Recursion Pharmaceuticals: FAP Registrational Trial and RBM39 Early Signal Emerge as Near-Term Catalysts, But Cash Discipline Remains the Defining Story
Bank of America Global Healthcare Conference, May 12, 2026 — CFO Ben Taylor outlines pipeline milestones and capital framework
Recursion Pharmaceuticals used its appearance at the Bank of America Global Healthcare Conference to do something relatively rare for an AI-driven drug discovery company: anchor the investment case squarely in near-term clinical outcomes rather than platform promise. CFO Ben Taylor was direct that the company sees itself first as a therapeutics company, and the next 12 to 18 months will either validate or pressure-test that framing across five programs with incoming clinical data.
FAP and RX-4881: Moving Toward a Pivotal Trial With Real-World Data as a Weapon
The most advanced and commercially significant program remains RX-4881 in familial adenomatous polyposis (FAP), a hereditary condition in which patients develop hundreds to thousands of malignant GI polyps throughout their lives, face colectomies before age 30, and accumulate on average ten serious surgeries over a lifetime — a figure Recursion generated in roughly one week using an LLM querying 300 million patient records and identifying 250,000 FAP-related cases. Taylor emphasized this real-world dataset is actively informing FDA discussions on the registrational pathway, though he declined to characterize how it might influence trial design requirements. "The FDA loves data and loves transparency," he said. "Anyone who thinks differently hasn't worked with them."
The phase 2 data that gave Recursion confidence to pursue registration showed a roughly 50% reduction in polyp burden within three months of treatment, and critically, patients were able to maintain that benefit after discontinuing the drug — a meaningful finding for a chronic disease indication with no approved therapies. FDA discussions on the pivotal trial structure are ongoing, with an update expected later in 2026. Potential endpoints include polyp burden, the composite Spigelman score, excision frequency, and serious surgical resections, all of which represent genuine unmet need given the current standard of care. The real-world data infrastructure also allows Recursion to model enrollment pools based on inclusion/exclusion criteria and site demographics, which Taylor said has driven 30% to 60% improvement in enrollment rates across trials using the company's ClinTech tools.
RBM39: A Novel Degrader With Early PK/PD Validation, Full Data Set Due Second Half
RBM39 is arguably the most scientifically novel program in the pipeline and the one with the most binary risk profile. The target — identified through Recursion's phenomics platform as a proxy for CDK12 biology — has never been therapeutically exploited. CDK12 itself has been of longstanding interest in oncology given its relevance to high mutational burden cancers, but direct inhibition has been frustrated by the near-identical binding pockets of CDK12 and CDK13, with CDK13 inhibition generating significant toxicity. Recursion's approach was to identify a separate target that recapitulates the desired biology, then design a degrader molecule against it.
At the Q1 earnings call just prior to the conference, the company disclosed initial first-in-human data showing dose-proportional pharmacokinetics and pharmacodynamics aligned with design targets — specifically, the protein knockdown profile needed to sustain greater than 70% target suppression throughout the day was being observed as dose escalation approached predicted therapeutic levels. Taylor described this as a "futility analysis" mindset: move efficiently, confirm the mechanism is behaving as modeled, and then look for early efficacy signal in monotherapy before committing broader resources. A more comprehensive dataset is expected in the second half of 2026. The addressable patient population, particularly in MSI-high and synthetically lethal tumor types, is large, and Taylor noted the program could "be really, really exciting" — though the degree of clinical signal from monotherapy remains genuinely uncertain at this stage.
RX-7735: Mutant-Selective PI3K Inhibitor Targeting a Gap in the Existing Competitive Field
Recursion also discussed RX-7735, a mutant-selective PI3K alpha inhibitor announced earlier in 2026. The space is competitive — Incyte, Relay Therapeutics, and others are active — but Taylor argued that Recursion's molecule offers approximately an order of magnitude greater selectivity over wild-type PI3K than anything currently advancing through development. The clinical implication is twofold: first, a potential to enroll patients currently excluded from approved PI3K therapies like Piqray due to diabetes or prediabetes — a population Taylor estimated at roughly half of relevant cancer patients — and second, the possibility of pushing to deeper doses before reaching maximum tolerated doses, potentially driving higher response rates. Real-world analysis showed that the median time on Piqray is only a few months, driven primarily by the hyperglycemia burden. IND-enabling work is ongoing, with data expected in the second half of 2026.
Sanofi and Roche Partnerships: Breakeven Today, Optionality Tomorrow
The partnership portfolio with Sanofi and Roche has generated over $500 million to date, with seven milestones hit in approximately two years — five with Sanofi, two with Roche. Taylor characterized the current economics as "breakeven to a mild profit" because partners prepay operational expenses. The more interesting dynamic is what happens as programs advance: Sanofi opt-ins and Roche milestone progression would convert these partnerships from cost-covered activities into pure profit streams with no remaining operational obligations for Recursion. That inflection point is not imminent but is a meaningful component of the medium-term financial model that deserves monitoring.
Capital Discipline: 35% Budget Reduction Post-Merger and a Sub-$390 Million Expense Target
Perhaps the most underappreciated aspect of Taylor's presentation was the rigor he described around cost management. Post-merger with Exscientia, Recursion applied an analytical impact-probability framework to the entire budget and cut everything that could not clearly demonstrate high expected value. The result was a 35% reduction in total spend. The 2026 expense guidance sits below $390 million — covering five clinical programs, two preclinical programs, and two major pharma partnerships — and Taylor said the company is "focused on getting to that less number." He also pushed back on a common perception of AI drug discovery companies: approximately 70% of Recursion's costs flow directly into pipeline programs and partnerships, not into standalone platform development. "That's not how we do it," he said. "It's applied platform and pipeline development."
Platform Science: Multimodal Biology and the Virtual Cell Are Real, But Distant
Taylor offered one of the more grounded assessments of AI drug discovery's actual state of progress heard at a public conference. He framed the industry's remaining challenge as navigating from point solutions — where most competitors operate — to integrated systems capable of addressing the multiple, distinct reasons clinical trials fail: biology, chemistry, patient selection, and trial design simultaneously. Recursion's claim is that its unified system and 50-plus petabyte proprietary dataset, built through years of applied work rather than public data aggregation, enables more predictive modeling than the industry norm. He cited a recent publication showing that smaller, highly annotated datasets outperform massive but poorly annotated ones in predictive power — a finding with direct implications for how the company thinks about data quality versus scale.
On the virtual cell concept — a term Taylor acknowledged is heavily used and often abused — Recursion published work in Nature Biotechnology showing that multimodal biology combining cellular phenomics, transcriptomics, and other data streams can generate experimental predictions for cell lines that were entirely absent from training data. "We're just tip of the iceberg on it," Taylor said, resisting any suggestion that this represents a near-term commercial advantage. The honest read is that these capabilities are scientifically meaningful but operationally early, and investors should not price them as near-term value drivers.
The next several months will determine whether Recursion's blend of platform science and clinical execution deserves a re-rating. The company's credibility rests on 4881's registrational path crystallizing, RBM39 showing early tumor signal, and 7735 reaching the clinic — all against a backdrop of tightening capital discipline that leaves limited room for program slippage.
Recursion Pharmaceuticals Deep Dive
Business Model and Strategy
Recursion Pharmaceuticals operates at the intersection of technology and biology, pioneering a TechBio business model designed to industrialize the notoriously inefficient drug discovery process. Historically, pharmaceutical research has relied on artisanal, hypothesis-driven science, which suffers from high attrition rates and skyrocketing costs. Recursion flips this paradigm by utilizing a massive, automated wet-lab infrastructure combined with machine learning to map cellular phenotypes. Through its proprietary operating system, the company conducts millions of automated cellular experiments each week, capturing high-resolution biological images that allow computational models to identify complex patterns of human disease without human bias. This data-first approach transforms drug discovery from a sequential, low-throughput process into a highly parallelized, industrialized engine.
The company monetizes its platform through a dual-pronged strategic model. First, it advances a wholly owned and co-developed pipeline of therapeutic candidates targeting oncology and rare genetic diseases. This internal pipeline offers the highest potential for long-term value creation but carries traditional clinical and regulatory risks. Second, Recursion mitigates this cash-intensive risk profile through lucrative, milestone-driven discovery partnerships with major biopharmaceutical companies. By licensing access to its platform and collaborating on target discovery, the company generates crucial non-dilutive capital, effectively shifting late-stage clinical risk and commercialization costs to its larger partners while retaining milestone and royalty upside.
A transformative evolution in Recursion's strategy occurred in late 2024 with the $688 million all-stock acquisition of United Kingdom-based artificial intelligence drug designer Exscientia. Prior to this merger, Recursion was heavily indexed toward target identification and phenotypic biology but lacked deep capabilities in the subsequent step of precision chemistry. Exscientia filled this void by bringing an automated small-molecule synthesis platform and advanced generative chemistry tools. The integration of these two platforms effectively created a full-stack, end-to-end drug discovery pipeline, allowing the combined entity to seamlessly transition from identifying a novel biological target to rapidly designing and synthesizing a highly optimized small molecule.
Customers, Partners, and Suppliers
Recursion's immediate commercial focus relies heavily on enterprise-level partnerships with top-tier pharmaceutical companies, which serve as its primary source of operating revenue. The most significant of these is a sweeping collaboration with Roche and Genentech. Under this agreement, the partners are exploring up to 40 programs across neuroscience and oncology, carrying a staggering theoretical deal value in excess of $12 billion in long-term milestones. Similarly, the company has an extensive multi-target collaboration with Sanofi, focusing on up to 15 best-in-class programs in immunology and oncology. Sanofi has already contributed well over $130 million in upfront and milestone payments, with each program eligible for over $300 million in downstream milestones plus tiered royalties. A strategic partnership with Bayer further validates the platform's utility for large-scale discovery.
The ultimate end-customers for Recursion are patients suffering from high-unmet-need conditions, particularly in precision oncology and rare orphan disorders. By targeting diseases with well-defined genetic drivers, such as familial adenomatous polyposis, the company seeks to leverage expedited regulatory pathways and address populations where standard-of-care treatments remain severely limited.
On the supply side, the company's reliance on immense computational power makes high-performance hardware vendors its most critical suppliers. Recursion's capabilities are anchored by its BioHive-2 supercomputer, which is powered by 504 Nvidia H100 GPUs and stands as the largest supercomputing cluster in the pharmaceutical industry. Nvidia has historically been a critical strategic partner, injecting $50 million into Recursion in 2023 to optimize foundational models on the BioNeMo platform. While Nvidia liquidated its public equity position in Recursion at the end of 2025, the deep technical collaboration remains active. The company also relies heavily on cloud infrastructure providers, notably expanding a partnership with Google Cloud to support data storage and compute scalability for its 50 petabyte proprietary dataset.
Market Landscape and Competitive Advantages
The artificial intelligence drug discovery landscape is highly fragmented and aggressively funded, characterized by distinct technical lanes. Recursion operates in a crowded arena alongside computationally heavy peers like Schrodinger, which relies on physics-based molecular simulation, and Relay Therapeutics, which utilizes its Dynamo platform to target dynamic protein structures. In the biologics space, competitors like Generate Biomedicines use generative models to design novel therapeutic proteins from scratch. Meanwhile, Insilico Medicine has established itself as a formidable direct competitor in small molecules, becoming one of the first AI-native companies to advance a fully computationally designed asset into Phase 2 human efficacy trials.
Despite this fierce competition, Recursion commands a distinct competitive advantage derived from the sheer scale and integration of its proprietary data ecosystem. While many AI startups rely on public, fragmented, or retrospective datasets to train their models, Recursion generates its own ground-truth biological data through a closed-loop system. The company synthesizes vast amounts of multi-omic and chemical data in real-time, feeding insights back into its algorithms to continuously refine their predictive accuracy. This integration of a high-throughput wet lab with predictive computation is incredibly difficult for software-only startups to replicate.
This structural advantage manifests in tangible operational efficiencies and low input costs. Recursion reports that it now synthesizes an average of just 330 compounds per development candidate, representing a 90 percent reduction compared to the industry standard of 2,500 to 5,000 compounds. Furthermore, the platform accelerates the timeline from discovery to advanced development candidate to roughly 17 months, less than half the industry norm of over 40 months. These metrics highlight an industrial-scale competitive moat where capital and time efficiency offer a structural edge over legacy pharmaceutical research.
New Products and Clinical Pipeline
Following the Exscientia integration, Recursion initiated a rigorous portfolio optimization in 2025, discontinuing several mid-stage assets, including REC-994 for cerebral cavernous malformation, which met safety endpoints but failed to demonstrate meaningful clinical benefit. This clinical reality check forced management to pivot resources toward its most promising oncology and rare disease candidates, resulting in a leaner but more focused pipeline heading into late 2026.
The centerpiece of the company's clinical narrative is currently REC-4881, an allosteric MEK1/2 inhibitor being evaluated for familial adenomatous polyposis. In early 2026, the company reported strong Phase 2 proof-of-concept data for the asset, demonstrating a median 43 percent reduction in total polyp burden with a 75 percent patient response rate. Crucially, the drug showed durable effects even after patients were taken off treatment. With regulatory engagement actively underway to define a registrational pathway, REC-4881 represents the first significant clinical validation of Recursion's end-to-end platform.
Further down the pipeline, the company is advancing a new generation of sophisticated targeted therapies. REC-1245, a novel RBM39 degrader targeting biomarker-enriched solid tumors and lymphoma, is currently in a Phase 1 study where it has demonstrated a well-tolerated safety profile with no dose-limiting toxicities among the first cohorts. Additionally, REC-4539, a highly selective LSD1 inhibitor developed entirely through the platform's generative chemistry capabilities in under 20 months, recently dosed its first patient. The rapid progression of these assets from computational prediction to human trials serves as live ammunition for management's thesis that their platform can reliably churn out high-quality development candidates at scale.
Industry Dynamics and New Entrants
The biopharmaceutical industry is facing a severe productivity crisis, often referred to as Eroom's Law, wherein the cost of developing a new drug doubles roughly every nine years. This macroeconomic pressure creates a highly receptive environment for TechBio platforms. Large pharmaceutical companies are eager to offload the earliest, riskiest stages of target discovery to specialized tech partners. This dynamic ensures that companies with validated platforms will continue to see robust demand for licensing and co-development deals.
However, the industry is simultaneously undergoing a period of hyper-innovation that threatens to commoditize certain layers of the AI drug discovery stack. The most existential threat comes from Alphabet spin-off Isomorphic Labs. Armed with a monumental $2.1 billion Series B round secured in May 2026, Isomorphic is advancing a proprietary drug design engine that has demonstrated unprecedented accuracy in predicting antibody-antigen interfaces and complex molecular structures. Backed by the computational dominance of Google DeepMind and substantial capital from sovereign wealth funds, Isomorphic represents a heavyweight new entrant that could monopolize premium discovery partnerships.
Simultaneously, the open-source community is rapidly eroding the proprietary moats of foundational biological models. The release of open-weight models like Chai-1 and Boltz-2 over the last year successfully replicated the capabilities of Google's AlphaFold 3, democratizing access to biomolecular structure prediction. As the software algorithms become commoditized, the ultimate battleground shifts away from in silico predictive accuracy and toward in vivo clinical efficacy. The persistent translation gap, defined as the reality that a highly optimized digital molecule can still fail unpredictably in human biology, remains the industry's most significant hurdle, serving as both a headwind for the sector and a potential differentiator for companies that can bridge the divide.
Management Track Record
Under the leadership of Chief Executive Officer Chris Gibson, management has successfully navigated an unforgiving macroeconomic environment for unprofitable biotechnology companies. Gibson has proven to be a highly effective capital allocator and narrator of the TechBio thesis, securing billions in potential deal value from legacy pharmaceutical giants while managing the narrative arc from a pure screening company to a fully integrated drug discovery powerhouse. The execution of the Exscientia merger showcased management's self-awareness regarding their internal deficits in chemistry and their willingness to execute bold, transformative acquisitions to complete their platform.
Financially, the executive team has demonstrated necessary austerity as capital markets tightened. In the first quarter of 2026, management successfully reduced operating expenses by roughly 30 percent year-over-year. By guiding operational cash burn below $390 million for the year, they extended the company's cash runway into early 2028 based on a robust $665 million liquidity position. While these cost containment measures have been applauded, management's track record is not unblemished. The failure of earlier clinical assets like REC-994 highlighted the limitations of the platform's early iterations, and the optics of Nvidia quietly unwinding its equity stake required careful investor relations maneuvering. Nevertheless, the team has delivered on partnership milestones and maintained the necessary capital buffer to see their next-generation assets through critical clinical inflection points.
The Scorecard
Recursion Pharmaceuticals has undeniably built one of the most formidable, vertically integrated drug discovery platforms in the biopharmaceutical industry. The marriage of its immense, proprietary phenomics dataset with Exscientia's precision chemistry capabilities creates a closed-loop engine that demonstrably accelerates the timeline from target identification to clinical candidate. The company's ability to reduce compound synthesis requirements by 90 percent while attracting massive validation deals from Roche and Sanofi provides a sturdy floor of non-dilutive capital. Furthermore, the promising Phase 2 efficacy data from REC-4881 offers a much-needed glimpse of clinical validation, proving that the platform can indeed yield durable, disease-modifying therapies in humans.
Conversely, the path forward is fraught with intense competitive and scientific risks. The TechBio landscape is experiencing a capitalization arms race, punctuated by Isomorphic Labs' massive war chest and the rapid proliferation of open-source foundational models that threaten to commoditize in silico discovery. Furthermore, while Recursion has proven it can generate clinical candidates with unprecedented speed, the binary risk of human biology remains the ultimate arbiter of value. The company's long-term viability hinges not on its computational elegance or the size of its supercomputer, but on its ability to consistently translate digital predictions into approved, commercialized therapeutics in a market that heavily penalizes clinical failure.