Next™ BriefAI Foundation Models for Biology and the Future of R&D
Buy Now
Meticulous Next™HealthcareOct 2026116 ppMRN-1065

Protein and DNA Language Models Market Outlook 2026–2036: Market Size, Growth Drivers, Key Players, Strategic Developments & Adoption Forecast for Protein Structure and Design Models, Genome Language Models, Generative Antibody and Enzyme Design, AI-Discovered Drugs in the Clinic and Pharma Partnerships — A Meticulous Next™ Foresight Brief

Brief ID: MRN-1065Format: PDF + Summary DeckDelivery: InstantHorizon: 10-yr horizonSignal: High-impact
Buy Now · $850
Adoption maturity (indexed)
Mainstream inflection: ~2030
Horizon: 2026–2036 · Signal: High-impact
10 yrs
Forward horizon
~2030
Mainstream inflection
High impact
Signal strength

What This Brief Covers

This Meticulous Next™ brief examines how AI foundation models for biology — large models trained on protein sequences and structures, genomes and experimental data that predict how molecules fold and interact and generate new proteins, antibodies, enzymes and DNA constructs — will change the cost, speed and success rate of pharmaceutical and biotechnology R&D over the next ten years. The field has crossed from prediction into product. AlphaFold 3 extended structure prediction to protein–small-molecule interactions, Evo 2 and ESM3 brought genome-scale and generative protein models, and the companies built on them have reached the clinic: a peer-reviewed analysis at ASCO 2026 counted 117 AI-enabled drugs from 63 companies in interventional trials, with phase 1 success of 80–90% against a historical 40–65% and phase 2 success near the historical 40%; Generate Biomedicines' AI-engineered antibody golukibart entered phase 3 in severe asthma in December 2025 with about 1,600 patients; Insilico Medicine's rentosertib, which improved lung function by 98.4 mL against a 20.3 mL decline on placebo in phase 2a, dosed its first phase 3 patient in September 2026; and Recursion's REC-4881 cut polyp burden by 53% at 25 weeks. Capital has followed: Isomorphic Labs raised USD 2.1 billion in May 2026, Earendil Labs USD 787 million, Generate listed on Nasdaq for USD 400 million and Insilico in Hong Kong; Eli Lilly signed Insilico for up to USD 2.75 billion and Profluent for up to USD 2.25 billion in AI-designed recombinases, Merck KGaA signed Valo Health for more than USD 3 billion in milestones and Lilly and NVIDIA committed up to USD 1 billion to an AI lab. No AI-discovered drug has yet been approved. The brief maps the technology, its indicative market size and forecast, the factors behind its growth, the developments of the last 24 months, the key players operating in the space and the adoption trajectory to 2036.

It is a focused 116-page decision brief for pharmaceutical and biotechnology R&D leaders, AI-native drug discovery companies, contract research organisations, cloud and compute providers, data companies, regulators and investors. It presents an indicative trajectory rather than a segmented market model. Its purpose is to identify which model classes and applications create value first, how clinical readouts and approvals change the pace, and who captures the resulting value.

Brief Snapshot
ParameterDetails
Forward horizon2026–2036 (10 years)
Emerging forceAI foundation models for biology: protein structure and interaction models, protein and antibody language and diffusion models for generative design, genome and DNA language models, multimodal models combining sequence, structure, imaging and experimental data, and the platforms, data, compute and partnerships through which they enter pharmaceutical and biotechnology R&D
Technology readinessDeployed across pharma for structure prediction and target assessment; generative design in clinical use with an AI-engineered antibody in phase 3 and AI-discovered small molecules in phase 2 and 3; 117 AI-enabled drugs in interventional trials; genome language models in research and first commercial applications; frontier model developers (Isomorphic, EvolutionaryScale, Profluent, Chai, Latent Labs, Arc Institute) licensed or partnered with pharma; no approved AI-discovered drug
Indicative market size & forecastUSD 1.5–2.0 billion in 2026 (biology foundation model and platform software and services revenue, partnership upfronts and recognised milestone value, model licensing, biological data licensing and dedicated compute for biology models, excluding the sales value of resulting drugs), rising to USD 18–25 billion by 2036; indicative CAGR 28–30% over 2026–2036
Mainstream inflection~2030, when the first AI-discovered and AI-engineered drugs hold approval, phase 2 and 3 success rates for AI-enabled assets are measured across dozens of readouts, foundation models are standard in target, antibody and enzyme design across large pharma and genome models guide gene therapy and editing design
Signal strengthHigh-impact — Isomorphic Labs USD 2.1 billion Series B (May 2026); Lilly–Insilico up to USD 2.75 billion with USD 115 million upfront (March 2026); Lilly–Profluent up to USD 2.25 billion for AI-designed recombinases (April 2026); Merck KGaA–Valo Health over USD 3 billion in milestones (November 2025); Generate golukibart phase 3 start (December 2025) and USD 400 million IPO (February 2026); Insilico rentosertib phase 3 first patient dosed (September 2026); ASCO 2026 count of 117 AI-enabled drugs in trials; Lilly–NVIDIA AI lab up to USD 1 billion (January 2026)
Primary beneficiariesAI-native discovery companies with clinical assets and pharma partnerships; frontier biology model developers; large pharma with in-house platforms and data; cloud, GPU and compute providers; biological data and CRO partners; patients in indications where design speed matters
Brief length / format116 pages · PDF + executive summary deck · instant delivery

Understanding the Technology

Foundation models for biology learn the language of molecules. Protein language models trained on hundreds of millions of sequences, structure models such as AlphaFold 3 that predict how proteins fold and bind small molecules, nucleic acids and other proteins, diffusion and generative models that design new proteins and antibodies to a specification, and genome models such as Evo 2 that read and write DNA at the scale of whole genomes together let researchers predict, design and test in silico what once required years of wet-lab iteration. Generate Biomedicines engineered golukibart, an anti-TSLP antibody, with generative models; Profluent's models design recombinases that insert kilobase-scale DNA at chosen genome sites; Insilico's platform proposed the TNIK target and designed rentosertib; Isomorphic's engine claims to more than double AlphaFold 3's accuracy on protein–ligand benchmarks.

The evidence is now clinical. The ASCO 2026 analysis of 117 AI-enabled drugs found 60 had completed phase 1 and 8 phase 2, with phase 1 success of 80–90% against a historical 40–65% — the models produce drug-like, safe molecules faster — and phase 2 success of about 40%, level with history, because efficacy depends on biology the models do not yet fully capture. Rentosertib's phase 2a lung-function gain and phase 3 start, golukibart's 1,600-patient phase 3 and REC-4881's polyp reductions are the lead assets; the failures — Recursion's 2025 pipeline cull, Schrödinger's SGR-2921 after two deaths, Verge's ALS miss — show that AI design does not remove clinical risk. Pharma has priced the capability: Lilly's deals with Insilico and Profluent, Merck KGaA's with Valo, Genentech's with Earendil, GSK's with Noetik and Servier's with Iktos total several billion dollars of milestones, and Lilly's lab with NVIDIA commits up to USD 1 billion to compute and models.

The constraints are data, validation, compute and proof. Models are only as good as the experimental data behind them, which is why data licensing — Tempus with AstraZeneca and Pathos at USD 200 million — and lab-in-the-loop platforms such as Lila Sciences matter; generative designs must be made and tested; frontier models need GPU budgets that favour well-funded companies and partnerships with Alphabet, NVIDIA and cloud providers; and the first approval of an AI-discovered drug, with zasocitinib's FDA decision due in the first quarter of 2027 as the nearest AI-enabled candidate, will set the valuation of the whole field. The decade's question is whether phase 2 and 3 success rates move as phase 1 has.

Market Outlook

The protein and DNA language models market — biology foundation model and platform software and services revenue, partnership upfronts and recognised milestone value, model licensing, biological data licensing and dedicated compute for biology models, excluding the sales value of resulting drugs — is estimated at USD 1.5–2.0 billion in 2026, led by pharma partnership payments to AI-native companies, platform software and services, data licensing and compute. Meticulous Next™ expects it to reach USD 18–25 billion by 2036, an indicative CAGR of 28–30%. Growth is driven by the first approvals of AI-discovered and AI-engineered drugs from 2027 to 2030, by milestone and royalty flows from the current deal base, by adoption of generative design across antibodies, enzymes, gene editors and cell therapies, by genome models guiding genetic medicine design and by compute and data partnerships. The forecast is sensitive to phase 2 and 3 outcomes and to the first approval. The United States leads on frontier models, AI-native companies and capital; the United Kingdom hosts Isomorphic and Latent Labs; China contributes Insilico's Hong Kong listing, XtalPi and Qilu; Switzerland, Germany and France contribute pharma demand and models.

Scenarios

The base case assumes first AI-discovered approvals around 2028, phase 3 success for one of the lead assets by 2029 and continued pharma partnership flows. An accelerated case assumes both rentosertib and golukibart succeed, phase 2 success rates rise measurably and genome models deliver first genetic-medicine products, bringing the inflection to 2029 and lifting the 2036 value to the top of the range. A delayed case assumes lead phase 3 failures and flat phase 2 rates, cooling partnerships and pushing the inflection to 2032, with value confined to prediction tools and services.

Factors Behind Growth

Growth drivers

  • Clinical validation: 117 AI-enabled drugs in trials, 80–90% phase 1 success, an AI-engineered antibody in phase 3 and AI-discovered molecules in phase 3.
  • Pharma commitment: Lilly–Insilico, Lilly–Profluent, Merck KGaA–Valo, Genentech–Earendil, GSK–Noetik and Servier–Iktos deals worth billions in milestones.
  • Capital: Isomorphic's USD 2.1 billion, Earendil's USD 787 million, Generate's and Insilico's IPOs and about USD 4.2 billion across the top five private rounds.
  • Model progress: AlphaFold 3, Evo 2, ESM3 and successors extending prediction to interactions, genomes and generative design.

Enablers

  • Open and licensed model weights and benchmarks that let pharma adopt and compare models.
  • GPU compute partnerships with NVIDIA, Alphabet and cloud providers.
  • Biological data licensing and lab-in-the-loop platforms generating training data.
  • Automated synthesis and screening that test generative designs quickly.

Restraints and barriers

  • No approved AI-discovered drug yet; phase 2 success level with history.
  • Clinical failures and discontinuations that AI design does not prevent.
  • Data quality and access limits; compute cost concentrating frontier models in a few companies.
  • Regulatory and intellectual-property uncertainty for AI-designed molecules.

The Forces at Play

Five converging forces will determine how fast, and how far, foundation models reshape biological R&D: (1) clinical readouts and the first approvals; (2) pharma partnerships and in-house adoption; (3) model progress across structure, generative design and genomes; (4) data and lab-in-the-loop access; and (5) compute economics and capital. The brief assesses each force for direction, speed and confidence.

Adoption Outlook

How the shift is likely to unfold across three time horizons.

Near term2026–2029
Clinical proof and the first approval

Rentosertib and golukibart phase 3 trials read out; zasocitinib's FDA decision in early 2027 and the first AI-discovered approvals around 2028 set precedent. Pharma partnership and milestone flows from the Lilly, Merck KGaA, Genentech, GSK and Servier deals grow; Isomorphic, Earendil, Generate and Insilico deploy new capital into pipelines; frontier models from EvolutionaryScale, Arc Institute, Chai and Latent Labs reach new scales. Generative design spreads from antibodies to enzymes, recombinases and cell-therapy components. Lab-in-the-loop and data licensing expand; compute partnerships with NVIDIA and cloud providers formalise.

Mid term2029–2032
Standard in design; measured in success rates

Foundation models are standard tools in target, antibody, enzyme and gene-editor design across large pharma, with phase 2 and 3 success rates for AI-enabled assets tracked across dozens of readouts and reflected in pricing of partnerships. Several AI-discovered drugs are approved; royalty streams begin. Genome models guide gene therapy, editing and cell-therapy construct design. Consolidation among AI-native companies; pharma builds in-house model teams and data assets; regulators issue guidance on AI-designed molecules.

Long term2032–2036
Design-first R&D

Most new biologics, many small molecules and genetic medicine constructs are designed with foundation models and validated through automated experiments; R&D cycle times and attrition fall measurably in model-suited modalities. Value concentrates in AI-native companies with approved products and royalties, frontier model developers with licensing across pharma, large pharma with in-house platforms and data, compute and data providers, and CROs that integrated model-driven design and testing.

Latest Strategic Developments

Date

Development

Type

Significance

Nov–Dec 2025

Merck KGaA signs Valo Health for more than USD 3 billion in potential milestones plus royalties; Generate Biomedicines starts the phase 3 SOLAIRIA-1 trial of golukibart, an AI-engineered anti-TSLP antibody, in severe asthma, with SOLAIRIA-1 and -2 to enrol about 1,600 patients; a peer-reviewed analysis for ASCO 2026 counts 117 AI-enabled drugs from 63 companies in interventional trials with 80–90% phase 1 success; Insilico Medicine lists in Hong Kong raising about USD 292 million

Clinical / Finance

First AI-engineered antibody in phase 3; field reaches public markets

Jan 2026

Eli Lilly and NVIDIA announce an AI laboratory worth up to USD 1 billion over five years; Servier signs Iktos for more than EUR 1 billion and Insilico for up to USD 888 million; GSK pays Noetik USD 50 million upfront; Insilico's ISM8969 receives FDA IND clearance

Partnership

Compute and model partnerships scale across pharma

Feb–Mar 2026

Generate Biomedicines completes a USD 400 million Nasdaq IPO; Earendil Labs raises USD 787 million; Eli Lilly signs Insilico Medicine for up to USD 2.75 billion with USD 115 million upfront

Finance / Partnership

Capital and pharma deals concentrate in clinical-stage AI companies

Apr 2026

Profluent announces a strategic partnership with Eli Lilly to design site-specific recombinases for kilobase-scale DNA insertion using its generative protein models, with an upfront payment, committed R&D funding, up to USD 2.25 billion in milestones and tiered royalties, and Lilly taking selected recombinases through development

Partnership

Generative models move into genetic medicine design

May 2026

Isomorphic Labs raises a USD 2.1 billion Series B led by Thrive Capital with Alphabet, GV, MGX, Temasek, CapitalG and the UK Sovereign AI Fund, following its USD 600 million round in March 2025, to move oncology and immunology programmes toward the clinic alongside its Lilly (up to USD 1.7 billion) and Novartis (up to USD 1.2 billion) collaborations

Finance

Largest private round for an AI biology company

Jun–Sep 2026

Insilico doses the first patient with ISM8969 (June), gains FDA Fast Track for ISM6331 (July) and doses the first patient in the 320-patient GENESIS-IPF-3 phase 3 trial of rentosertib (September) after phase 2a showed a 98.4 mL FVC gain at 60 mg versus a 20.3 mL decline on placebo; Genentech signs Earendil for USD 55 million upfront and more than USD 1.5 billion potential; XtalPi signs a GPCR partnership worth more than USD 400 million

Clinical / Partnership

AI-discovered small molecule enters phase 3

Key Players & Competitive Landscape

The key players operating in protein and DNA language models include Isomorphic Labs Limited (Alphabet Inc., Google DeepMind), EvolutionaryScale, Inc., Profluent Bio, Inc., Chai Discovery, Inc., Latent Labs, the Arc Institute, Generate Biomedicines, Inc., Insilico Medicine, Inc., Recursion Pharmaceuticals, Inc. (Exscientia), Earendil Labs, Xaira Therapeutics, Inc., Lila Sciences, Inc., Pathos AI, Inc., Valo Health, Inc., Iambic Therapeutics, Inc., Absci Corporation, Schrödinger, Inc., Relay Therapeutics, Inc., XtalPi Holdings Limited, Iktos SA, Noetik, Inc., Cradle Bio, Inc., Nabla Bio, Inc., BigHat Biosciences, Inc., Eli Lilly and Company, Novartis AG, Merck KGaA, Genentech, Inc. (Roche), GSK plc, AstraZeneca PLC, Servier, Sanofi S.A., Pfizer Inc., NVIDIA Corporation, Microsoft Corporation, Amazon Web Services, Inc., Alphabet Inc. (Google Cloud), Tempus AI, Inc., Ginkgo Bioworks Holdings, Inc., Charles River Laboratories International, Inc., WuXi AppTec Co., Ltd., the Institute for Protein Design at the University of Washington and the Broad Institute. The brief profiles representative players in each archetype and assesses which are positioned to capture value from AI-driven biological R&D.

The competitive landscape is forming around six archetypes. Frontier biology model developers. AI-native drug discovery companies with clinical assets. Large pharma building in-house platforms and partnerships. Compute, cloud and model infrastructure providers. Biological data, lab-automation and CRO partners. Academic institutes and open-model communities. Competitive intensity is high in 2026 and is expected to sort companies by clinical outcomes and model differentiation through 2030.

Archetype

Representative players

Position in 2026

Outlook to 2036

Frontier biology model developers

Isomorphic Labs, EvolutionaryScale, Profluent, Chai Discovery, Latent Labs, Arc Institute, Cradle, Nabla Bio

Frontier structure, generative and genome models; pharma licensing

Set capability benchmarks; licensing and partnerships

AI-native drug discovery companies

Generate, Insilico, Recursion, Earendil, Xaira, Lila Sciences, Pathos, Valo, Iambic, Absci, Schrödinger, Relay, XtalPi, Iktos, Noetik, BigHat

Clinical assets in phase 1–3; pharma deals

First approvals; royalties; consolidation

Large pharma

Lilly, Novartis, Merck KGaA, Genentech, GSK, AstraZeneca, Servier, Sanofi, Pfizer

Partnerships worth billions in milestones; in-house platforms

Internalise capability; own data

Compute, cloud & infrastructure

NVIDIA, Microsoft, AWS, Google Cloud

GPU, cloud and model infrastructure; joint labs

Capture compute spend; model hosting

Data, lab automation & CROs

Tempus, Ginkgo, Charles River, WuXi AppTec, lab-in-the-loop platforms

Training data, synthesis and testing

Supply the experimental loop

Academic institutes & open models

Institute for Protein Design, Broad Institute, Arc Institute, open-weight communities

Foundational research and open models

Seed companies; keep benchmarks open

Where value migrates.

In 2026 value sits in pharma partnership payments to AI-native companies, platform software and services, frontier model licensing, data licensing and compute. By 2030 it moves to the first approved AI-discovered and AI-engineered drugs, milestone and royalty flows, foundation models as standard design tools in pharma and genome models in genetic medicine. By 2036 it settles in AI-native companies with approved products and royalties, frontier model developers licensed across pharma, large pharma with in-house platforms and data, compute and data providers and CROs that integrated model-driven design. Prediction-only tools and AI-native companies without clinical assets lose ground.

Who Will Win — and Why

The archetypes best positioned to capture value as the shift matures.

AI-native companies with clinical assets and pharma deals

companies such as Insilico and Generate whose models have produced phase 3 assets and whose partnerships convert to milestones and royalties.

Frontier model developers with pharma licensing

developers such as Isomorphic, Profluent, EvolutionaryScale and Chai whose models set benchmarks and are licensed or partnered across several pharma companies.

Compute, data and experimental-loop providers

GPU, cloud, data and lab-automation companies that supply every model and every design-test cycle

Regulatory Landscape

Jurisdiction

Milestone

Indicative timing

Effect on adoption

United States

FDA review of AI-enabled assets (zasocitinib decision Q1 2027); FDA guidance on AI in drug development; IND clearances and Fast Track for AI-discovered molecules; patent treatment of AI-designed molecules

2026–2030

First approvals and precedent

United Kingdom

MHRA clearances (Recursion REC-3565); UK Sovereign AI Fund investment in Isomorphic; AI and life sciences strategy

2026–2030

Frontier model hub

European Union

EMA reflection on AI in the medicinal product lifecycle; EU AI Act obligations; EU Biotech Act

2026–2032

Harmonised expectations for AI-designed products

China

NMPA IND clearances for AI-discovered molecules (ISM8969, August 2026); Hong Kong listings; data rules

2026–2032

Second major pipeline and capital base

Standards & IP

Benchmarks for structure and design models; data-sharing and model-weight licensing norms; patentability of AI-generated sequences

2026–2032

Shapes licensing value

Investment Signals

Capital is concentrated at the top: Isomorphic's USD 2.1 billion, Earendil's USD 787 million, Pathos's USD 365 million, Lila's USD 350 million and Generate's and Insilico's IPOs, about USD 4.2 billion across the five largest private rounds since January 2025, while pharma commits billions in milestones through Lilly, Merck KGaA, Genentech, GSK, Servier and XtalPi's partners and USD 200 million in data licensing from Tempus to AstraZeneca and Pathos. Clinical signals are measured — 117 assets, 80–90% phase 1 success, phase 3 starts in asthma and IPF — and so are failures. Patent and research activity is concentrated in protein–ligand prediction, antibody and enzyme generative design, genome language models, multimodal biological models and lab-in-the-loop learning. The brief tracks four indicators: phase 3 readouts for rentosertib and golukibart, the first AI-discovered approvals, phase 2 success rates across AI-enabled assets and new pharma partnership values.

The United States leads on frontier models, AI-native companies, capital and clinical assets; the United Kingdom hosts Isomorphic and Latent Labs with sovereign backing; China contributes Insilico's listing, XtalPi and Qilu partnerships and NMPA clearances; Switzerland, Germany and France contribute pharma demand through Novartis, Merck KGaA, Roche, Sanofi and Servier. Oncology, immunology, fibrosis and genetic medicine lead the applications.

Questions This Brief Answers

01What are protein and DNA foundation models, and how do they predict structure and design proteins, antibodies and genomes?
02What is the market size of protein and DNA language models in 2026, and what is the forecast to 2036?
03What does the clinical record — 117 AI-enabled drugs, phase 1 and 2 success rates, phase 3 starts — show?
04What factors are driving growth, and what data, validation, compute and approval barriers remain?
05Which key players are operating in biology foundation models, and which archetypes are positioned to win?
06What are the latest funding rounds, pharma partnerships, IPOs and clinical milestones?
07How will FDA, MHRA, EMA and NMPA treatment of AI-designed molecules shape adoption between 2026 and 2036?
08What should pharma, AI-native companies, model developers, infrastructure providers and investors do now?

Strategic Implications

  • Pharma R&D leaders: secure access to frontier models and AI-native pipelines now through partnerships and in-house teams; the capability is being priced in multi-billion-dollar deals and the first approvals will reprice it.
  • AI-native companies: convert model advantage into clinical assets and pharma milestones; the companies raising at scale are those with phase 2 and 3 programmes, not prediction tools alone.
  • Frontier model developers: license across several pharma companies and pair models with experimental loops; benchmark leadership without wet-lab validation does not hold value.
  • Compute, data and CRO providers: build biology-specific offers — GPU partnerships, data licensing, automated synthesis and testing — around pharma's model adoption.
  • Investors and regulators: treat the rentosertib and golukibart phase 3 readouts, the first approvals and phase 2 success rates as the leading indicators; phase 1 has been proven, phase 2 has not.
Analyst Perspective

"Biology's foundation models have done what every new discovery technology is asked to do first: put molecules into people. A hundred and seventeen of them, with phase 1 success nearly double the historical rate. What they have not yet done is beat biology in phase 2, where the rate is unchanged, or win an approval. Lilly has bet five billion dollars across two deals that they will. The next three years of phase 3 readouts decide whether that was early or wrong."

Lead Foresight Analyst
Biotechnology, AI in R&D & Computational Biology · Meticulous Next™

Table of Contents

Access & Licensing

A focused foresight brief, priced to circulate. Every option is delivered instantly and backed by analyst support.

Single Brief
$850
One named user
Full PDF brief
Executive summary deck
One named user
Free outlook update
★ Most Popular
Team License
$1,350
Up to 10 users
Access for 2–10 users
Internal sharing rights
30-minute analyst briefing
Enterprise
$2,050
Organization-wide
Organization-wide access
Unlimited users
Priority analyst access
Custom extracts
Included with
every brief
Analyst-reviewed foresightMulti-signal validationInstant deliveryFree outlook updates

Frequently Asked Questions

Search Market Intelligence

Search across reports, blogs, press releases, and industries