AI-driven small-molecule design — now part of Recursion
Exscientia is an AI drug-design platform — now part of Recursion Pharmaceuticals — built around its Centaur generative-AI design technology.
AI Panel Score
6 AI reviews
Reviewed
AI Editor ApprovedApproved and published by our AI Editor-in-Chief after full panel analysis.Exscientia is an AI-first drug-design platform, now part of Recursion Pharmaceuticals following their 2024 merger, built around its Centaur generative-AI technology. It designs and optimizes small-molecule drug candidates, pairing AI design with closed-loop laboratory automation and patient-tissue data to select candidates and the patient subsets most likely to benefit. Engagement is quote-based through business development rather than self-serve software. Key capabilities include Centaur generative molecular design, a design-make-test-analyse closed loop, active-learning compound optimization, multi-parameter optimization, automated synthesis and robotics, and precision-medicine patient models that inform which patient subsets to target. TopReviewed's six-seat AI review panel scored it 7.4/10, praising that six Exscientia-designed molecules have entered clinical trials while noting that after the November 2024 merger it no longer exists as a standalone engagement option and every deal is a bespoke negotiation. It best fits biopharma R&D leaders seeking an AI-driven small-molecule discovery partner backed by public-company disclosure.
Exscientia is a UK-founded AI-first drug-discovery company, now part of Recursion Pharmaceuticals. Founded in 2012, it developed proprietary technology — branded Centaur — to design and optimize small-molecule drug candidates, integrating generative AI and machine-learning models with experimental data to explore chemical space far more efficiently than traditional methods. Its end-to-end approach spanned target identification, hit discovery, lead optimization, and candidate selection, repeatedly producing development candidates in roughly 12–15 months versus an industry norm of around four years.
The company operated a hybrid model, maintaining its own pipeline of therapeutic programs while partnering with major pharmaceutical and biotech companies, including Sanofi, Bristol Myers Squibb, and Bayer. These collaborations typically applied Exscientia's AI platform to a partner's therapeutic targets in exchange for milestone payments and royalties.
A defining aspect of Exscientia's approach was closed-loop automation, in which AI-designed molecules are synthesized and tested by robotic systems and the resulting data feeds back into the models. It also pioneered a precision-medicine method that uses data from primary human tumor samples and multi-omics sequencing to predict a candidate's efficacy and identify the specific patient subset most likely to respond.
In 2024 Exscientia merged with Recursion Pharmaceuticals in an all-stock deal valuing Exscientia at roughly $650 million, completing in November 2024. The combination pairs Exscientia's Centaur design platform and precision chemistry with Recursion's large-scale biological datasets and automated wet labs, enabling fully automated cycles from biological screen to optimized lead. Exscientia's technology now operates within Recursion's combined platform.
As a drug-discovery company rather than a software-as-a-service product, Exscientia does not publicly list subscription pricing; access to its platform and capabilities was negotiated through enterprise partnerships and collaborations.
Synthesizes and assays AI-designed molecules with robotic laboratory automation to accelerate each design cycle.
Uses generative AI to design and optimize novel small-molecule candidates against a target product profile, exploring chemical space beyond human-enumerated libraries.
Uses data from primary human tumor samples and multi-omics sequencing to predict efficacy and identify the patient subgroups most likely to benefit.
Prioritizes which compounds to make and test next using active learning, reducing the number of molecules needed to reach a development candidate.
Applies the platform to partners' targets under collaborations with companies such as Sanofi, Bristol Myers Squibb, and Bayer, structured around milestones and royalties.
Balances potency, selectivity, ADME, and safety properties simultaneously when designing candidates rather than optimizing one property at a time.
Following the 2024 merger, combines Centaur design with Recursion's biological datasets and automated wet labs for end-to-end discovery within Recursion's platform.
Couples AI design with automated synthesis and testing so experimental results feed back into the models in a continuously improving cycle.
Advances a proprietary pipeline of small-molecule programs, several of which have progressed into clinical trials.
Applies machine learning to biological and clinical data to identify and validate new drug targets for discovery programs.
Exscientia operates as a drug-discovery partner rather than a self-serve SaaS — engagements are structured as multi-year discovery collaborations or platform licensing deals. Contact the business-development team for terms.
Exscientia is now Recursion — evaluate the combined balance sheet, not the legacy AI drug-design brand.
“Exscientia closed its all-stock merger with Recursion in 2024, valued at roughly $193 million, and now operates inside Recursion Pharmaceuticals. The combined company is publicly traded with cash runway guided into 2027.”
Three questions before this review even matters. Did the 2024 Recursion merger close? Yes. Is Exscientia still an independent vendor? No. Is the combined company worth a 36-month bet? That's the actual conversation.
Recursion OS is the asset you're actually licensing — closed-loop AI design feeding a wet lab that captures millions of cell experiments weekly, all crunched on BioHive-2, the NVIDIA-built supercomputer. Sanofi signed for $100 million upfront and up to $5.2 billion in milestones in 2022. That's the validation Schrödinger and Insilico Medicine still chase.
But this isn't software you procure. There's no pricing page, no self-serve tier — engagements are multi-year discovery partnerships negotiated by BD. Pilot a single target program before anyone talks about platform licensing, and watch the cash runway, which management has guided into 2027.
Most pipeline-advanced AI drug discovery platform; ahead of Schrödinger and Insilico on clinical progression.
Sanofi, BMS, Bayer, and NVIDIA collaborations make the choice defensible to any pharma board.
Drug discovery cycles measured in years; no quarterly payback even with closed-loop automation.
Genuine fit for pharma R&D buyers seeking AI-led discovery partners, less so for software procurement.
Combined Recursion-Exscientia entity reported ~$850M cash post-merger with guided runway into 2027.
Biopharma R&D leaders who want a discovery partner backed by public-company disclosure.
Software buyers expecting a SaaS tier or a self-serve trial.
Two AI drug-discovery pipelines under one Recursion OS — the moat is data scale, not molecule design.
“Recursion closed the $688M all-stock Exscientia acquisition in late 2024, inheriting REC-3565 MALT1 and REC-4539 LSD1 alongside its own RBM39 program REC-1245. The catch is that closed-loop AI discovery still has to clear Phase 2 readouts — Insilico Medicine and Isomorphic Labs are racing the same clock.”
Recursion finished absorbing Exscientia in late 2024, an all-stock deal valued near $688M, leaving roughly $850M cash and a projected $100M in annual synergies. Exscientia's REC-3565 MALT1 and REC-4539 LSD1 began Q1 2025 dosing, joining Recursion's REC-1245 RBM39 degrader.
Recursion OS is the integrating layer — phenomics, transcriptomics, proteomics and de-identified patient data on the same substrate, with BioHive-2 (built with NVIDIA in 2024) running the models. 50 petabytes of fit-for-purpose data is the moat against Insilico Medicine's Pharma.AI and Isomorphic Labs, neither of which owns a closed-loop wet lab at this image-throughput scale.
However, the three-year call is biology, not software. Recursion deprioritized REC-4881 in 2024 and surviving oncology programs are still early Phase 1; no AI-discovered molecule has cleared Phase 3 anywhere in the category. The bet is whether closed-loop phenomics shortens cycles before pipeline attrition forces another consolidation.
Post-merger, the combined company is the most-capitalized public AI drug discovery platform with $850M cash and a four-program oncology pipeline.
Enterprise discovery-partnership model with Sanofi, Bristol Myers Squibb and Bayer matches how biopharma R&D actually procures platform tech.
Partnership-only access fits pharma enterprise stacks but rules out self-serve integration for smaller biotech teams.
Recursion OS lock-in is real and pipeline still has to deliver Phase 2 wins before the architecture pays off.
BioHive-2 with NVIDIA plus 50 petabytes of phenomics, transcriptomics and ADME data is best-in-class craft depth for AI drug discovery.
Biopharma R&D leaders who need closed-loop AI discovery at petabyte data scale.
Solo researchers who want a self-serve molecular design tool.
Recursion absorbed Exscientia in a $193M all-stock deal November 2024 — no standalone SKU survives.
“No published pricing — Exscientia engagements now invoice through Recursion Pharmaceuticals as multi-year discovery partnerships. Sanofi's 2022 deal anchors the comp at $100M upfront against $5.2B in clinical milestones.”
Recursion absorbed Exscientia in a $193M all-stock deal closed November 2024. There is no Exscientia SKU anymore — the platform invoices through Recursion Pharmaceuticals. Public NASDAQ counterparty under ticker RXRX, audited financials, real cap table. Procurement at least gets a real legal entity.
No published pricing. Engagements run as multi-year discovery partnerships, milestone-plus-royalty paper. Sanofi's 2022 deal anchors the comp — $100M upfront, up to $5.2B across 15 small-molecule programs. Most biotechs will not see that scale. Mid-size buyers negotiate target-by-target.
Recursion OS and the NVIDIA-built BioHive-2 supercomputer with 504 H100 GPUs are real assets. Compare to Isomorphic Labs — Google-backed, partnership-only, similar gating on access. But the catch is ROI on a discovery deal does not land for five to seven years. Milestones tie to clinical phases, not quarterly budget reviews.
NASDAQ-listed counterparty (RXRX) with audited financials reduces legal-entity vendor risk.
Multi-year discovery partnerships with milestone-plus-royalty paper, negotiable per deal but not standardized.
No published pricing anywhere — every engagement requires a business-development sales call.
Discovery ROI tied to clinical phases over five to seven years, not quarterly metrics.
Sanofi anchor is $100M upfront against $5.2B milestones across 15 programs — most buyers cannot model their own version.
Biopharma R&D leaders who need AI-driven small-molecule discovery.
SMB buyers who need self-serve SaaS pricing.
Centaur Chemist designs molecules a med chem team can defend, but the platform now lives inside Recursion.
“Exscientia's Centaur Chemist platform compresses lead optimization by scoring potency, selectivity, and ADME in one design pass instead of three separate assay rounds. After the November 2024 Recursion merger, the tool is no longer accessible as a standalone partnership — it ships as part of Recursion OS.”
Centaur Chemist scores potency, selectivity, ADME, and synthetic tractability in one multi-parameter pass — work a med chem team normally splits across three weekly assay reviews. DSP-1181 with Sumitomo Dainippon reached candidate after about 350 compounds, versus the roughly 2,500 a conventional program burns through. That's the number every deck cites.
The closed loop matters in practice. Designs go to the automated wet lab, synthesis and assays run, results feed the next iteration — no chemist hand-copying CSVs into Schrödinger LiveDesign. Atomwise stayed virtual-screen-only; Exscientia owned the synthesis loop, which is where most AI design platforms fall over.
But after the November 2024 Recursion merger, Exscientia stopped existing as a standalone partner. Centaur Chemist now ships inside Recursion OS alongside BioHive-2. Chemists who just wanted the small-molecule design layer are buying into a phenomics-first platform.
Closed-loop design-make-test is real, but post-merger practitioner access runs through Recursion business development.
Public technical material is limited to SEC filings, partner press releases, and a handful of conference papers.
No self-serve entry — every engagement is a multi-year discovery collaboration negotiated through sales.
Multi-parameter optimization across potency, selectivity, ADME, and tractability is genuinely deep for a chemistry workflow.
Owns the synthesis loop end-to-end, eliminating CSV handoffs to external design tools like Schrödinger LiveDesign.
Medicinal chemists who want closed-loop AI design tied to automated synthesis.
Teams who need standalone small-molecule tooling without committing to Recursion OS.
Exscientia is a tab on Recursion's website now, and the merger price was $193 million in stock.
“Recursion absorbed Exscientia in 2024 for about $193 million all-stock, and the combined company runs as Recursion Pharmaceuticals. The platform is enterprise-only — partnerships and licensing deals, no self-serve, no pricing page.”
The link from this page lands you on recursion.com. The Exscientia name lives on press releases and SEC filings, but the product is the Recursion OS now. One company's stack, another's brand on the side.
The technical bench is real. BioHive-2, the NVIDIA H100 supercomputer they stood up in 2024, hit #35 on the TOP500 list and runs four times faster than BioHive-1. A 50-petabyte phenomics dataset is the moat. Schrödinger sells chemistry simulation software, not 2 million wet-lab experiments a week feeding the models.
But there's no product to buy. No pricing tier, no trial, no docs — just a contact button and a multi-year discovery collaboration on the other side. Mobile is moot when there's no app at all; the whole platform is web-only and behind enterprise sales. Day three for a researcher means MSA negotiation, not a sandbox.
Website and brand collateral are clean post-merger, though Exscientia-branded pages now redirect to Recursion.
No public docs or changelog; the platform is taught through partner engagements, not self-discovery.
Neutral score for enterprise scientific infrastructure where mobile is not a relevant surface.
There is no onboarding — every engagement starts with a contact-sales button and a partnership conversation.
BioHive-2 with 504 NVIDIA H100 GPUs and multi-year deals with Sanofi, Bayer, and Roche signal a real operating bench.
Pharma R&D teams who can sign multi-year discovery deals.
Solo researchers who need self-serve access today.
Recursion absorbed Exscientia in an all-stock deal valued at $193M in August 2024.
“Exscientia closed into Recursion Pharmaceuticals last year, and the platform now ships under the Recursion OS banner with BioHive-2 compute behind it. The AI drug-design pitch is real, but the standalone Exscientia thesis is gone.”
The brand on the door says Exscientia. The website redirects to Recursion. That's the first thing to know — the all-stock merger closed in 2024 at roughly $193M, and Exscientia shareholders ended up with 26% of the combined entity. Not the merger of equals it was sold as.
What survived is the platform. Recursion OS now sits on 50 petabytes of phenomics and ADME data, BioHive-2 runs the compute via NVIDIA, and the closed-loop wet lab keeps shipping experimental cycles weekly. Real infrastructure, not a slide deck.
But the clinical scoreboard is mixed. DSP-1181, the AI-designed OCD candidate Exscientia trumpeted in 2020, was discontinued by Sumitomo by 2022. Isomorphic Labs has Alphabet money and DeepMind talent. Insilico Medicine has more Phase 2 readouts. The platform is durable; the differentiation isn't obvious anymore.
Real wet-lab plus compute moat, but Isomorphic Labs and Insilico Medicine crowd the same pitch.
Discovery partnership engagements are not extractable software; you exit by ending the contract.
Combined entity held roughly $850M cash at close with runway into 2027 and public-market discipline.
Product page still markets Exscientia, but the URL and brand have already shifted to Recursion post-merger.
DSP-1181 discontinuation and the absorbed-by-acquirer pattern mirror earlier struggling AI-bio cohort.
Pharma R&D teams who want a multi-year discovery collaboration with embedded AI design and wet-lab cycles.
Biotech buyers who need self-serve software or short engagement contracts.
Common questions answered by our AI research team
Exscientia applies AI-driven molecular design to accelerate drug discovery, combining machine learning with robotic laboratory automation to design and optimize drug candidates faster and at lower cost than traditional methods.
Yes, Exscientia combines AI-driven molecular design with robotic laboratory automation, integrating computational and physical lab processes to streamline preclinical drug development.
Exscientia covers drug development from target identification through preclinical candidate selection, spanning the early discovery phases before clinical trials begin.
Yes, Exscientia partners with major pharmaceutical companies alongside developing its own internal drug pipeline.
Exscientia's core mission is to use AI and automation to radically reduce the time and cost of developing new medicines, addressing the high failure rates of traditional drug discovery.
Exscientia was an Oxford-based AI drug discovery company using machine learning to design novel small molecules, acquired by Recursion in October 2024.
