Recursion Pharmaceuticals
Last reviewed
Jun 7, 2026
Sources
20 citations
Review status
Source-backed
Revision
v1 ยท 2,132 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 7, 2026
Sources
20 citations
Review status
Source-backed
Revision
v1 ยท 2,132 words
Add missing citations, update stale details, or suggest a clearer explanation.
Recursion Pharmaceuticals (NASDAQ: RXRX) is a clinical-stage biotechnology company based in Salt Lake City, Utah, that describes itself as a "TechBio" company built to industrialize drug discovery by combining artificial intelligence with large-scale automated wet-lab experiments. Founded in 2013, Recursion generates massive proprietary biological datasets, principally through automated microscopy of human cells, and trains machine learning models on that data to identify drug targets and candidate molecules. The company has assembled one of the largest privately held biological and chemical datasets, secured collaborations worth billions of dollars in potential milestones with Roche, Genentech, Bayer and Sanofi, taken a $50 million strategic investment from Nvidia, and in November 2024 completed an all-stock merger with the British AI drug discovery firm Exscientia. Despite this high profile, Recursion has not yet brought a drug to market, has recorded large recurring net losses, and in 2025 discontinued several clinical programs and cut roughly 20 percent of its staff. Its founder, Chris Gibson, has served as chief executive throughout.
Recursion was founded in 2013 in Salt Lake City. The core idea originated in the University of Utah laboratory of physician-scientist Dean Li, where Chris Gibson was completing a Ph.D. in bioengineering. Gibson, who took a leave of absence from medical school to build the company, became chief executive. The founding team is generally described as comprising Gibson, Li, and computer scientist Blake Borgeson. The early thesis was to repurpose existing drugs for rare genetic diseases by screening them against cellular models of disease and reading out the results with automated imaging, an approach the company later generalized into a broader discovery engine.
The company stayed private for several years, raising venture capital before going public. Recursion priced its initial public offering at $18.00 per share, and its shares began trading on the Nasdaq Global Select Market under the ticker "RXRX" on April 16, 2021. With the underwriters' option exercised, the IPO closed on April 20, 2021 on the sale of roughly 27.9 million shares for gross proceeds of about $502 million; the base deal raised approximately $436 million.
Recursion's central asset is what it calls the Recursion Operating System (OS), a closed-loop platform that pairs industrial-scale biology with machine learning and AI. The biology side is built on phenomics: the company runs large automated laboratories that culture human cells, perturb them with genetic edits or compounds, stain them, and photograph them at scale using high-content microscopy. Each image captures hundreds of subtle morphological features, and Recursion uses deep learning to convert these cellular pictures into numerical "embeddings" that can be compared to map relationships between genes, diseases and drugs. The company has stated that its proprietary dataset spans phenomics, transcriptomics, proteomics, ADME (absorption, distribution, metabolism and excretion) and de-identified patient data, and by 2026 it described holding more than 50 petabytes of multimodal data. Recursion has also developed in-house foundation models, including Phenom-1 for extracting biological features from cell images and a family of transcriptomics models it has referred to as TxPert and TxFM.
Training models on datasets of this size requires substantial computing power, which Recursion brought in-house through a series of supercomputers branded "BioHive." BioHive-1, its first on-premises system, was built on Nvidia's DGX SuperPOD architecture. In November 2023 the company announced a roughly fourfold expansion, adding more than 500 Nvidia H100 GPUs to the more than 300 Nvidia A100 GPUs already installed. Its successor, BioHive-2, was completed in 2024. According to Recursion and Nvidia, BioHive-2 is built from 63 Nvidia DGX H100 systems containing 504 H100 Tensor Core GPUs connected by Nvidia Quantum-2 InfiniBand networking, and at launch the company described it as the most powerful supercomputer owned and operated by any pharmaceutical company.
On August 8, 2024, Recursion announced a definitive agreement to combine with Exscientia, an Oxford, England-founded AI drug discovery company whose strength was generative chemistry and molecular design, complementing Recursion's biology-first platform. The deal was structured as an all-stock transaction in which each Exscientia share was exchanged for 0.7729 shares of Recursion Class A common stock. Shareholders of both companies approved the combination on November 13, 2024, and the merger closed on November 20, 2024, with Exscientia becoming a wholly owned subsidiary of Recursion. Exscientia's American depositary shares (Nasdaq: EXAI) were delisted.
The combined company kept the Recursion name and Salt Lake City headquarters while retaining a significant presence in the United Kingdom. Gibson remained chief executive, and Exscientia's David Hallett joined as chief scientific officer. At closing, Recursion described a combined pipeline of more than 10 clinical and preclinical programs, around 10 advanced discovery programs, and more than 10 partnered programs.
Recursion is balanced in its own framing as having a high-profile platform but, as of mid-2026, no approved drug. In the months after the Exscientia merger the company moved to streamline a pipeline that had grown through the combination. On May 5, 2025, Recursion announced it was halting or pausing several programs to concentrate resources on oncology and rare disease.
Among the discontinued programs, REC-994, an oral superoxide scavenger for cerebral cavernous malformation, was stopped after long-term data from its SYCAMORE trial showed outcomes the company characterized as not clearly distinguishable from the natural course of the disease. REC-2282, a brain-penetrant HDAC inhibitor for neurofibromatosis type 2-related meningiomas, was dropped despite meeting a futility threshold because it showed limited tumor shrinkage and clinical activity. REC-3964, a non-antibiotic toxin inhibitor for Clostridioides difficile infection, was halted for strategic reasons that the company tied to competing advances in that field. The preclinical program REC-4209 in idiopathic pulmonary fibrosis was also cut, and REC-4539, an LSD1 inhibitor for small-cell lung cancer, was paused pending data. The company said the totality of the available data did not justify continued investment in the discontinued assets.
The retained portfolio centered on a smaller set of oncology and rare-disease programs. The table below summarizes the prioritized programs the company highlighted after the 2025 restructuring; clinical statuses are as reported by Recursion and evolve over time.
| Program | Type / target | Indication | Status (per Recursion) |
|---|---|---|---|
| REC-617 | CDK7 inhibitor | Advanced solid tumors | Clinical (oncology) |
| REC-1245 | RBM39 degrader | Solid tumors / lymphoma | Phase 1 (DAHLIA trial) |
| REC-3565 | MALT1 inhibitor | B-cell malignancies | Clinical (oncology) |
| REC-7735 | PI3Kalpha inhibitor | Solid tumors | Clinical (oncology) |
| REC-4881 | MEK1/2 inhibitor | Familial adenomatous polyposis (FAP) | Clinical proof of concept reported |
| REV102 | Rare disease candidate | Hypophosphatasia | Preclinical / early development |
In its Q1 2026 update, Recursion reported encouraging signals for REC-4881 in FAP, describing a reduction in precancerous polyps and the start of discussions with the U.S. Food and Drug Administration about a registrational path, while cautioning that these were early findings. The company also reported that REC-1245's first-in-human DAHLIA study had dosed 16 patients across four dose levels without dose-limiting toxicities at that point, and that REC-4539 had entered Phase 1 in April 2026.
Recursion has signed a series of large pharmaceutical collaborations that provide non-dilutive funding and validate its platform. On December 7, 2021, it announced a collaboration with Roche and Genentech in neuroscience and oncology that included a $150 million upfront payment; under the agreement the partners could initiate up to 40 programs, each potentially worth more than $300 million in milestones to Recursion plus tiered royalties. Recursion has also maintained collaborations with Bayer (later refocused on precision oncology) and Sanofi, and Exscientia brought additional partnerships including work with Merck KGaA. The company has stated that it has received more than $500 million in cumulative platform and partnered-program inflows to date, with substantial additional milestones theoretically available across the partnerships. Individual milestone payments are typically modest; for example, in early 2026 Recursion noted a fifth Sanofi milestone of $4 million.
The relationship with Nvidia has been both commercial and financial. On July 12, 2023, Nvidia made a $50 million strategic investment in Recursion alongside a collaboration to build large foundation models in biology and chemistry, give Recursion priority access to Nvidia GPUs through DGX Cloud, and explore distributing some of Recursion's models through Nvidia's BioNeMo platform. Nvidia hardware also underpins the BioHive systems.
In June 2025, Recursion and the Massachusetts Institute of Technology released Boltz-2, an open-source biomolecular foundation model, with engineering contributions from Nvidia. Developed primarily by researchers at MIT's CSAIL and Jameel Clinic, Boltz-2 was the first openly released model to combine three-dimensional structure prediction with binding-affinity prediction in a single pass, building on the late-2024 Boltz-1 model, which had approached the accuracy of systems such as AlphaFold 3 for predicting biomolecular complexes. The collaborators reported that Boltz-2 approaches the accuracy of physics-based free-energy perturbation calculations while running roughly 1,000 times faster, and released it under a permissive MIT license for both academic and commercial use, an unusually open posture for a company that otherwise guards its data tightly.
Recursion's ambitions carry a high cost, and the company has consistently operated at a loss while investing in its platform. For the full year 2025 it reported total revenue of about $74.7 million (up from $58.8 million in 2024) and a net loss of roughly $644.8 million (versus $463.7 million in 2024), with collaboration milestones, particularly from Roche, Genentech and Sanofi, driving a strong fourth quarter. It ended 2025 with about $753.9 million in cash, cash equivalents and restricted cash.
For the first quarter of 2026, reported on May 6, 2026, Recursion posted revenue of $6.5 million (down from $14.7 million a year earlier) and a narrowed net loss of $117.5 million (versus $202.5 million). Research and development expense fell to $87.9 million from $129.6 million, and general and administrative expense to $34.6 million from $54.7 million, reflecting post-merger cost cuts. The company ended the quarter with about $665.2 million in cash and reiterated guidance for full-year 2026 operational cash burn of less than $390 million, supporting a runway it said extends into early 2028.
These figures followed a difficult 2025. In June 2025 Recursion announced a workforce reduction of about 20 percent, affecting roughly 160 of about 800 employees and triggering an estimated $11 million in charges; Gibson attributed the cuts to the maturation of the company's AI tools and tight biotech capital markets. On the leadership side, the company appointed Vicki Goodman as chief medical officer in early 2026, citing her prior oncology development experience in large pharmaceutical companies. The combination of platform investment, an unproven late-stage pipeline, and a multi-year cash runway leaves Recursion a closely watched test of whether AI-driven drug discovery can translate into approved medicines.