, sequence analysis, computational chemistry through tool-use). Not biological-foundation-model specialists, but credible substitutes for many of the biology-adjacent workloads that would otherwise route through ESM3 or comparable specialist APIs. | ChatGPT, Claude.ai, Gemini app plus the OpenAI / Anthropic / Vertex AI developer APIs; hyperscaler distribution at consumer-AI default-brand scale. | Medium — horizontal-vs-vertical risk that biology-adjacent workloads default to the cheapest available general-purpose API rather than a biology-specific platform; structurally larger compute base and distribution reach but no biological-foundation-model specialisation. |
Pricing benchmark: EvolutionaryScale does not publish per-call API pricing in line with the closed-API frontier labs. The closest pricing comparables in biological foundation models are the AlphaFold Server hosted-API surface (free-tier research access plus partner-tier commercial access) and the closed-API biology specialists at per-call rates aligned with horizontal frontier model APIs within roughly a 2x band. The competitive frame is research-capability-and-partial-open-research-release rather than headline per-call price.
Potential Risks
The case for EvolutionaryScale at IM Framework 7.61 rests on the ESM research lineage carried over from Meta FAIR, the November 2025 Chan Zuckerberg Biohub Network acquisition that takes runway to effectively unbounded philanthropic backing on top of the prior $142M seed, and the category-defining biological-foundation-model research signal from ESM3 and successor releases. The case against splits into five risks of differing magnitude — with customer portability the most structural (partial-open-research-release design is intentional) and the pre-commercial revenue profile the most active near-term constraint on the defensibility composite.
Customer portability at the open-research-release layer is structural
ESM3 model weights were largely open-research published, and the partial-open-research release model is itself the brand differentiator versus closed-API biological-foundation-model competitors. The trade-off is structural customer portability — pharma R&D customers and biotech research teams can self-host the open-research-released ESM3 weights at low switching cost, and downstream applied-protein-engineering platforms (Cradle and adjacent SaaS plays) can route to either ESM3 or alternative foundation models with limited friction at the inference layer. The sub-rubric score on D1c portability was held at 6 on this evidence: substitution risk is real and structural by design, and the strategic bet relies on continued frontier capability cadence rather than lock-in.
Pre-commercial revenue profile at the biology-FM-as-platform stage
EvolutionaryScale does not publish a revenue figure, and the biology-FM-as-platform stage is structurally pre-commercial — the pharma R&D partnership cadence is the closest published indicator of commercial-traction signal, and the November 2025 Chan Zuckerberg Biohub Network acquisition reorients the commercial-GTM frame toward philanthropic-research distribution rather than a conventional API revenue ramp. The sub-rubric score on P1d time-to-revenue stage-appropriateness was held at 5 on this evidence; the disruption composite at 8.59 is essentially priced for continued frontier-capability cadence and a credible pharma-and-biotech commercial conversion path on a longer horizon than horizontal frontier model APIs.
Foundation-model dependency on horizontal frontier labs covering biology-adjacent workloads
Horizontal frontier model labs (OpenAI GPT-5, Anthropic Claude Opus, Google Gemini) increasingly cover biology-adjacent workloads (literature synthesis, sequence analysis, computational chemistry through tool-use) through their general-purpose APIs and large-context-window reasoning. None of these substitutes for a biology-specific frontier model on the canonical protein sequence-to-structure-to-function workload, but collectively they cap the share of biology-adjacent enterprise spend that routes through specialist biological-foundation-model APIs rather than the cheapest available general-purpose API. The competitive frame is structurally horizontal-vs-vertical, and the strategic bet relies on the biology-specific capability margin holding release-by-release against horizontal frontier model substitution.
Founder concentration and bench depth at the ex-Meta-FAIR research-team scale
EvolutionaryScale is unusually founder-and-research-concentrated — the Rives / Sercu / Candido trio carries the ESM research lineage out of Meta FAIR into the founding cohort, and the company has so far chosen to scale headcount slowly (press-triangulated 30–60 range pre-acquisition; approximately 50 employees referenced at the CZ Biohub absorption) rather than build out a conventional executive bench. No CFO, CRO or CTO has been publicly disclosed as a separate appointment at the time of writing. A coordinated founder departure of the magnitude seen when the same research lineage left Meta FAIR to found EvolutionaryScale would be a material disruption to the research roadmap. The sub-rubric score on D4e key-person dependency was held at 5 on this evidence.
Regulatory and biosecurity exposure binds through downstream pharma and biotech partners
Biological foundation models face a distinctive regulatory-and-biosecurity environment binding largely through downstream pharma R&D and biotech partners rather than directly on the model layer — FDA / EMA drug-approval pathways on therapeutic candidates derived from ESM3 outputs, dual-use research-of-concern frameworks on protein-design models with potential biosecurity implications, and the EU AI Act’s general-purpose-AI obligations where the biology-FM is classified under the high-risk-application tier. None of these is fatal to EvolutionaryScale, but collectively they cap the velocity at which the biological-foundation-model layer can ship release-by-release into the most consequential therapeutic-discovery workloads. The sub-rubric score on D4c regulatory exposure was held at 9 on this evidence — favourable because the binding is downstream rather than direct, but a tracked risk inside the IM Framework cohort.
Recent IM Coverage
- Information Matters AI Tracker May 2026.
Show recent press coverage of EvolutionaryScale
- Nov 2025 — Chan Zuckerberg Biohub Network acquires EvolutionaryScale — ESM technology and approximately 50 employees brought in-house on a philanthropic-research footing.
- Jun 2024 — EvolutionaryScale launches with $142M seed round — Nat Friedman and Daniel Gross co-lead; Lux Capital, Amazon and NVentures participating.
- Jun 2024 — ESM3 release — simulating 500 million years of evolution with a language model: the canonical biological-foundation-model preprint.
- Jun 2024 — EvolutionaryScale raises $142M seed to build foundation models for biology — ex-Meta FAIR ESM research team founding announcement.
- Jun 2024 — EvolutionaryScale — foundation models for biology: $142M seed and ESM3 release.
- Jun 2024 — NVIDIA BioNeMo partnership: EvolutionaryScale ESM3 integrated into the BioNeMo biological-foundation-model platform.
- 2025 — ESM3 in Science: peer-reviewed publication of the simulating-evolution biological-language-model research.
Curated feed of named-source coverage — EvolutionaryScale’s own newsroom and blog, the bioRxiv preprint and Science peer-reviewed publication of the ESM3 research, the Chan Zuckerberg Biohub Network acquisition announcement, and named-author biotech and tech press (Endpoints News, SynBioBeta, NVIDIA Blog). Excludes paywalled article bodies of The Information, WSJ, FT and Bloomberg (headline + free-snippet only), wire-aggregator reposts and unsourced biology-AI round-up pieces.
Show the source register for the figures on this page
IM operates a primary-source-where-possible discipline. The figures above come from:
- Revenue (basis-disclosure note): EvolutionaryScale is private and does not publish a revenue figure; the biology-FM-as-platform stage is structurally pre-commercial and the November 2025 Chan Zuckerberg Biohub Network acquisition reorients the commercial-GTM frame toward philanthropic-research distribution rather than a conventional API revenue ramp. We decline-to-publish a precise revenue figure pending a primary disclosure and reference the pre-commercial profile only with the caveat that the pharma R&D partnership cadence is the closest published indicator of commercial-traction signal.
- Usage — ESM3 research and partner distribution: ESM3 was released as a partial-open-research model with weights distributed for research use alongside the bioRxiv preprint and subsequently published in Science as the canonical biological-language-model peer-reviewed reference. The NVIDIA BioNeMo integration distributes ESM3 into the BioNeMo biological-foundation-model platform across the NVIDIA-customer biotech and pharma footprint. EvolutionaryScale does not publish a precise download or API-call usage figure; we decline-to-publish a precise number pending a primary disclosure.
- Headcount (basis-disclosure note): EvolutionaryScale does not publish precise headcount. Named press across the June 2024 founding and the November 2025 Chan Zuckerberg Biohub Network acquisition triangulates a 30–60 employee range pre-acquisition, with approximately 50 employees referenced at the CZ Biohub absorption announcement. We decline-to-publish a precise figure pending a primary disclosure and reference the 30–60 range only with the caveat that it is press-triangulated, not company-disclosed.
- Funding to date: Cumulative external equity $142M through the June 2024 seed round — one of the largest seed rounds for an AI research lab at founding — co-led by Nat Friedman and Daniel Gross alongside Lux Capital, Amazon and NVentures (NVIDIA’s strategic-investment arm). References: Endpoints News $142M seed coverage; SynBioBeta $142M seed coverage; EvolutionaryScale founding announcement. The November 2025 Chan Zuckerberg Biohub Network acquisition reshaped the funding profile entirely — EvolutionaryScale is now backed on a philanthropic-research footing through CZ Biohub, with effectively unbounded runway at the model-research layer. Acquisition financial terms have not been publicly disclosed.
Methodology & Disclaimer
For metric definitions, source-tier hierarchy, and decline-to-publish rules, see the tracker methodology. Confidence dots (• green / • amber / • red) follow the same convention as the AI Tracker.
Spotted a figure you believe is wrong? Send corrections to info@informationmatters.net.
Information Matters Framework scores are the considered opinion of the IM team — human and AI — applied to publicly-available evidence under a disclosed methodology. They are not statements of fact about the companies scored and they are not investment advice.
