In-Silico Drug Discovery Market

The global In-Silico Drug Discovery Market generated USD 2.556 billion in 2025 and is estimated at USD 2.878 billion in 2026. Revenue is forecast at USD 7.438 billion by 2034, equivalent to a 12.6% CAGR for 2026–2034.

In-silico drug discovery encompasses paid molecular modelling, virtual screening, predictive ADMET, target prioritisation, computational chemistry and AI-assisted design delivered through licences, cloud subscriptions or contracted discovery services. Charles River sells computer-aided drug-design work from hit identification through lead optimisation, while Schr - dinger licenses physics-based simulation and enterprise molecular design software. Revenue from downstream approved drug sales, wet-lab-only testing and general-purpose computing hardware is excluded from the computational discovery boundary.

The FDA issued draft guidance on 6 January 2025 describing risk-based credibility assessment for AI models supporting drug and biologic regulatory decisions. Its scope includes the intended context of use and evidence that a model performs reliably for that context. The June 2026 M15 guidance addresses general principles of model-informed drug development. These frameworks increase demand for documented model validation, reproducible calculations and qualified biosimulation workflows rather than replacing confirmatory experiments.

Commercial demand extends from molecular docking to phenotypic data integration. Recursion reported five Sanofi discovery-program milestones by February 2026, and its Genentech collaboration advanced a neuroscience target into joint early discovery in August. Certara released Simcyp version 25 in March 2026 after receiving an EMA qualification opinion for specified PBPK uses in August 2025. These examples connect contracted computational work, recurring platform licences and regulatory-grade modelling with identifiable pharmaceutical customers.

Asia-Pacific suppliers are expanding alongside U.S. and European incumbents. Insilico Medicine signed a July 2026 Takeda collaboration with about USD 60 million in project-initiation and near-term consideration, while Shanghai authorities described its LabClaw laboratory orchestration system in May. Japan’s RIKEN selected a new AI-for-science computing system in July 2025, and India’s CSIR held an AI drug-discovery workshop in April 2026. Growth depends on reproducible wet-lab validation, access to suitable datasets and buyers’ willingness to commit multiyear discovery budgets.

Source: FDA AI credibility guidance; Schr - dinger BMS agreement; Insilico–Takeda collaboration.

Revenue covers commercial software, subscriptions and specialist services used to identify and optimise drug candidates computationally before clinical approval; downstream drug sales and generic cloud infrastructure are outside scope.

Drivers Impact Analysis*

Restraints Impact Analysis*

*CAGR contribution shows the percentage points each factor adds to or removes from the 12.6% CAGR for 2026–2034; driver contributions less restraint contributions equal the market CAGR. Period columns rate the strength of each factor in that window.

AI and physics-based platforms broaden discovery throughput

Schr - dinger introduced Bunsen to Bristol Myers Squibb in August 2026, combining agentic planning with physics-based simulation and retrosynthesis. NVIDIA expanded BioNeMo in January 2026 with molecular modelling and biological foundation-model components. These tools allow scientists to rank hypotheses before ordering chemical synthesis, so project budgets can support larger virtual compound spaces without matching growth in wet-lab screening. Buyers still require prospective experimental confirmation, because a high predicted affinity does not establish clinical efficacy. Integration into medicinal-chemistry workflows converts platform use into recurring licence and collaboration revenue.

Pharmaceutical collaborations create contracted demand

Insilico Medicine and Takeda announced a July 2026 collaboration with about USD 60 million in initiation fees, near-term payments and milestones, alongside contingent longer-term consideration. In October 2026, insitro and AstraZeneca disclosed USD 40 million upfront for computational discovery in MASH and cardiomyopathies. Such agreements pay for target selection, model development and candidate nomination before a medicine reaches the market. Collaboration economics vary by achieved research milestones; future development and commercial milestone maxima are conditional and cannot be recognised as current discovery sales.

Regulatory credibility requirements stimulate validation spending

The FDA’s January 2025 draft AI guidance sets out a risk-based approach to model credibility for a defined context of use. EMA qualified Certara’s Simcyp simulator in August 2025 for specified PBPK drug-interaction applications, and Certara launched version 25 in March 2026. Sponsors therefore need traceable input data, version-controlled models, documented performance and clear limits on intended uses. This raises demand for specialist validation and simulation services while limiting adoption of black-box models that cannot explain predictions to reviewers or medicinal chemists.

Data fragmentation and experimental failure restrain contracts

The NIH Biomedical Data Translator initiative addresses fragmented biomedical data across research, clinical and adverse-event systems. Discovery vendors encounter similar incompatibilities when integrating assays, omics, molecular structures and laboratory metadata. A model trained on inconsistent measurements can prioritise compounds that fail confirmatory tests, adding cost to hit-to-lead work. Clients may restrict cross-border transfer of proprietary structures and patient-derived data, reducing the usefulness of shared cloud models. Specialist data curation, access controls and assay replication increase project costs before a computational lead becomes a validated candidate.

Skilled staff and compute expense influence purchasing decisions

RIKEN selected an AI-for-science computing system in July 2025, illustrating the scale of infrastructure supporting modern simulation and foundation models. Smaller biotech companies cannot always justify dedicated GPU clusters or computational chemistry teams, which favours subscription software and external project services. UK Research and Innovation offered GBP 12 million for doctoral training in AI and bioscience during 2025, reflecting the need for specialised talent. Licence cost, integration effort and reproducibility requirements may slow procurement, particularly where drug-discovery teams operate on fixed grant budgets.

Sources: FDA; EMA; NIH NCATS; Schr - dinger; NVIDIA; Insilico Medicine; insitro; RIKEN; UKRI.

Agentic molecular design moves into pharmaceutical workflows

Bristol Myers Squibb agreed in August 2026 to deploy Schr - dinger’s Bunsen agentic co-scientist, with integration into molecular simulation and synthesis planning. The system is designed to plan computational experiments and interpret their results within validated scientific workflows. Recursion’s platform connects cell imaging, multimodal data and computational design, while its Genentech collaboration advanced a neuroscience target in August. These deployments support larger enterprise software contracts, but medicinal chemists must still evaluate synthetic feasibility, selectivity and assay confirmation before compounds progress into development.

Qualified biosimulation is expanding beyond early discovery

Certara’s Simcyp platform received an EMA qualification opinion in August 2025 for specified drug-interaction modelling scenarios. Version 25, announced in March 2026, added capabilities for drug development and regulatory submissions. The FDA issued M15 general principles for model-informed drug development in June 2026, strengthening expectations for documented assumptions and uncertainty. Computational vendors can sell integrated services from lead optimisation to clinical pharmacology, provided each application is validated for its intended decision and does not substitute unsupported simulation for necessary trials.

Laboratory automation links predictions with measured outcomes

Shanghai authorities described Insilico Medicine’s LabClaw system in May 2026 as coordinating target discovery, screening, experiments and analysis through multiple AI agents. Recursion operates automated cell-based experiments to generate phenotypic data used by its discovery models. Evotec’s Enricture offering combines affinity-selection mass spectrometry with machine-learning hit enrichment. By connecting experimental feedback to predictions, suppliers can improve the next screening cycle and reduce the number of compounds purchased for confirmation, creating differentiated service contracts beyond stand-alone docking software.

Cross-border research investment develops regional supplier capacity

China’s Pudong authorities described an AI new-drug research alliance with more than 40 members in June 2025. India’s CSIR trained scientists in AI-assisted drug discovery in April 2026, while South Africa’s University of Cape Town co-hosted a September 2025 AI-for-drug-discovery workshop. RIKEN is combining AI computing with existing Japanese scientific infrastructure. These initiatives widen the future customer and supplier base for software licences, collaborative discovery and contract screening, although funding maturity and access to specialised datasets differ substantially across countries.

Hybrid AI and molecular physics address prediction limitations

Selvita’s virtual-screening platform uses a target-aware model to prioritise compounds from very large libraries, while BIOVIA Discovery Studio supports molecular mechanics and simulation. Schr - dinger’s 2026 platform roadmap emphasises combined physics and AI, including predictive toxicology and synthesis planning. Computational chemists increasingly compare learned activity predictions with binding energetics and experimentally observed structure–activity relationships. Hybrid validation raises software differentiation but also requires curated chemical data and trained scientists, making service-led deployment attractive to sponsors with limited internal capabilities.

In-silico drug discovery spending is divided by contract economics rather than by the software architecture alone. Consultancy, hosted subscriptions, installed licences and earned platform-collaboration consideration are separate revenue streams, with drug-product royalties excluded from discovery services.

Enterprise integration and recurring access are expected to gain weight through 2034 as computational tools become embedded in routine medicinal chemistry, while project-based consultancy remains important for difficult targets and sponsors lacking specialised staff.

Contracts are allocated to the first substantive research decision they support: target identification, lead discovery and optimisation, or preclinical property prediction. This avoids counting the same molecular-design project across successive workflow stages.

Target discovery benefits from larger multimodal datasets, while lead optimisation retains the highest base-year value because repeated structure–activity experiments require sustained computational support. Certara’s March 2026 Simcyp release supports preclinical modelling projects that require documented assumptions for regulators.

Structure-based, ligand-based and AI-driven methods can coexist in a single research programme, but commercial revenue is assigned to the dominant contracted method. Assigning contracts to one dominant method prevents duplicate technology revenue across combined molecular modelling projects.

Physics-based methods provide interpretable molecular mechanisms, while AI methods expand searchable chemical and biological spaces. Integrated software and service providers are positioned to capture projects requiring both approaches.

Purchasers are grouped by the organisation paying for the computational discovery deliverable. Biopharmaceutical sponsors, CROs and academic or government research bodies form non-overlapping demand categories.

Biopharmaceutical sponsors are expected to remain the largest paying customer group, while CROs extend access to smaller drug developers. Public research procurement grows where grants fund computational infrastructure and interdisciplinary training.

North America leads commercial spending on computational discovery, while Europe contributes qualified modelling tools and Asia-Pacific adds research capacity. Latin America and Middle East & Africa remain smaller procurement markets centred on public institutes and specialist collaborations.

In-silico drug discovery competition spans licensed molecular software, contracted chemistry services and AI-native platform partnerships. Schr - dinger sells physics-based molecular design, Certara supplies model-informed drug-development tools, and Charles River and Evotec integrate computational work with experimental services. Recursion, Insilico Medicine and insitro use proprietary datasets and models in research collaborations, while NVIDIA provides foundational computing software used in biological AI workflows.

NVIDIA supplies BioNeMo biological model infrastructure, but general-purpose GPU hardware revenue is excluded from the computational drug-discovery boundary. Suppliers differentiate through validated algorithms, proprietary biological data, domain expertise and direct integration with experiments. A platform that can prioritise thousands of compounds still needs assay confirmation and credible uncertainty estimates, so sponsors assess end-to-end scientific performance rather than model speed alone.

Partnership structures vary from fixed annual subscriptions to fee-for-service chemistry and research collaborations with conditional milestones. The October 2026 insitro–AstraZeneca agreement includes USD 40 million upfront, whereas the July 2026 Insilico–Takeda arrangement includes around USD 60 million in near-term consideration. These payments cannot be treated as equivalent to the maximum contingent deal values. Buyers also compare intellectual-property terms, cross-border data controls and portability between modelling environments.

European providers benefit from established medicinal chemistry and EMA-facing modelling experience, while U.S. platform companies serve large pharmaceutical R&D teams. China-based AI developers increasingly sign international discovery agreements. Competition is likely to intensify as cloud infrastructure reduces deployment costs, but scientific validation, proprietary datasets and laboratory integration remain substantial barriers to interchangeable commodity software.

• Schr - dinger, Inc.
• Certara, Inc.
• Charles River Laboratories International, Inc.
• Evotec SE
• Dassault Syst - mes SE (BIOVIA)
• Selvita S.A.
• Recursion Pharmaceuticals, Inc.
• Insilico Medicine
• insitro
• NVIDIA Corporation

Revenue from in-silico discovery platforms and contracted computational services is forecast to increase from USD 2.878 billion in 2026 to USD 7.438 billion in 2034, a 12.6% CAGR. Recurring software adoption, research collaboration payments and qualified simulation tools support the projected expansion.

Pharmaceutical sponsors are likely to expand computational discovery into routine hit-to-lead decisions rather than reserve modelling for difficult targets. Schr - dinger’s 2026 BMS deployment demonstrates the move from specialist use toward wider research-team adoption. Recursion’s Genentech programme shows how AI-based biological mapping can generate targets for joint chemistry work. Each new enterprise integration can support annual licence renewals, scientific services and collaboration payments, while revenue recognition remains tied to delivered obligations.

Europe is expected to sustain demand for regulatory-grade modelling because EMA qualification and AI guidance create defined validation requirements. Certara’s Simcyp example demonstrates a route from software performance evidence to accepted regulatory contexts of use. In Asia-Pacific, Chinese platform developers are expanding global partnerships, and public investments in Japan and India increase the available research talent and computing infrastructure. Latin American and African growth will remain more dependent on academic budgets and externally funded research programmes.

The largest risks are poor training-data quality, model drift across chemical series, high compute costs and weak translation from virtual predictions into laboratory results. NIH NCATS continues to address interoperability across biomedical datasets, highlighting the challenge of combining incompatible sources. Drug sponsors also need to protect unpublished molecular structures and manage cross-border data access. Vendors that provide transparent uncertainty estimates, reproducible workflows and experimental confirmation can defend premium pricing against lower-cost general AI tools.

The 2034 market mix should contain a greater share of hosted software and integrated AI services, while consultancy retains a substantial role in complex programmes. The value of contingent downstream drug royalties is excluded from computational discovery revenue, even where a platform provider also owns clinical assets. Stronger validation standards may lengthen procurement cycles initially but can increase renewal rates for tools that repeatedly deliver useful, experimentally confirmed candidates.

Coverage includes computational drug-discovery software and contracted services from target selection through preclinical modelling, excluding approved-drug sales and unrelated infrastructure hardware.