RXRX US Equity☆

rxrx-20251231

Securities registered pursuant to Section 12(b) of the Act:

Title of each class Trading symbol(s) Name of each exchange on which registered

Class A Common Stock, par value $0.00001 RXRX Nasdaq Global Select Market

Large accelerated filer x Non-accelerated filer ☐

Accelerated filer ☐ Smaller reporting company ☐

Emerging growth company ☐

Table of Contents

TABLE OF CONTENTS

PART 01 Item 1. Business 9

Item 1A. Risk Factors 72

Item 1B. Unresolved Staff Comments 135

Item 1C. Cybersecurity 135

Item 2. Properties 136

Item 3. Legal Proceedings 137

Item 4. Mine Safety Disclosures 137

Item 6. [Reserved] 139

Item 7A. Quantitative and Qualitative Disclosures About Market Risk 152

Item 8. Financial Statements and Supplementary Data 154

Item 9A. Controls and Procedures 193

Item 9B. Other Information 195

Item 9C. Disclosure Regarding Foreign Jurisdictions that Prevent Inspections 195

PART 03 Item 10. Directors, Executive Officers and Corporate Governance 196

Item 11. Executive Compensation 196

Item 14. Principal Accounting Fees and Services 196

PART 04 Item 15. Exhibits and Financial Statement Schedules 197

3

Table of Contents

PART I

Risk Factor Summary

Below is a summary of the principal factors that make an investment in the common stock of Recursion Pharmaceuticals, Inc. (Recursion, the Company, we, us, or our) risky or speculative. This summary does not address all of the risks we face. Additional discussion of the risks summarized below, and other risks that we face, can be found in the section titled “Item 1A. Risk Factors” in this Annual Report on Form 10-K.

Risks Related to Our Limited Operating History, Financial Position, and Need for Additional Capital

◦We are a clinical-stage biotechnology company with a limited operating history and no products approved by regulators for commercial sale, which may make it difficult to evaluate our current and future business prospects.

◦We have incurred significant operating losses and anticipate that we will incur continued losses for the foreseeable future, and will need to raise substantial additional funding, which may cause dilution to stockholders, restrict operations, require us to relinquish rights to our technologies or drug candidates, and divert management’s attention from our core business.

◦We may be required to repurchase for cash, or to facilitate the purchase by a third party of, the shares of Class A common stock that were issued to the Bill & Melinda Gates Foundation if we default under the global access commitments agreement with Exscientia, which could have an adverse impact on us.

◦We are engaged in strategic collaborations and we intend to seek to establish additional collaborations, including for the clinical development or commercialization of our drug candidates. If we are unable to do so, or if current and future collaborations are not successful, we may have to alter our development and commercialization plans.

◦We have no products approved for commercial sale and have not generated any revenue from product sales. We or our current and future collaborators may never successfully develop and commercialize our drug candidates, which would negatively affect our results of operation and our ability to continue our business operations.

◦If we engage in future acquisitions or strategic partnerships, this may increase our capital requirements, dilute our stockholders’ equity, cause us to incur debt or assume contingent liabilities, and subject us to other risks.

Risks Related to the Discovery and Development of Drug Candidates

◦Our approach to drug discovery is unique and may not lead to successful drug products for various reasons, including, but not limited to, challenges identifying mechanisms of action for our candidates.

◦Our drug candidates are in preclinical or clinical development, which are lengthy and expensive processes with uncertain outcomes and the potential for substantial delays, including due to difficulties in the enrollment of patients in clinical trials.

◦Our planned clinical trials, or those of our current and potential future collaborators, may not be successful and may not receive regulatory approval or market acceptance.

◦We may develop drug candidates for use in combination with other therapies, which exposes us to additional risks.

◦We conduct clinical trials for our drug candidates outside the United States, and the FDA and similar foreign regulatory authorities may not accept data from such trials.

◦It is difficult to establish with precision the incidence and prevalence for target patient populations of our drug candidates. If such data is not accurate our revenue and ability to achieve profitability will be adversely affected, possibly materially.

◦We may never realize a return on our investment of resources and cash in our drug discovery collaborations.

◦Our competitors may discover, develop, or commercialize products before, or more successfully than, we do.

◦Because we have multiple programs and drug candidates in our pipeline and are pursuing a variety of target indications and treatment modalities, we may expend our resources to pursue a particular drug candidate and fail to capitalize on development opportunities or candidates that may be more profitable or for which there is a greater likelihood of success.

◦Our product candidates may cause significant adverse events, toxicities or other undesirable side effects when used alone or in combination with other approved products or investigational new drugs that may result in a safety profile that could prevent regulatory approval or market acceptance, limit commercial potential, or result in material negative consequences.

Risks Related to Our Platform and Data

◦We have invested, and expect to continue to invest, in research and development efforts to further enhance our drug discovery platform, which is central to our mission. If the return on these investments is lower or develops more slowly than we expect, our business and operating results may suffer.

◦Our information technology systems and infrastructure may fail or experience security breaches and incidents that could adversely impact our business and operations and subject us to liability.

◦Interruptions in the availability of server systems or communications with internet or cloud-based services, or failure to maintain the security, confidentiality, accessibility, or integrity of data stored on such systems, could harm our business.

4

Table of Contents

◦Our solutions utilize third-party open source software (OSS), which presents risks that could adversely affect our business and subject us to possible litigation.

◦Issues relating to the use of artificial intelligence and machine learning in our offerings could adversely affect our business.

Risks Related to Our Operations/Commercialization

◦Even if any drug candidates we develop receive marketing approval, they may fail to achieve the degree of market acceptance by physicians, patients, healthcare payors, and others in medicine necessary for commercial success.

◦If we are unable to establish sales and marketing capabilities or enter into agreements with third parties to sell and market any drug candidates we may develop, we may not be successful in commercializing those drug candidates, if approved.

◦We are subject to regulatory and operational risks associated with the physical and digital infrastructure at both our internal facilities and those of our external service providers.

◦The manufacture of drugs is complex, and our third-party manufacturers may encounter difficulties in production or supply chain. If any of our third-party manufacturers encounter such difficulties, our ability to provide adequate supply of our product candidates for clinical trials or our products for patients, if approved, could be delayed or prevented.

Risks Related to Our Intellectual Property

◦Our success significantly depends on our ability to obtain and maintain patents of adequate scope covering our proprietary technology and drug candidate products. Obtaining and maintaining patent assets is inherently challenging, and our pending and future patent applications may not issue with the scope we need, if at all.

◦Our current proprietary position for certain drug product candidates depends upon our owned or in-licensed patent filings covering components of such drug product candidates, manufacturing-related methods, formulations, and/or methods of use, which may not adequately prevent a competitor or other third party from using the same drug candidate.

◦We may not be able to protect our intellectual property and proprietary rights throughout the world.

◦If we do not obtain patent term extension and data exclusivity for any drug product candidates we may develop, our business may be materially harmed.

◦We may need to license certain intellectual property from third parties, and such licenses may not be available or may not be available on commercially reasonable terms.

◦Changes in U.S. patent law could diminish the value of patents, thereby impairing our ability to protect our products.

◦Issued patents covering our drug product candidates and proprietary technology that we have developed or may develop in the future could be found invalid or unenforceable if challenged in the United States or abroad.

Risks Related to Acquisitions

◦The anticipated benefits of the business combination with Exscientia may vary from expectations.

◦As a company with substantial operations outside of the United States, we are subject to economic, political, regulatory and other risks associated with international operations.

Risks Related to Government Regulation

◦We may be unable to obtain regulatory approval and, as a result, may be unable to commercialize our product candidates.

◦Regulatory authorities may not accept data from trials conducted in locations outside of their jurisdiction.

◦Even if we receive FDA or other regulatory approval for any of our drug candidates, we will be subject to ongoing regulatory obligations and other conditions that may result in significant additional expense, as well as the potential recall or market withdrawal of an approved product if unanticipated safety issues are discovered.

◦Regulatory agencies actively enforce the laws and regulations prohibiting the promotion of off-label uses.

◦Though we have been granted orphan drug designation for certain of our drug candidates, we may be unsuccessful or unable to maintain the benefits associated with such a designation, including the potential for market exclusivity.

◦We are subject to U.S. and foreign laws regarding privacy, data protection, and data security that could entail substantial compliance costs, while the failure to comply could subject us to significant liability.

◦Regulatory and legislative developments related to the use of AI could adversely affect our use of such technologies in our products, services, and business.

Other Risks

◦Third parties that perform some of our research and preclinical testing or conduct our clinical trials may not perform satisfactorily or their agreements may be terminated.

◦Third parties that manufacture our drug candidates for preclinical development, clinical testing, and future commercialization may not provide sufficient quantities of our drug candidates or products at an acceptable cost, which could delay, impair, or prevent our development or commercialization efforts.

◦We may not realize all of the anticipated outcomes and benefits of our Acquisitions.

◦Our future success depends on our ability to attract, retain, and motivate key personnel.

◦We have identified material weaknesses in our internal control over financial reporting.

5

Cautionary Note Regarding Forward-Looking Statements

This Annual Report on Form 10-K contains “forward-looking statements” about us and our industry within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended. All statements other than statements of historical facts are forward-looking statements. In some cases, you can identify forward-looking statements by terms such as “may,” “will,” “should,” “would,” “expect,” “plan,” “anticipate,” “could,” “intend,” “target,” “project,” “contemplate,” “believe,” “estimate,” “predict,” “potential,” or “continue” or the negative of these terms or other similar expressions. Forward-looking statements contained in this report may include without limitation those regarding:

•our research and development programs;

•the initiation, timing, progress, results, and cost of our current and future preclinical and clinical studies, including statements regarding the design of, and the timing of initiation and completion of, studies and related preparatory work, as well as the period during which the results of the studies will become available and key milestones will be met;

•our continued ability to achieve milestones and receive associated milestone payments and royalties from current and future collaborations;

•our ability to use our combined assets from our business combination to create a fully integrated, technology-first drug discovery platform;

•our ability to reduce our cash burn;

•the timing and likelihood of our ability to shift our wet-lab from a source of data generation to a model for validating data from AI-generated results and the projected impact of our ClinTech platform on our business;

•the ability and willingness of our collaborators to continue research and development activities relating to our development candidates and investigational medicines;

•future agreements with third parties in connection with the commercialization of our investigational medicines and any other approved product;

•the timing, scope, and likelihood of regulatory filings and approvals, including the timing of Investigational New Drug applications and final approval by the U.S. Food and Drug Administration, or FDA, of our current drug candidates and any other future drug candidates, as well as our ability to maintain any such approvals;

•the timing, scope, or likelihood of foreign regulatory filings and approvals, including our ability to maintain any such approvals;

•the size of the potential market opportunity for TechBio companies, including the expected impact of AI-enabled technologies;

•the size of the potential market opportunity for our drug candidates, including our estimates of the number of patients who suffer from the diseases we are targeting;

•our ability to identify viable new drug candidates for clinical development and the rate at which we expect to identify such candidates, whether through an inferential approach or otherwise;

•our expectation that the assets that will drive the most value for us are those that we will identify in the future using our datasets and tools;

•our ability to develop and advance our current drug candidates and programs into, and successfully complete, clinical studies;

•our ability to reduce the time or cost or increase the likelihood of success of our research and development relative to the traditional drug discovery paradigm, including the use of data sets from our partners to accelerate the development of our AI-enabled technologies;

•our ability to improve, and the rate of improvement in, our infrastructure, datasets, biology, technology tools, and drug discovery platform, and our ability to realize benefits from such improvements;

•our ability to effectively use machine learning and artificial intelligence in our drug development process;

•our ability to leverage our collaborations and partnerships to develop our products and grow our business;

•our expectations related to the performance and benefits of our BioHive-2 supercomputer, Recursion OS, and our digital chemistry platform;

•our ability to realize a return on our investment of resources and cash in our drug discovery collaborations;

•our ability to sell or license assets and re-invest proceeds into funding our long-term strategy;

•our ability to scale like a technology company and to add more programs to our pipeline each year;

•our ability to acquire and generate datasets to train and develop our AI-enabled technologies;

•our ability to successfully compete in a highly competitive market;

•our manufacturing, commercialization, and marketing capabilities and strategies;

•our plans relating to commercializing our drug candidates, if approved, including the geographic areas of focus and sales strategy;

•our expectations regarding the approval and use of our drug candidates in combination with other drugs;

•the rate and degree of market acceptance and clinical utility of our current drug candidates, if approved, and other drug candidates we may develop;

6

•our competitive position and the success of competing approaches that are or may become available, including with respect to our AI-enabled technologies;

•our estimates of the number of patients that we will enroll in our clinical trials and the timing of their enrollment;

•the beneficial characteristics, safety, efficacy, and therapeutic effects of our drug candidates;

•our plans for further development of our drug candidates, including additional indications we may pursue;

•our ability to adequately protect and enforce our intellectual property and proprietary technology, including the scope of protection we are able to establish and maintain for intellectual property rights covering our current drug candidates and other drug candidates we may develop, receipt of patent protection, the extensions of existing patent terms where available, the validity of intellectual property rights held by third parties, the protection of our trade secrets, and our ability not to infringe, misappropriate or otherwise violate any third-party intellectual property rights;

•the impact of any intellectual property disputes and our ability to defend against claims of infringement, misappropriation, or other violations of intellectual property rights;

•our ability to keep pace with new technological developments, including with respect to AI;

•our ability to utilize third-party open source software and cloud-based infrastructure, on which we are dependent;

•the adequacy of our insurance policies and the scope of their coverage;

•the potential impact of a pandemic, epidemic, or outbreak of an infectious disease, such as COVID-19, or natural disaster, global political instability, or warfare, and the effect of such outbreak or natural disaster, global political instability, or warfare on our business and financial results;

•our ability to maintain our technical operations infrastructure to avoid errors, delays, or cybersecurity breaches;

•our continued reliance on third parties to conduct additional clinical trials of our drug candidates, and for the manufacture of our drug candidates for preclinical studies and clinical trials;

•our ability to obtain, and negotiate favorable terms of, any collaboration, licensing or other arrangements that may be necessary or desirable to research, develop, manufacture, or commercialize our platform and drug candidates;

•the pricing and reimbursement of our current drug candidates and other drug candidates we may develop, if approved;

•our estimates regarding expenses, future revenue, capital requirements, and need for additional financing;

•our financial performance;

•the period over which we estimate our existing cash and cash equivalents will be sufficient to fund our future operating expenses and capital expenditure requirements;

•our ability to raise substantial additional funding;

•the impact of current and future laws and regulations, and our ability to comply with all regulations that we are, or may become, subject to;

•the need to hire additional personnel and our ability to attract and retain such personnel;

•the impact of any current or future litigation, which may arise during the ordinary course of business and be costly to defend;

•our ability to maintain effective internal control over financial reporting and disclosure controls and procedures, including our ability to remediate the material weaknesses in internal control over financial reporting;

•our anticipated use of our existing resources and the net proceeds from our public offerings; and

•other risks and uncertainties, including those listed in the section titled “Risk Factors.”

We have based these forward-looking statements largely on our current expectations and projections about our business, the industry in which we operate, and financial trends that we believe may affect our business, financial condition, results of operations, and prospects. These forward-looking statements are not guarantees of future performance or development. These statements speak only as of the date of this report and are subject to a number of risks, uncertainties and assumptions described in the section titled “Risk Factors” and elsewhere in this report. Because forward-looking statements are inherently subject to risks and uncertainties, some of which cannot be predicted or quantified, you should not rely on these forward-looking statements as predictions of future events. The events and circumstances reflected in our forward-looking statements may not be achieved or occur and actual results could differ materially from those projected in the forward-looking statements. Except as required by applicable law, we undertake no obligation to update or revise any forward-looking statements contained herein, whether as a result of any new information, future events, or otherwise.

In addition, statements that “we believe” and similar statements reflect our beliefs and opinions on the relevant subject. These statements are based upon information available to us as of the date of this report. While we believe such information forms a reasonable basis for such statements, the information may be limited or incomplete, and our statements should not be read to indicate that we have conducted an exhaustive inquiry into, or review of, all potentially available relevant information. These statements are inherently uncertain and you are cautioned not to unduly rely upon them.

7

Table of Contents

Item 1. Business.

Business Overview

Recursion is a clinical-stage TechBio company with a mission to decode biology to radically improve lives. We have advanced a portfolio of differentiated internal programs and strategic partnerships powered by our integrated drug discovery and development platform, the Recursion Operating System (OS). This platform provides end-to-end, AI-native capabilities that span from novel biological ideas through the clinic, integrating multimodal biological data generation, AI-powered small molecule synthesis, and AI-enabled clinical development. All of our technologies are designed to translate complex science into medicines that matter — faster, better, and at scale — for patients who are waiting.

Historically, it has taken over ten years and an average capitalized R&D cost of approximately $2-3.5 billion to move a drug discovery project from early discovery to an approved therapeutic, with less than 4% of drug discovery programs initiated resulting in an approved medicine.1,2,3,4,5,6 Today, we are working to transform the traditional high-attrition, “V-shaped” discovery funnel by pivoting to a ‘T-shaped’ model. By leveraging advanced computational tools across biology, chemistry and clinical development, we aim to rapidly narrow a broad set of potential medicines to the candidates with the highest probabilities of success, with the goal to move programs through development more efficiently and with less attrition.

Figure 1. Illustrative. Reshaping the drug discovery funnel. Recursion’s goal is to leverage technology to reshape the typical drug discovery funnel towards its ideal state by moving failure as early as possible to rapidly narrowing the funnel into programs with the highest probability of success.

In recent years, advances in artificial intelligence and machine learning (“AI/ML”) have increasingly influenced both the technology and biopharmaceutical industries. Industry reports estimate that a majority of large biopharmaceutical companies now employ AI/ML in some aspect of drug discovery or development, and global investment in AI-enabled drug discovery has grown to several billions of dollars annually. Regulators and policymakers have also engaged more actively in this area, with AI/ML-enabled approaches being applied across multiple stages of drug discovery and development, including target identification, molecular design, chemical synthesis, clinical development, and manufacturing. We believe the increasing adoption of these technologies reflects a growing industry consensus that AI/ML has the potential to improve efficiency, decision-making, and productivity in drug discovery and development — the extent and timing of these benefits remain subject to ongoing validation and focus on proof-of-concept by leading players, including Recursion.

1 Zhou, S. and Johnson, R. (2018). Pharmaceutical Probability of Success. Alacrita Consulting, 1-42.

2 Steedman, M, and Taylor, K. (2024). Measuring the return from pharmaceutical innovation. Deloitte. 1-28.

3 DiMasi et al. (2016). Innovation in the pharmaceutical industry: New estimates of R&D costs. Journal of Health Economics. 47, 20-33.

4 Paul, et al. (2010). How to improve R&D productivity: the pharmaceutical industry’s grand challenge. Nature Reviews Drug Discovery. 9,203-214.

5 Martin et al. (2017). Clinical trial cycle times continue to increase despite industry efforts. Nature Reviews Drug Discovery. 16, 157.

6 European Federation of Pharmaceutical Industries and Associations (EFPIA). (2024). The pharmaceutical industry in figures: Key data 2024.

9

Table of Contents

Our Strategic Focus Across Three Pillars

The Recursion OS provides a common foundation for mapping biology, navigating disease space, designing molecules, and optimizing clinical trials across therapeutic areas and modalities. We deploy these capabilities to meet the specific differentiation and risk-reward needs of each program, tailoring our approach based on medical, market, regulatory, and capital considerations through a combination of internal pipeline development and strategic partnerships.

This approach allows us to balance near-term learning and proof generation with longer-term platform innovation, while allocating capital where Recursion has the highest confidence and greatest potential for differentiation. The three strategic pillars described below reflect how we operationalize this model to drive disciplined value creation and impact.

Figure 2. Recursion’s strategy is organized around three core pillars, described below.

10

Table of Contents

Pillar 1 – Translate insights to proof points – on the path to new medicines

A core pillar of our strategy is demonstrating that our AI platform can consistently translate scientific insights into medicines that deliver meaningful patient impact. In 2025, we made tangible progress against this objective, including a positive clinical readout from our familial adenomatous polyposis (FAP) program, where our novel platform-derived insight, identifying the potential therapeutic benefit of MEK1/2 inhibition in FAP, translated into clinically meaningful reductions in polyp burden in a disease with no approved therapies. We have several clinical-stage and multiple preclinical programs that are differentiated by novel biology, chemistry, and/or patient understanding from our platform. We have achieved key progress-based milestones across multiple strategic partnerships that further validate the applicability of our platform across diverse aspects of discovery and therapeutic areas. To date, we have received over $500 million in partner payments for novel data generation (e.g. maps optioned by Roche and Genentech) and advancing AI-designed small molecule programs with Sanofi and others. We expect to receive additional milestone payments as programs continue to progress. Together, these advancements across our internal and partnered pipeline provide growing evidence that our approach can convert insights into early proof points.

Figure 3. Recursion: Progress, by the numbers. 1. Includes preclinical programs that are expected to enter the clinic within the next 18 months. 2: Milestones: Potential Roche and Genentech and Sanofi milestones per small molecule program. Royalties: Recursion is eligible for tiered royalties up to high single digits (Roche and Genentech) and up to double digits (Sanofi).

Pillar 2 – Focused innovation, grounded in clear impact

We have built an end-to-end, AI-native platform that spans biological discovery, small molecule design, and clinical development, and our strategy is to continue investing selectively in capabilities that improve the probability of success, speed, and confidence of scientific and clinical decision-making. For example, a key area of focus in 2025 was the build out of our ClinTech capabilities, where we are applying data, automation, and AI to enable more efficient trial design, patient stratification, and evidence generation. Leveraging our Recursion OS platform, we have also been able to advance small molecule drug candidates that potentially solve complex design problems, while synthesizing approximately 90% fewer compounds than the industry average. Looking ahead, we will continue to direct resources toward platform capabilities that address critical bottlenecks in research and development and that we believe can drive durable differentiation and long-term value creation.

11

Table of Contents

Figure 4. The Recursion OS. The Recursion OS is designed to use AI to advance and accelerate decision-making and insight across the entire R&D value chain with the end goal to make novel medicines that matter. The Recursion OS generates value by advancing a pipeline of differentiated investigational medicines, in addition to building pipelines for our partners.

Pillar 3 – Pair bold ambition with disciplined execution

Our strategy emphasizes pairing long-term ambition with disciplined execution, clear prioritization, and prudent capital allocation. We apply rigorous go/no-go decision-making across our portfolio and focus resources on programs and initiatives where we believe we have a true strategic advantage. Actions taken in 2025, including prioritizing our clinical portfolio and streamlining operations, reflect this disciplined approach and are intended to support sustainable execution over the long term. Going forward, we will continue to balance investment in innovation with operational and financial discipline, aligning resources with our highest-impact opportunities while maintaining flexibility as we advance our mission.

Foundation – Empowering exceptional, bilingual teams to deliver impact with humanity

As Recursion works to transform how better medicines are brought to patients, we believe a new, integrated culture is essential to success. A core foundation of our strategy is our people, which we view as a critical operating advantage in translating platform capability into real-world impact. Recursion has intentionally built integrated, bilingual teams that operate fluently across science, computation, and engineering, enabling tight collaboration between wet-lab experimentation, model development, and clinical strategy. We continue to invest in our people and teams to reduce friction across workflows, accelerate iteration, and ensure that experimental design, AI models, and development decisions are informed by a common context. Paired with a culture that emphasizes rigor, accountability, and disciplined execution, this talent model is expected to enable us to pursue bold scientific ambition while consistently delivering progress toward medicines that matter — with speed, confidence, and humanity.

Building a Pipeline – Wholly Owned and Partnered Discovery

Our combined wholly-owned and partnered pipeline represents the primary vehicle for translating novel insights and capabilities from the Recursion OS into tangible medicines. We utilize a wide range of AI and automation to achieve differentiation in biology, chemical design, and clinical development, targeting areas of high unmet need with a speed and precision unique to our AI-native approach. The progression of the portfolio through clinical development represents a critical step in validating our OS-driven methodology. Our goal remains for these proprietary insights to be translated into successful clinical outcomes across our internal focus areas of oncology and rare disease, as well as for our partners across oncology, neuroscience, immunology, and other therapeutic areas with high unmet need. All of our programs target differentiated medicines in select patient populations.

12

Table of Contents

Advancing our Wholly Owned Pipeline

We are accelerating critical clinical milestones while delivering measurable progress against diseases with high unmet medical needs. To focus resources on programs with the strongest scientific rationale and the highest potential for near- and long-term impact, such as REC-4881 in FAP and REC-617 in advanced solid tumors, we streamlined our portfolio in May 2025. As part of this prioritization, the clinical programs REC-2282 for NF2, REC-994 for CCM, and REC-3964 for C. difficile were discontinued and/or partnering opportunities are being pursued.

Figure 5. Recursion’s wholly-owned clinical pipeline includes differentiated medicines across oncology and rare disease. The current pipeline consists of 5 clinical programs and 2 preclinical programs with the potential to enter Phase 1 pending go/no go decision.

Pipeline Highlights from 2025

In 2025, we reported the first clinical validation of the Recursion OS, with positive Phase 1b/2 results from our REC-4881 MEK1/2 inhibitor program in FAP. The rapid and durable reduction in polyp burden observed in the Phase 2 portion of the TUPELO study shows how unbiased phenotypic and mechanistic insights from the Recursion OS, such as MEK1/2 rescue of APC loss-of-function, can translate to novel, differentiated therapeutics for diseases like FAP. We expect to engage with the FDA to define a potential registration path for REC-4881 while further optimizing dosing schedule in the ongoing TUPELO trial, to continue to progress in this disease with no approved pharmacotherapies.

In parallel, Recursion has three other clinical studies ongoing: ELUCIDATE (Phase 1/2, REC-617, CDK7i), DAHLIA (Phase 1/2, REC-1245, RBM39 degrader) and EXCELERIZE (Phase 1, REC-3565, MALT1i). A fourth study, ENLYGHT (REC-4539, LSD1i) is expected to enter Phase 1 for solid tumors in 2026. IND-enabling studies are ongoing for REC-7735 (PI3Kα H1047Ri) and REC-102 (ENPP1i), with the potential to enter Phase 1 studies pending go/no go decision.

Anticipated Near-term Catalysts

Recursion is poised for a catalyst-rich period, with multiple programs reaching meaningful milestones over the next 24 months. In the first half of 2026, we will engage with the FDA to define a registration path for REC-4881, and we will report early monotherapy safety and PK data for REC-1245 (RBM39 degrader) during the same period. Go/no-go decisions on the initiation of Phase 1 studies for REC-7735 (PI3Kα H1047Ri) and REC-102 (ENPP1i) are expected in the second half of 2026. Additional clinical data for REC-4881 (MEK1/2i), early combination safety and PK data for REC-617 (CDK7i), and early monotherapy safety and PK data for REC-3565 (MALT1i) will be reported in the first half of 2027, with early monotherapy safety and PK data for REC-4539 (LSD1i) reported in the second half of that year.

13

Table of Contents

Impact Through Partnered Pipeline

Through our partnerships with leading pharmaceutical companies including Roche and Genentech, Sanofi, Bayer, and Merck KGaA (Darmstadt, Germany), we have secured more than $500 million in upfront and progress-based milestone payments to date, with the potential for over $20 billion in additional milestones before royalties. These global collaborations not only provide near-term cash flows but also combine our scaled biology, precision chemistry, and automated synthesis capabilities to pave the way for transformative therapies in oncology, neuroscience, immunology, and other therapeutic areas with high unmet need. By partnering with some of the best biopharmaceutical companies in their respective areas, our platform and team have an opportunity to learn from some of the most experienced in the industry. By uniting our AI-driven platforms, vast proprietary data, and deep scientific expertise, we continue to unlock powerful innovations and expand patient impact. Below are some of the latest developments illustrating this momentum:

Sanofi: Designing molecules against difficult and diverse protein targets in challenging data-poor and data-rich environments

•Small Molecule Joint Portfolio: Recursion is using its platform to discover and advance a joint portfolio of 5+ AI-driven and differentiated novel small molecule programs in immunology and oncology therapeutic areas. The joint collaboration has the potential for up to 15 AI-driven small molecule programs.

•Milestones and Collaboration: In February 2026, we achieved our fifth milestone across the collaboration, generating a $4 million payment from Sanofi. In total, we have achieved $134 million in upfront and progress-based milestones to date. There is potential for additional near-term milestones as the first programs advance towards development candidate milestones and earlier-stage programs progress.

Roche and Genentech: Turning novel insights from proprietary digital maps of complex biology into potential novel therapeutics

•Neuron Map: In partnership with Roche and Genentech, Recursion built the first whole-genome CRISPR knockout map generated from a subset of 1 trillion internally manufactured iPSC-derived neuronal cells ($30 million milestone payment, accepted in 2024). This proprietary dataset is being used in partnership with Roche and Genentech to identify potential new targets in neuroscience, a field which has historically suffered from limited new discoveries.

•Microglia Map: Recursion built and Roche and Genentech accepted a second neuroscience Phenomap, a first-of-its-kind whole-genome CRISPR knockout map generated from over 100 billion internally manufactured iPSC-derived microglial cells ($30 million milestone payment, accepted in 2025). With approximately 46 million images, the scale and quality of this proprietary map enables us, in partnership with Roche and Genentech, to leverage the power of AI to explore novel targets and pathways.

•Gastrointestinal-Oncology Advancements: We have built four proprietary Phenomaps which are being leveraged under the collaboration to identify novel insights that can be used to initiate programs for a gastrointestinal-oncology indication including continuing to advance one program optioned by Roche and Genentech.

•Milestones and Collaboration: In total, Recursion has received $213 million in upfront and milestone payments from the collaboration. Roche and Genentech have accepted six Phenomaps and initiated one small molecule program based on Phenomap insights to date. The companies have also identified a number of biological insights from Phenomaps that are now being validated or advanced as potential novel targets.

Bayer: Developing programs in challenging oncology indications with high unmet need

•Oncology: With our partners at Bayer, we are advancing multiple programs towards lead series milestones in precision oncology.

Merck KGaA (Darmstadt, Germany): Leveraging Recursion’s discovery engine to identify differentiated targets across oncology and immunology

•Oncology and immunology: With our partners at Merck KGaA, we are focused on identifying differentiated targets across oncology and immunology and assessing target tractability using our precision design chemistry platform.

14

Table of Contents

The Recursion OS – A Platform that Powers a Portfolio

The Recursion OS is a unified, AI-native operating system for drug discovery and development that integrates biology, chemistry, and clinical execution end-to-end. Built on proprietary, multimodal data generated at unprecedented scale through automated wet and dry labs and partnerships, the OS combines large-scale phenomics, emerging omics layers, AI-driven chemistry design, and clinical development intelligence into a single, closed-loop system. Powered by purpose-built models, scalable compute, and bilingual teams fluent in both science and AI, the Recursion OS enables faster translation of insights into proof points, reduces R&D bottlenecks, and supports the delivery of better medicines at scale for patients who are waiting.

Figure 6. Recursion combines proprietary multimodal data, purpose-built models and compute, and our bilingual teams and culture to create the first AI-native, end-to-end platform spanning idea through the clinic.

Rather than optimizing isolated steps, the Recursion OS improves decision-making across the entire R&D value chain—from decoding unknown biology and generating first-in-class targets, to designing synthetically feasible molecules, to selecting the right patients and executing trials more efficiently. By systematically generating, integrating, and analyzing high-dimensional experimental and real-world data, we train purpose-built machine learning and foundation models that translate complex biology into actionable insights across discovery and development. Throughout these processes, we are deploying AI agents and automated systems to help orchestrate our wet-lab experimentation and dry-lab modeling, standardizing workflows, coordinating data generation and analysis, and enabling faster, more consistent, and higher-confidence decisions at scale.

Figure 7. The Recursion OS. The Recursion OS is designed to use AI to advance and accelerate decision-making and insight across the entire R&D value chain with the end goal to make novel medicines that matter. The Recursion OS generates value by advancing a pipeline of differentiated investigational medicines, in addition to building pipelines for our partners.

15

Table of Contents

As models are increasingly integrated across biology, chemistry, and clinical data, Recursion is building a systems-level representation of how biology and chemistry function. This enables high-confidence predictions about previously untested hypotheses and shifts the role of the wet lab from primarily generating data to scaled validation of model-derived insights. In practice, the platform is used to simulate and prioritize targets, mechanisms, and chemistries with the highest probability of clinical success and a well-defined target product profile, followed by rapid experimental validation.

Bilingual Teams and Culture – Fluent in Tech and Science

Our mission at Recursion, Decoding Biology to Radically Improve Lives, flows naturally from our vision. We interpret our mission expansively and believe it to be a durable direction and source of inspiration for our team. We seek not only to radically improve the lives of patients who could benefit from the medicines we help to deliver, but the lives of those who care for those patients, the lives of our employees and their families, as well as the communities in which we operate our company.

We’ve intentionally designed our culture to fuel the pursuit of our mission. Our Guiding Principles are guideposts for scientific and technical decisions, and our Values underpin how our employees engage day-to-day with colleagues inside and outside the company. The Recursion Mindset, a deep commitment to achieving impact at unprecedented scale through new industrialized approaches, is an essential component of building our TechBio ecosystem. Our employees bring all these to life, contributing their unique expertise and experiences from their incredible breadth of fields and industries.

Figure 8. Recursion’s teams operate at the interface of many diverse fields. We have bilingual teams and cultures, scientists that understand AI, and AI researchers that understand science.

16

Table of Contents

Figure 9. Recursion’s Guiding Principles and Values support our ambitious mission. Together, these elements shape Recursion’s culture by guiding our people to high-impact decision-making and behaviors.

17

Table of Contents

Recursion In-Depth

AI-native end-to-end platform from idea to clinic: making novel medicines that matter

We have built an integrated, AI-native platform to decode complex biology and chemistry from multi-modal data into potentially transformative medicines. The strength of our platform is not defined by a single asset or model, but by the scale and quality of our core capabilities and underlying infrastructure, including:

•Proprietary Data: We have generated one of the largest relatable data sets in biopharma using our automated high throughput labs, which can run over 2 million experiments per week. Our data includes cellular phenomics, captured using brightfield microscopy, as well as chemical synthesis, transcriptomics, proteomics, ADMET, genomics, and patient data.

•Models & Compute: We use our proprietary data to train purpose-built AI models that accelerate learning and address specific bottlenecks across the R&D value chain. We largely train models using our own supercomputer, BioHive-2, one of the largest supercomputers in biopharma, built in collaboration with NVIDIA.

•People & Culture: A core differentiator is our people and culture—a unique, "bilingual" team of experts fluent in both life sciences (biology/chemistry) and technology (data science/engineering).

Figure 10.The Recursion OS. The Recursion OS is designed to use AI to advance and accelerate decision-making and insight across the entire R&D value chain with the end goal to make novel medicines that matter. The Recursion OS generates value by advancing a pipeline of differentiated investigational medicines, in addition to building pipelines for our partners.

Our Portfolio

Our portfolio reflects the industrial scale of our discovery engine, comprising a robust pipeline of wholly owned programs alongside strategic partnerships with leading pharmaceutical companies. This pipeline currently includes approximately five wholly owned programs in clinical development and roughly 15 discovery-stage programs spanning our internal and partnered efforts. Each milestone across this diverse portfolio serves as a critical proof point for our ability to translate AI-native insights into meaningful clinical candidates, with the breadth of the pipeline providing multiple, concurrent opportunities to validate the Recursion OS as a transformative engine for drug discovery.

18

Table of Contents

Figure 11. All pipeline programs leverage our platform, which has evolved, across biology to insight, insight to molecule, and molecule to patient. We leverage and track how we use our platform across every single program. For example. some programs are focused on novel biological insight, design, or both.

Our Internal Pipeline: Clinical Programs Overview in Oncology and Rare Disease

REC-4881 – Familial Adenomatous Polyposis

We are developing REC-4881, a highly potent and selective, potential best-in-class MEK1/2 inhibitor, for familial adenomatous polyposis (FAP). FAP is a genetic condition characterized by the development of adenomas throughout the GI tract. It is an orphan disease caused by inactivating mutations in APC, with most patients undergoing prophylactic colectomy due to nearly 100% likelihood of CRC by age 40.

During a collaboration with Takeda, we leveraged machine learning and automated analysis to quantify hundreds of cellular parameters linked to APC siRNA knockdown. We screened numerous compounds in this genetic background within 24 hours and identified REC-4881 as a potent molecule that rescued the phenotype in a concentration dependent manner. In preclinical studies, REC-4881 demonstrated over 1,000-fold selectivity in APC-mutant tumor cell lines and effectively inhibited spheroid growth and organization. In the APCmin mouse model of FAP, REC-4881 showed up to a 70% reduction in total polyps, surpassing celecoxib’s 30% reduction, highlighting its potential as a highly selective and efficacious therapy for FAP.

The IND was reactivated by Recursion and the Phase 1b/2 trial (TUPELO) of REC-4881 was initiated. As of December 31, 2025, Part 1 of the study is complete and Part 2 remains ongoing. In Part 1, which assessed safety, tolerability, and PK in FAP patients, REC-4881 was observed to have a safety profile consistent with other MEK inhibitors. A 4 mg dose of REC-4881 was shown to be pharmacologically active in FAP and progressed to Part 2. In May 2025, preliminary Phase 1b/2 data was shared at Digestive Disease Week 2025 for 6 patients following 13 weeks of treatment with REC-4881, demonstrating reduced polyp burden and an early safety profile generally consistent with that of prior MEK1/2 inhibitors. Expanded data was shared in December 2025 for a larger cohort of FAP patients treated for 12 weeks with REC-4881, followed by a 12 week off-treatment phase. Rapid reductions in polyp burden were demonstrated by week 13 (median polyp burden reduction of 43%), with a durability of effect and reductions maintained through the off-treatment phase at week 25 (median polyp burden reduction of 53%). The safety profile of REC-4881 was consistent with MEK1/2 inhibition, with adverse events predominantly low grade and N=4 discontinuations. This data provided the first clinical validation of the Recursion OS, from an unbiased phenotypic signal identifying MEK1/2 inhibition as a rescue mechanism for APC loss-of-function, through mechanistic confirmation and clinical translation, to positive clinical data. In the first half of 2026, we expect to engage with the FDA to define a registration path while further optimizing dosing schedule in the ongoing TUPELO trial. We expect to provide additional clinical data in the first half of 2027.

19

Table of Contents

REC-617 – Advanced Solid Tumors

REC-617 is a potential best-in-class, potent and selective oral small molecule inhibitor of CDK7 with demonstrated activity in preclinical studies. CDK7 controls cell cycle progression and gene transcription, often overexpressed in advanced stage cancers reliant on transcriptional pathways. This program utilized our generative AI and active learning platform to optimize molecule design, including non-covalent binding and improved ADME/PK for rapid absorption. This rapid design cycle enabled us to synthesize 136 novel compounds and select REC-617 as our lead candidate in under 11 months.

A multicenter, open-label, Phase 1/2 dose escalation and dose expansion study (ELUCIDATE) is currently ongoing in advanced solid tumors. Initial results from 19 patients were presented at the 2024 AACR Special Conference in Cancer Research, with data from a larger cohort of 29 heavily pretreated patients reported in November 2025. From this monotherapy dose escalation (QD and BID) portion of the study, REC-617 demonstrated signs of preliminary efficacy. One heavily pre-treated ovarian cancer patient achieved a confirmed durable partial response (PR), which correlated with significant reductions in clinical tumor markers (CA125 and TK1). Five additional patients achieved durable stable disease (SD) as their best response. REC-617 was generally well-tolerated, with adverse events predominantly low grade and the most common DLTs being nausea and thrombocytopenia. 7% (N=2) discontinued due to a treatment-related adverse event. The MTD was established at 10 mg once daily.

Monotherapy dose escalation remains ongoing to assess alternative dosing schedules, and in 2025 the ELUCIDATE study was expanded into platinum-resistant ovarian cancer (PROC), with a Phase 2 dose expansion monotherapy cohort ongoing and a Phase 1 dose escalation combination arm also initiated. Initial combination regimens include bevacizumab plus paclitaxel or pegylated liposomal doxorubicin (PLD). We expect to provide early safety and PK combination data in 2027.

REC-1245 – Biomarker-enriched Solid Tumors and Lymphoma

REC-1245 is a potential first-in-class, novel, potent, and selective molecular glue degrader of RBM39, a critical RNA-binding protein involved in alternative splicing and DNA damage repair (DDR) pathways. Leveraging the Recursion OS, we discovered that genetic knockout of RBM39 can phenotypically mimic CDK12 loss – a validated DDR target – without impacting CDK13 which, to our knowledge, is the first report of this novel biological insight. Utilizing our phenomics based platform for SAR, we synthesized 204 candidates and advanced this program from target ID to IND-enabling studies in 18 months (vs. industry average of 42 months).

Preclinical data confirmed strong anti-tumor activity, including tumor regressions in a BRCA-proficient ovarian cancer model, minimal off-target effects, and no CDK12 kinase inhibition. With over 100,000 addressable patients in the US and EU5 each year, REC-1245 has the potential to be a novel therapy in a biomarker-enriched advanced solid tumor and lymphoma patient population – either as a monotherapy and/or in combination regimens.

Following IND clearance, we initiated a Phase 1/2 study (DAHLIA) to evaluate the safety, tolerability, PK/PD, and preliminary efficacy of REC-1245 in unresectable, locally advanced, or metastatic cancers. This includes a biomarker-enriched population that may benefit most from targeted RBM39 degradation. In the third quarter of 2025, we reported updated information on the population being enrolled into the DAHLIA study, to include cancers with high genomic instability (for example endometrial cancer) and to confirm specific biomarker-enriched populations (for example 2L+ MSI-H/dMMR) based on early preclinical data that showed that REC-1245 reduces viability in tumors characterized by replication stress and DNA repair vulnerabilities (DDR defects) across multiple solid tumor types. The trial is currently enrolling at sites in the US and Canada, and we expect to share early safety and PK data from the Phase 1 monotherapy dose-escalation portion of the study in the first half of 2026.

REC-3565 – Relapsed / Refractory B-cell Malignancies

We are advancing REC-3565, our reversible allosteric potential best-in-class MALT1 inhibitor, for the treatment of patients with relapsed or refractory B-cell malignancies. A variety of mutations seen in lymphomas induce constitutive MALT1 protease activation, leading to aberrant NF-κB signaling that drives survival and proliferation of B-cell tumors. Key preclinical data demonstrates sustained anti-tumor activity as a single-agent or in combination with BTK inhibitors.

We leveraged physics-based predictive modelling using our molecular dynamics toolkit and AI-powered hotspot analysis to deliver a candidate with lower predicted safety risk in the clinic. We synthesized 344 novel compounds and advanced this program from hit ID to lead candidate in 15 months.

The molecule’s unique profile minimizes UGT1A1 inhibition risk, demonstrating superior target selectivity compared to oral competitors, both of which reported treatment-related hyperbilirubinemia in early Phase 1/2 studies. As a result, REC-3565’s enhanced selectivity supports the potential for a more favorable therapeutic index not only as a monotherapy, but also in combinations with BTK and BCL2 inhibitors. A multicenter, open-label, dose escalation Phase 1 study (EXCELERIZE) is ongoing, with the first patient dosed in April 2025. We expect to share early safety and PK monotherapy data in the first half of 2027.

20

Table of Contents

REC-4539 – Solid Tumors and Hematology Oncology

REC-4539 is a reversible, CNS penetrant, orally bioavailable, and potential best-in-class inhibitor of LSD1. LSD1 is an epigenetic enzyme that removes methyl groups from histones to control gene expression. LSD1 is abnormally overexpressed in a broad spectrum of solid tumors including lung, breast, prostate, esophageal, and bladder cancers, as well as acute myeloid leukemia, with evidence suggesting that LSD1 is a promising therapeutic target. This is exemplified within lung cancer by small cell lung cancer (SCLC). SCLC is particularly dependent on LSD1 to maintain a neuroendocrine phenotype that drives tumor cell survival in this aggressive lung cancer subtype. In AML, LSD1 has been shown to disrupt normal hematopoiesis by modulating key oncogenic pathways and transcriptional regulators like GFI1 and SNAI1. Preclinical studies demonstrate that REC-4539 shows anti-tumor activity in SCLC and AML human xenografts with limited impact on platelets.

Our program used multi-parameter optimization to design a unique candidate combining reversibility with CNS penetrance. We synthesized 414 novel candidates to arrive at our lead candidate in 22 months. Following IND clearance in January 2025, the program was placed on strategic pause in May 2025. While the broader field has faced safety challenges, REC-4539 remains highly differentiated by its optimized profile, with a potential improved therapeutic index through better management of on-target toxicities e.g. reduced impact on platelets. We now expect to initiate the Phase 1 trial (ENLYGHT) in the first half of 2026, with early monotherapy safety and PK data expected in the second half of 2027.

Deep Dive into Clinical Programs

REC-4881 for Familial Adenomatous Polyposis (FAP) - Phase 1b/2

REC-4881 is an orally bioavailable, non-ATP-competitive, allosteric small molecule inhibitor of MEK1 and MEK2 currently under development for familial adenomatous polyposis (FAP). REC-4881 demonstrated dose-dependent increases in exposure and pharmacological activity, with a safety profile consistent with other MEK inhibitors. We are currently enrolling patients in TUPELO, a Phase 1b/2, open-label, multicenter study to evaluate the effect of REC-4881 on polyp burden reduction. Orphan Drug Designation in the US and EU as well as Fast Track Designation in the US were granted to REC-4881 for FAP. Following positive clinical data from TUPELO shared in 2025, in the first half of 2026 we expect to engage with the FDA to define a registration path while further optimizing dosing schedule within the trial. We expect to provide additional clinical data from TUPELO in the first half of 2027.

Disease Overview

FAP is a rare, inherited tumor predisposition syndrome affecting more than 50,000 patients in the US and EU5, resulting from autosomal dominant mutations in the APC gene, a key negative regulator of the Wnt signaling pathway. FAP is a lifelong continuum of disease progression and intervention driven by chronic polyposis, with an almost 100% lifetime risk of colorectal cancer by the age of approximately 40 if untreated.

In adolescence and early adulthood, patients typically develop hundreds to thousands of precancerous adenomas in their colon and rectum. As disease burden increases, most patients will require a colectomy to remove the colon and manage disease progression and cancer risk. While this surgery addresses immediate cancer risk in the colon, it does not stop the development of further adenomas in the remaining rectum, pouch, or duodenum. Post-colectomy, patients with FAP still require decades of repeat endoscopies and excisional procedures. Approximately 50% of these patients will eventually require removal of the remaining rectum pouch in order to manage uncontrolled polyposis, a life-altering surgery that impacts quality of life. Disease progression continues in the upper GI tract, where approximately 90% of FAP patients will develop duodenal adenomas, which can often be difficult to manage endoscopically. Around 6% of these patients will undergo duodenectomy or Whipple surgeries, which are some of the most significant life-altering surgeries associated with high morbidity and mortality. Despite this substantial disease burden, no approved therapies currently exist for FAP.

Insights from Recursion OS

REC-4881 was identified as a potential first-in-disease therapy for FAP using a high-content phenotypic screening approach targeting APC-deficient human cells. In this screen, REC-4881 emerged as a potent allosteric MEK1/2 inhibitor that rescued an APC siRNA genetic knockdown-associated morphological phenotype. Compared to other MEK inhibitors, REC-4881 demonstrated a highly selective and concentration-dependent response, suggesting best-in-class potential. As a result, REC-4881 was in-licensed from Takeda and subsequently advanced into preclinical studies.

21

Table of Contents

Figure 12. Discovery of REC-4881 in Recursion OS. Compared to thousands of other molecules tested, REC-4881 rescued phenotypic defects associated with APC siRNA genetic knockdown.

REC-4881 is an orally bioavailable, non-ATP-competitive allosteric inhibitor of MEK1 (IC50: 2-3 nM) and MEK2 (IC50: 3-5 nM) being developed as a potential first-in-disease therapy for FAP. Loss of APC disrupts β-catenin regulation, leading to uncontrolled Wnt signaling, RAS stabilization, and ERK pathway activation, which drives MYC-dependent proliferation. REC-4881 inhibits MEK1/2, and blocks ERK phosphorylation downstream. This reduces MYC expression levels in the cell and potentially restores Wnt pathway control. Given ERK signaling activity in both adenoma epithelium and tumor stroma, as well as frequent MAPK-activating mutations in FAP, MEK inhibition offers a targeted strategy to suppress disease progression.

Figure 13. REC-4881 inhibits APC-mutation induced MAPK signaling to block cell proliferation in the context of FAP. A potential mechanism of action of REC-4881 in cells with loss of function mutations in APC.7

7 Jeon, WJ, et al. (2018). Interaction between Wnt/β-catenin and RAS-ERK pathways and an anti-cancer strategy via degradations of β-catenin and RAS by targeting the Wnt/β-catenin pathway. NPJ Precision Oncology, 2(5).

22

Table of Contents

Preclinical

REC-4881’s activity was validated in tumor cell lines and spheroid models derived from APC-mutant human epithelial tumor cells. In these systems, REC-4881 inhibited spheroid growth and disrupted cellular organization, demonstrating over 1,000-fold selectivity in APC-mutant cells. In a disease-relevant FAP model, ApcMin/+ mice were treated with multiple oral doses of REC-4881 or celecoxib over eight weeks. While celecoxib reduced polyp formation by approximately 30% compared to vehicle, REC-4881 treatment led to a reduction of 50% (1-3 mg/kg), and 70% (10 mg/kg). Mice that were treated with 10 mg/kg REC-4881, the highest dose tested, exhibited an approximately 70% reduction in total polyps. Histological analysis of gastrointestinal tissues further revealed that, unlike celecoxib, which primarily affected benign polyps, REC-4881 significantly reduced both benign polyps and high-grade adenomas. These findings suggest that REC-4881 not only limits early polyp formation but may also inhibit progression to advanced adenomas, highlighting its potential to address both pre- and post-colectomy FAP populations.

Figure 14. REC-4881 reduces GI polyp count and pre-cancerous, high-grade adenomas in the APCMin/+ mouse model of FAP. GI polyp count (left) and the percentage of high-grade adenomas (right) after oral administration of indicated dose of REC-4881, celecoxib, or vehicle control for 8 weeks. Polyp count at the start of dosing reflects animals sacrificed at the start of study (15 weeks of age). P < 0.001 for all REC-4881 treatment groups vs. vehicle control. Quantification of high-grade adenomas versus total polyps was based on blinded histological review by a pathologist. While celecoxib reduces benign polyps, most remaining lesions are high-grade adenomas. By contrast, REC-4881 reduces both polyps and high-grade adenomas.8

Clinical

REC-4881 has been evaluated in multiple clinical studies, demonstrating a well-tolerated safety profile and pharmacological activity.

Phase 1 Oncology Studies

In a prior dose-escalation study (C20001) conducted by Millennium Pharmaceuticals in 51 participants with advanced solid tumors, REC-4881 (formerly TAK-733) was administered at doses ranging from 0.2 mg to 22 mg once daily on days 1–21 of 28-day treatment cycles. The maximum tolerated dose (MTD) was determined to be 16 mg. The most common adverse events (AEs) were rash (67%; 57% Gr1-2, 10% Gr3), diarrhea (29%, All Gr1-2), and increased blood CPK (20%, 10% Gr1-2, 10% Gr3). Treatment-related serious adverse events (SAEs) were infrequent. No unexpected safety concerns emerged, and pharmacokinetic analyses showed a less-than-dose proportional increase in exposure.

REC-4881-101 (Healthy Volunteers)

We conducted a Phase 1 study to evaluate the safety and pharmacokinetics of REC-4881 in 25 healthy participants receiving single doses of 4 mg, 8 mg, and 12 mg. REC-4881 was well tolerated, with no SAEs or dose-related safety concerns. The most common treatment-emergent adverse events (TEAEs) were mild and self-limiting, including transient blurred vision and vitreous floaters. No QTcF abnormalities were observed.

8Data on file.

23

Table of Contents

Phase 1b/2 in FAP (TUPELO)

We are currently enrolling patients in a Phase 1b/2 open-label, multicenter study (TUPELO) evaluating the efficacy, safety, pharmacokinetics, and pharmacodynamics of REC-4881 in FAP. Part 1 assessed safety, tolerability, and pharmacokinetics in FAP patients receiving 4 mg once daily for 14 days. REC-4881 was generally well-tolerated, with a safety profile consistent with other MEK inhibitors. Pharmacodynamic data showed that the 4 mg dose was pharmacologically active in FAP, and this dose progressed to Part 2 of the study. Part 2 is evaluating efficacy, safety, and pharmacokinetics in post-colectomy FAP patients with confirmed germline APC mutations. Participants will receive once-daily REC-4881 for three months (the on-treatment phase, with readout at week 13), followed by a 3 month off-treatment phase (with a readout at week 25).

Figure 15. TUPELO study design. Phase 1b/2 clinical study to assess the efficacy, safety, and pharmacokinetics of REC-4881 in patients with classical familial adenomatous polyposis (FAP)

As of December 2025, treatment with REC-4881 (4 mg QD) demonstrated meaningful and durable reductions in polyp burden in patients with FAP within the Phase 2 portion of TUPELO. A rapid reduction in polyp burden was reported at week 13. The majority of evaluable patients responded, with 75% showing reductions in polyp burden, and a median 43% reduction in total polyp burden observed among 12 efficacy-evaluable patients. 40% of patients also achieved a ≥1-point improvement in Spigelman stage from baseline, which is a clinically meaningful measure of upper GI disease severity to assess surveillance and clinical management. Durability of effect was maintained at week 25, following the 12 week off-therapy phase. 82% (N=9) of 11 evaluable patients responded to treatment with REC-4881, with a 53% median reduction in total polyp burden observed from baseline, which is shown in the figure below. 40% of patients also maintained a ≥1-point improvement in Spigelman stage from baseline.

24

Table of Contents

Figure 16. Durable reductions in polyp burden at week 25 of the TUPELO study. Percent change in baseline in total polyp burden at week 25 (12 weeks on therapy / 12 weeks off therapy) for patients with FAP treated with 4mg REC-4881. Percent (%) change from baseline calculates the change between post-resection value from screening visit to the pre-resection value at Week 25/EOT visit. Note: Polyp burden defined as the sum of all diameters of polyps in the GI. 1Non-responder with 233% increase – polyp burden increased from 3mm to 10mm due to one polyp growth at Week 25. Efficacy Evaluable Population: Defined as all participants who have measurable disease (non-zero polyp burden) at end of baseline endoscopy, received at least 75% of study drug, and have at least one post-baseline on study endoscopic assessment. One patient who had a week 13 endoscopy did not have a Week 25 endoscopy.9

The safety profile of REC-4881 4 mg QD across the combined Phase 1b/2 cohort (19 safety evaluable patients) was consistent with prior MEK1/2 inhibitors. Treatment-related adverse events were predominantly low grade and N=4 discontinuations occurred due to TRAEs. The most frequent TRAEs (at greater than or equal to 10%) included dermatitis acneiform (57.9%; 52.6% Grade 1/2, 5.3% Grade 3) / rash (31.6%; all Grade 1/2) and blood CPK increase (36.8%; 26.3% Grade 1/2, 10.5% Grade 3).

Natural History Analysis

In December 2025, we reported on a natural history analysis in collaboration with Amsterdam University Medical Centre to contextualize the single-arm efficacy TUPELO trial of REC-4881, and to better understand the natural history of FAP. The study analyzed a subset of 55 patients from a FAP registry which met key inclusion criteria of TUPELO. Results suggested that the natural history of FAP is to progress: 87% of untreated patients in the registry experienced annualized increase in polyp burden, with 10% being stable and 3% experiencing a modest decrease in polyp burden, shown in the figure below. A mean increase of 60% and median increase of 28% in annualized polyp burden was observed.

9 Data on file.

25

Table of Contents

Figure 17. Annualized percent change in polyp burden in a natural history cohort. Each bar represents one patient. 6 patients had separate endoscopic evaluations for upper and lower GI involvement and have repeated bars. Data includes 55 patients aged ≥55 with a history of colectomy and measurable polyp burden at baseline endoscopy. In routine care, endoscopies for lower and upper GI are performed annually with variability. Therefore, polyp burden percent change was annualized. 52 (97%) of the 55 patients had an increase or stable polyp burden.10

Competitors

No drugs have been approved for FAP patients, though some generic drugs are used in selected patients to reduce polyp burden. The following programs represent the most clinically advanced efforts specifically evaluated in FAP populations:

•Flynpovi (Panbela Therapeutics) — A fixed-dose combination of sulindac and eflornithine. The combination of the individual drugs was previously evaluated vs. the individual drugs alone in a randomized Phase 3 trial in FAP. Post hoc analyses suggested a potential effect on delaying lower gastrointestinal surgery; Panbela had indicated plans for a new Phase 3 study in FAP patients, including those with an intact colon, although the company has provided no updates on development plans since 2023.

•eRapa (Biodexa Pharmaceuticals) — A formulation of rapamycin currently being evaluated in a Phase 3 study in FAP patients both prior to and following colectomy, with primary completion in 2030.

◦Phase 2 data among adult patients with and without intact colon in cohort 2 (Phase 3 dose) showed a 29% median reduction in total polyp burden at 12 months.

•Eicosapentaenoic Acid (SLA Pharma) — A derivative of an omega-3 fatty acid which was under evaluation in a Phase 3 study for polyp suppression in FAP, completed over 6 years in 2024 and is pending data updates. It has previously shared data from a Phase 2/3 study:

◦A 17% reduction in polyp size (diameters) vs baseline at 6 months, which translates to 29.8% mean reduction vs. placebo.

◦A decrease 34% from baseline in global rectal polyp burden was reported vs. a 9% increase with placebo.

Depending on jurisdiction, this product may be regulated differently from traditional prescription pharmaceuticals.

The following programs are in early clinical development or include FAP as a subset of a broader development strategy:

•FOG-001 (zolucatetide, Parabilis Medicines) — A peptide-based investigational therapy with early clinical evidence reported in FAP. Additional clinical data are expected in 2026.

•TPST-1495 (Tempest Therapeutics) — A dual EP2/EP4 antagonist with a Phase 2 study in FAP(NCI run study) anticipated to initiate in 2026.

•ZKN-013 (Eloxx Pharmaceuticals / Almirall) — A small-molecule read through agent currently in a Phase 1 clinical trial that includes a cohort of patients with FAP.

10 Data on file.

26

Table of Contents

REC-617 for Advanced Solid Tumors – Phase 1/2

REC-617 is an orally bioavailable, cyclin-dependent kinase 7 (CDK7) inhibitor currently under development for the treatment of advanced solid tumors. Inhibiting CDK7 targets both cell cycle dysregulation and transcriptional "addiction", which are hallmarks of multiple aggressive cancers including, but not limited to, CDK4/6 resistant breast cancer, ovarian cancer, and other solid tumors. There are currently no CDK7 inhibitors approved by the FDA. ELUCIDATE, a Phase 1/2 open-label, multicenter, safety, PK, PD and preliminary efficacy study is currently underway. Interim monotherapy Phase 1 safety, PK, PD, and efficacy data were shared in the fourth quarter of 2024, with expanded data from a larger cohort of patients shared in the fourth quarter of 2025. Phase 1 monotherapy dose escalation is ongoing in 2026, to evaluate alternative dosing schedules. Phase 1 dose escalation combination cohorts were initiated in 2L+ PROC in 2025, with initial combination regimens including bevacizumab plus paclitaxel or pegylated liposomal doxorubicin (PLD). A Phase 2 dose expansion monotherapy cohort in PROC was also initiated in 2025. We expect to provide early safety and PK combination data in 2027.

Disease Overview

The importance of cell cycle inhibitors in oncology has been established with CDK4/6 inhibitors, which generated approximately $10.5 billion in sales in 2023. Aberrant CDK7 overexpression is common in many cancer indications and associated with poor prognosis. CDK7 presents an opportunity to improve treatment outcomes over CDK4/6 inhibitors due to CDK7’s dual role in cell cycle progression and transcription. Potential specific indications include ovarian cancer, HR+ breast cancer, triple negative breast cancer, pancreatic cancer, ovarian cancer, head and neck cancer, and NSCLC for which we estimate an addressable population of approximately 150,000 drug-treatable patients per year in the US and EU5.

Insight from Recursion OS

CDK7 inhibitor development has faced significant challenges, primarily due to off-target effects and suboptimal pharmacokinetics. Previous attempts often employed covalent binding mechanisms or exhibited poor oral bioavailability, leading to undesirable side effects in the clinic. Current candidates in development for CDK7 feature covalent binding or extended half-lives potentially resulting in substantial on-target toxicity. In addition, the reversible inhibitors under investigation are transporter substrates, likely compromising their absorption and exacerbating gastrointestinal adverse events. These limitations underscore the critical need for novel CDK7 inhibitor designs that optimize both safety and efficacy profiles.

Leveraging our AI-driven multi-parameter optimization approach, we identified critical design limitations in existing CDK7 inhibitors. This insight led to an improved target product profile and a novel molecule design. REC-617 is an orally bioavailable, potent and selective CDK7 inhibitor with enhanced oral bioavailability. It has a non-covalent, reversible mechanism of action, and a predicted shorter human half-life compared to other drugs in development. These characteristics potentially offer an improved therapeutic index, less off-target effects, and more consistent absorption.

Preclinical

REC-617 has demonstrated strong anti-tumor activities in preclinical studies, and in vivo experiments showed potent tumor regression across multiple solid tumor types. Notably, in the OVCAR3 ovarian cancer xenograft model as shown below, complete tumor regression was observed in all 8 mice treated with 10 mg/kg by Day 27. Importantly, no significant body weight loss was observed across treatment arms. Mouse PK studies revealed that maintaining 8-10 hours of CDK7 IC80 coverage resulted in potent tumor regression with minimal side effects, while coverage beyond 10 hours led to significant body weight loss. This defined an optimal therapeutic window that guided target efficacious exposures in the clinic.

27

Table of Contents

Figure 18. REC-1245 anti-tumor activity and PK in preclinical tumor models. (Left) REC-617 induces tumor regression in the OVCAR3 cell line derived xenograft mouse model. N=8, 28 days of treatment, REC-617 administered QD PO. (Right) REC-617 administration results in 8-10 hours of therapeutic coverage at IC80. PK studies conducted in CD1 mice, single-dose administration PO.11,12

Clinical

In the third quarter of 2023, we initiated a Phase 1/2 open-label, multicenter study (ELUCIDATE) in patients with advanced solid tumors, with the design shown in the figure below.

Figure 19. ELUCIDATE study design. Phase 1/2 trial design to assess the safety, PK, exploratory PD, and efficacy of REC-617 in patients with advanced solid tumors.

11 Besnard, et al. (2022). AI-driven discovery and profiling of GTAEXS-617, a selective and highly potent inhibitor of CDK7 [abstract]. AACR; Cancer Res 2022;82(12_Supplement): 3930.

12 Hallett, et al. (2024). Overcoming traditional design limitations with AI-based discovery. AACR Special Conference in Cancer Research: Optimizing Therapeutic Efficacy and Tolerability through Cancer Chemistry; Plenary Session 1

28

Table of Contents

In December 2024, we presented results from the initial 18 response evaluable patients from Phase 1 monotherapy dose escalation at an AACR Special Conference in Cancer Research. In November 2025, updated results from a monotherapy cohort of 29 heavily pre-treated patients who received 6 dose levels of REC-617 (QD and BID) were reported. REC-617 was well-tolerated with predominantly Grade 1-2 adverse events, and fewer GI side-effects than reported for other CDK7 inhibitors. The most common DLTs were nausea and thrombocytopenia, and 6.9% (N=2) of patients discontinued due to a TRAE. The MTD was established at 10 mg once daily. PK data support dose-proportional exposure (see figure below), rapid absorption, and a short half-life (~5h).

Figure 20. REC-617 clinical plasma pharmacokinetics. REC-617 demonstrates dose-proportional exposures exceeding CDK7 IC50. Exposures remain below CDK2 IC80, supporting selective target inhibition.13

Encouraging antitumor activity included a confirmed partial response (PR), in a heavily pre-treated metastatic ovarian cancer patient. The patient had a maintained durable response and was treated with REC-617 for approximately 7 months. Patient LDH levels were also normalized, and reductions were observed in CA125 (-44%) and TK1 (-68%). Five additional patients achieved the best response of stable disease (SD) lasting up to six months.

Competitors

Several investigational CDK7 inhibitor programs have entered clinical development; however, the competitive landscape remains relatively limited, with only a small number of programs currently advancing in active clinical development. These programs vary by mechanism, combination strategy, geographic focus, and development priority.

Programs with active clinical development and strategic focus

•Samuraciclib (Carrick Therapeutics) — An oral CDK7 inhibitor currently in Phase 2 clinical development, primarily in combination with selective estrogen receptor degraders (SERDs) for patients with HR-positive, HER2-negative breast cancer following progression on CDK4/6 inhibitors.

◦Recent data update from Ph2 randomized study evaluating samuraciclib + fulvestrant vs. fulvestrant alone in, fulvestrant naive, post CDK4/6+AI patient population was directionally positive but with limited quantitative power

▪Positive direction holds true across ORR, CBR, PFS and in sub-cohorts (TP53wt)

•ORR: 28% with combo (n=32) vs. 14% (n=14) with fulvestrant alone

•mPFS: 7.8mo to 8.5mo (n=39) vs. 5.6mo (n=20)

•In TP53mut not detected: mPFS of 9.6-14.5mo (n=30) vs 6.8mo (n=11)

▪The drug continues to show GI toxicities in >80% patients at RP2D (360mg)

•Q-901 (Qurient) — An intravenous CDK7 inhibitor in Phase 1/2 clinical development as monotherapy and in combination with PD-1 inhibitors across solid tumors. No efficacy data has been shared so far. Qurient is also developing a HER2-targeted antibody–drug conjugate (QP-101) that incorporates a CDK7 inhibitor payload in combination with a topoisomerase I inhibitor, currently in preclinical development.

13 Data on file.

29

Table of Contents

Other notable programs with limited or geographically-constrained clinical presence

•TY-2699a (TYK Medicines) — A CDK7 inhibitor currently in Phase 1 clinical development in China.

•EOC-237 (EOC Pharma) — A CDK7 inhibitor currently in Phase 1 clinical development in China.

REC-1245 for Solid Tumors and Lymphoma – Phase 1/2

REC-1245 is a novel, potent and selective molecular glue degrader of RNA-binding motif protein 39 (RBM39) currently under development for the treatment of biomarker-enriched solid tumors and lymphoma. There are currently no RBM39 degraders approved by the FDA. We initiated a Phase 1/2 open-label, multicenter study (DAHLIA) to evaluate the safety, tolerability, PK, PD, RP2D, and preliminary efficacy of REC-1245. In the third quarter of 2025, we reported updated information on the population being enrolled into the DAHLIA study, to include cancers with high genomic instability (for example endometrial cancer) and to confirm specific biomarker-enriched populations (for example 2L+ MSI-H/dMMR). We expect to share early safety and PK data from the Phase 1 monotherapy dose-escalation portion of the study in the first half of 2026.

Disease Overview

Alternative splicing and RNA-binding proteins (RBPs) have emerged as attractive therapeutic targets for cancer due to their critical roles in the regulation of post-transcriptional modifications, impacts on DNA damage repair pathways, and modulation of cell cycle functions. Of these, RNA-binding motif protein 39 (RBM39) is a critical splicing factor that many high-risk cancers rely on to maintain transcriptional integrity and drive tumor progression. As a target, RBM39 is vulnerable to a 'molecular glue' approach, where its selective degradation triggers a cascade of lethal splicing errors across key oncogenic pathways, including those involving DNA damage repair. With over 100,000 addressable patients, with biomarker-enriched solid tumors and other select histologies where RBM39 could be targeted in the US and EU5 each year, REC-1245 has the potential to be used as a single agent or in combination with chemotherapy and/or immunotherapy.

Insight from Recursion OS

Reports suggest that genetic or pharmacologic depletion of CDK12 can reduce the expression of several genes involved in the homologous recombination repair pathway such as BRCA1 and BRCA2, inducing a BRCA-like phenotype and DDR response. Therefore, CDK12 has received considerable interest as a therapeutic target and tumor biomarker for HR-proficient cancers; however, success has been limited by toxicity associated with CDK12 inhibitors also inhibiting the structurally related CDK13. Despite reports of functional redundancy, we observed that the genetic knockout of CDK12 could be clearly distinguished phenotypically from that of CDK13. Using map-based inference to characterize and relate cellular phenotypes, we identified RBM39 as an alternative target that selectively mimics CDK12 loss, but not CDK13, providing a novel approach for targeting CDK12 biology while circumventing any toxicities that may arise due to CDK13. We subsequently discovered REC-1245 as an RBM39 molecular glue degrader that closely mimics the phenotypic loss of CDK12 and RBM39, but not CDK13. Functionally, REC-1245 treatment globally impacts the expression of many DDR genes but does so in a CDK12 independent manner.

Figure 21. Inferred map relationships between CDK12, CDK13, RBM39, and REC-1245. Map representation demonstrates a high degree of phenotypic similarity between CDK12, RBM39, and multiple concentrations of REC-1245. CDK13 shows little or no functional similarity to CDK12, RBM39, or any concentration of REC-1245.

30

Table of Contents

Preclinical

REC-1245 is a potent, potential first-in-class RBM39 molecular glue degrader with compelling preclinical activity. It showed no significant in vitro safety concerns (CEREP, hERG), no CDK12 kinase activity, and minimal ITGA2 liability – an off-target effect seen with prior RBM39 degraders. As shown in the figures below, REC-1245 demonstrated strong antitumor activities as a single-agent, including tumor regression in an ovarian cancer BRCA-proficient, p53 mutant, OVK18 in vivo cell line derived xenograft (CDX) model. In addition, dose-dependent anti-tumor activity correlated with increases in RBM39 degradation confirming target engagement and an exposure-response-efficacy relationship.

Figure 22. REC-1245 single-agent activity and target engagement. (Left) REC-1245 administered BID PO at doses noted for 15 days. N=8 mice per group. (Right) Percent RBM39 degradation (PD) evaluated at REC-1245 doses noted after 5 days BID oral administration of REC-1245. N=3 mice per group.14

Emerging preclinical data has shown that REC-1245 reduces viability in tumors characterized by replication stress and DNA repair vulnerabilities (DDR defects) across multiple solid tumor types, including MSI-H/dMMR, HRR altered cancers, and other tumors, as shown below, which could provide a potential signature for REC-1245 sensitivity. The data also suggests greater sensitivity to REC-1245 for tumors with high replicative stress signatures and DNA repair vulnerabilities, as shown below.

Figure 23. In vitro cell viability following REC-1245 treatment. Cell lines were assigned to broad pathway dysregulation contexts based on 1 or more documented alteration from CCLE/GSDC databases.15

14 Data on file.

15 Data on file.

31

Table of Contents

Figure 24. Reduction in tumor volume across different tumors with high replication stress and DNA repair vulnerabilities. N=4 mice per group.16

Clinical

In December 2024, we initiated a Phase 1/2 open-label, multicenter study to characterize the safety, tolerability, PK, PD, and preliminary anti-tumor activity of REC-1245 in participants with unresectable locally advanced or metastatic cancer. As of December 31, 2025, the trial is currently active and enrolling at sites in the US and Canada. We expect to share early safety and PK data from the Phase 1 monotherapy dose-escalation portion of the study in the first half of 2026.

Figure 25. DAHLIA study design. Phase 1/2 trial design to assess the safety, tolerability, PK, PD, and preliminary anti-tumor activity of REC-1245 in participants with unresectable locally advanced or metastatic cancer, and who are refractory to, had a relapse on, or intolerant of, established standard of care treatment.

16 Data on file.

32

Table of Contents

The clinical development landscape for RBM39 degraders remains limited. While the mechanism of RBM39 degradation has been explored historically through legacy aryl-sulfonamide compounds, only a small number of programs are currently believed to be under active clinical development as purpose-built RBM39 degraders.

Active RBM39-focused development programs

•ST-01156 (SEED Therapeutics) — An investigational RBM39 degrader that initiated a Phase 1a open-label clinical trial in patients with advanced solid tumors, with first patient dosed in January 2026.

•E7820 (Eisai) — A legacy aryl-sulfonamide compound now understood to induce RBM39 degradation via DCAF15. Eisai is collaborating with the National Cancer Center (NCC) Japan, which is conducting a Phase 1 investigator-initiated trial (CIRCUS) evaluating E7820 in Japanese patients with unresectable tumors. Advancement to Phase 2 is expected to be considered following determination of tolerability and a recommended Phase 2 dose.

REC-3565 for B-Cell Malignancies – Phase 1

REC-3565 is an orally bioavailable, highly potent and selective, potential best-in-class MALT1 inhibitor currently under development for the treatment of B-cell malignancies, including chronic lymphocytic leukemia (CLL). MALT1 is a protease crucial for activation of the NF-κB pathway, which drives the proliferation of malignant B-cells in hematological cancers. There are currently no MALT1 inhibitors approved by the FDA. Following clearance of a CTA by the MHRA in December 2024, we initiated EXCELERIZE, a Phase 1 open-label, multicenter, dose escalation study to evaluate the safety, tolerability, PK, PD, and preliminary anti-tumor activity of REC-3565 in 2025. The first patient was dosed in April 2025, and we expect to share early safety and PK monotherapy data in the first half of 2027.

Disease Overview

B-cell malignancies encompass a range of hematological cancers, including lymphomas such as diffuse large B-cell lymphoma (DLBCL), follicular lymphoma (FL), mantle cell lymphoma (MCL), and leukemias such as CLL and small lymphocytic lymphoma (SLL). These diseases are characterized by the dysregulated growth or function of B-cells and are often driven by chronic B-cell receptor (BCR) signaling, which leads to unchecked NF-κB activation. MALT1 functions downstream of the BCR and the widely targeted Bruton’s tyrosine kinase (BTK), mediating pro-tumorigenic signals in malignant B-cells. Current therapies (e.g. BTK inhibitors) have transformed the treatment landscape, yet resistance remains a significant challenge. By inhibiting MALT1, REC-3565 may help overcome resistance and improve therapeutic outcomes, either as a monotherapy or in combination with BTK and/or BCL2 inhibitors. Notably, the total addressable population for MALT1 inhibitors spans multiple hematologic indications, with approximately 41,000 relapsed and/or refractory (R/R) patients with CLL and B-cell lymphomas in the US and EU5 annually.

Insight from Recursion OS

BTK inhibitors and other therapies for B-cell malignancies can cause drug-induced liver injury (DILI), limiting combination treatment options. Current MALT1 inhibitor scaffolds significantly inhibit UGT1A1, leading to dose-limiting toxicities, potentially restricting their utility in combination. Leveraging our AI-driven, multi-parameter optimization approach, we focused on an allosteric mechanism to enhance potency, selectivity, and safety for REC-3565. Hotspot analyses and physics-based molecular dynamics guided our design strategy, helping us address the hydrophobic and highly mobile nature of the allosteric binding site. As a result, REC-3565 does not significantly inhibit UGT1A1, potentially mitigating liver toxicity risks and facilitating higher target engagement. This profile also supports combination strategies with agents known to affect liver function like BTK and BCL2 inhibitors, offering a path to potentially deeper and more durable responses.

Preclinical

REC-3565 demonstrated significant antitumor activity across multiple B-cell lymphoma models. As a monotherapy, it drove tumor regressions in ABC-DLBCL xenografts, and in combination with zanubrutinib – a next-generation BTK inhibitor – it produced durable responses, with 70% of mice displaying no palpable tumors 10 days after the last dose. Additional in vitro analyses revealed minimal UGT1A1 inhibitory effects relative to other MALT1 inhibitor scaffolds in clinical development, suggesting an improved safety and combination therapy profile.

33

Table of Contents

Figure 26. Preclinical data highlighting REC-3565 as a potential best-in-class MALT1 inhibitor. (Left) REC-3565 showed tumor growth regression as a single agent and when combined with zanubrutinib. N=10 per group mice per group, REC-3565 and zanubrutinib dosed BID. PD evaluated after 5 days BID oral administration of REC-1245 at doses noted. N=3 mice per group in PD portion. N=8 mice per group REC-3565 administered BID PO at doses noted. (Right) REC-3565 has best-in-class potential, especially given REC-3565 has >10 uM vs. <1 uM for other MALT1 inhibitors in clinical development. Development candidate criteria: MALT1 IC50 nM: green <100 nM; yellow >100-<300 nM; red>300 nM; OCI-Ly3 IC50 nM: green <400 nM; yellow >400-<1000 nM; red>1000 nM; UGT1A1 IC50 uM: green >10 uM; yellow <10->1 uM; red<1 uM; Caco-2 A2B (efflux): green >5(<3); yellow >1-<5(>3-<10); red <1(>10).17,18

Clinical

EXCELERIZE is a Phase 1 open-label, multicenter, dose escalation study designed to evaluate the safety, tolerability, PK, PD and preliminary anti-tumor activity of REC-3565 in patients with R/R B-cell malignancies. Part A will assess monotherapy dosing to identify a recommended dose for combination in Part B, which will evaluate combination regimens to inform future studies in B-cell cancers. The first patient was dosed in the second quarter of 2025, and we expect to share early safety and PK monotherapy data in the first half of 2027.

Figure 27. EXCELERIZE study design. Phase 1 trial to evaluate safety, tolerability, PK, PD and preliminary anti-tumor activity of REC-3565 in patients with R/R B-cell malignancies.

17 Payne, et al. (2024). Combining next-generation BTK and MALT1 inhibitors to enhance efficacy and therapeutic utility in B-cell malignancies [poster]. EORTC-NCI-AACR (ENA) Symposium: PB206.

18 Data on file.

34

Table of Contents

Competitors

Early MALT1 inhibitors have shown UGT1A1 liability that has led to instances of hyperbilirubinemia in the clinic and limited combinability, while also potentially leaving efficacy on the table. The following competitors have or are currently generating data in the clinic with a strategic focus on B-cell lymphomas:

•JNJ-6786633 (Johnson & Johnson) -An oral MALT1 inhibitor, that showed significant hyperbilirubinemia in the clinic. Ph2 trials in combination with BTK inhibitors have since started and been marked complete, though no further data has been disclosed.

•SGR-1505 (Schrödinger) - An oral MALT1 inhibitor with Phase 1 monotherapy results reported for the ongoing trial in R/R B-cell lymphomas. Received US orphan drug designation for Waldenström's macroglobulinemia (WM) and mantle cell lymphomas (MCL), along with a fast track designation for treatment of WM patients in post BTK 3L+ setting.

•ABBV-525 (AbbVie/Lupin) - An oral MALT1 inhibitor with a Phase 1/2a trial ongoing in R/R B-cell malignancies. Primary completion of the trial is expected in 2029.

•AUR-112 (Aurigene) -An oral MALT1 protease inhibitor, with Phase I trial (AUR112-101) ongoing for relapsed advanced lymphoma. Initial results were reported at December 2025 press release, which showed limited Gr3 hyperbilirubinemiand 64% response rate (6 PR, 1CR) across B-cell lymphomas.

Other assets that have been discontinued or have a different strategic focus includes:

•CTX-177/ONO-7018 (Chordia/ONO) - An oral MALT1 inhibitor that was in Phase 1 trial for patients with R/R NHL/CLL which has recently been discontinued with the company looking to out license the asset.

•RB-201 (Rarefied Biosciences) - An oral MALT1 inhibitor currently in a healthy volunteer's study, with a strategic focus on autoimmune diseases.

REC-4539 for Solid Tumors and Hematology Oncology – Phase 1

REC-4539 is an orally bioavailable, highly potent and selective, CNS penetrant, and potential best-in-class inhibitor under development for the treatment of solid tumors and hematological malignancies. LSD1 is an epigenetic regulator that removes methyl groups from histones, thereby controlling the expression of tumor suppressors and oncogenes. By inhibiting LSD1, REC-4539 promotes the reactivation of tumor suppressor pathways and may slow tumor growth or enhance sensitivity to cytotoxic agents. There are currently no LSD1 inhibitors approved by the FDA. In January 2025, the FDA cleared an IND application for ENLYGHT, a Phase 1 open-label multicenter study evaluating REC-4359. In May 2025, the program was placed on strategic pause, to allow review of emerging clinical data and to ensure the program has a competitive Target Product Profile. Following completion of this review, we now expect to initiate a Phase 1 trial in the first half of 2026, with the dose escalation study evaluating REC-4539 in patients with SCLC or other select solid tumors. We expect to share early safety and PK data in the second half of 2027.

Disease Overview

LSD1 is abnormally overexpressed in a broad spectrum of solid tumors including lung, breast, prostate, esophageal, and bladder cancers, as well as acute myeloid leukemia (AML), with evidence suggesting that LSD1 is a promising therapeutic target. One indication of focus is SCLC, a poorly differentiated neuroendocrine tumor, representing roughly 15% of all lung cancer diagnoses. The majority of SCLC patients present with metastatic (extensive) or unresectable disease and, notably, over 50% of patients eventually develop brain metastases. Despite some improvements in frontline therapy such as chemotherapy plus immunotherapy, treatment options after progression remain limited. Median survival in ES-SCLC is poor, with a 5-year overall survival rate of approximately 3%. Across the US and EU5, approximately 45,000 patients have a treatable Stage III/IV SCLC each year.

Within SCLC, LSD1 plays a key epigenetic role by demethylating histones that regulate critical tumor suppressor genes. Inhibiting LSD1 can reverse this epigenetic repression, upregulating pathways such as NOTCH, that promote differentiation of neuroendocrine tumor cells into a more quiescent state, potentially sensitizing them to cytotoxic therapies. However, effective LSD1 inhibition requires a reversible, brain-penetrant molecule with a short half-life to minimize risks such as thrombocytopenia. Many LSD1 inhibitors have failed to achieve these parameters, particularly brain penetration and controlled on-target effects, highlighting the unmet need that REC-4539 aims to address.

35

Table of Contents

A second indication of focus is acute myeloid leukemia (AML), which is the most prevalent adult leukemia. The disease accounts for approximately 80% of cases and is characterized by the aggressive expansion of immature "blast cells." While advancements in management have improved outcomes for younger patients, the prognosis for the elderly population remains poor, with cure rates as low as 15%. At the molecular level, LSD1 is abnormally overexpressed, acting as an epigenetic master regulator that complexes with GFI1 to arrest myeloid differentiation and drive the self-renewal of leukemic stem cells.

While the therapeutic potential of LSD1 inhibition is well-documented across various genetic profiles, the broader clinical field has been limited by significant safety hurdles. REC-4539 addresses this critical unmet need with an optimized profile designed to maximize anti-leukemic efficacy while mitigating the systemic risks that have hindered previous candidates, offering a highly differentiated solution for a patient population with few remaining options.

Insight from Recursion OS

Developing a selective LSD1 inhibitor for solid tumor indications requires a reversible mechanism, a short half-life to minimize on-target toxicity (e.g. thrombocytopenia), and the ability to penetrate the blood-brain barrier to address frequent metastases seen in indications such as SCLC. Many existing LSD1 agents fail to meet these criteria, resulting in dose-limiting toxicity and poor CNS exposure. Using our AI-driven, multi-parameter optimization approach, we generated and screened diverse chemical scaffolds for potency, selectivity, ADME properties, and CNS penetration. Active learning identified counterintuitive yet informative compounds, enabling a rapid design breakthrough. As a result, we created REC-4539 – a potent, selective, reversible, brain-penetrant, and potential best-in-class LSD1 inhibitor with a short predicted half-life and potential improved therapeutic index through better management of on-target toxicities such as reduced impact on platelets. We believe these key attributes provide competitive differentiation for REC-4539 versus prior LSD1-targeted molecules.

Preclinical

REC-4539 demonstrated potent anti-tumor activity across multiple preclinical models, including the NCI-H1417 human SCLC xenograft. In this model, dose-dependent tumor growth inhibition correlated with a corresponding decrease in the neuroendocrine tumor biomarker progastrin-releasing peptide (proGRP). Additionally, REC-4539 treatment was well-tolerated, with minimal impact on platelet counts.

Figure 28. REC-4539 preclinical assessment in SCLC xenograft model. BALB/c mice, REC-4539 dosed BID, 28 day study (Left) REC-4539 induces dose-dependent tumor growth inhibition in the NCI-H1417 SCLC cell line derived xenograft mouse model. (Right) REC-4539 induces dose dependent reductions in plasma proGRP.19,20

19 Payne, et al. (2023). Characterizing Antitumor Responses to EXS74539, a Novel, Reversible LSD1 Inhibitor with Potential in Small-cell Lung Cancer [poster]. American Association for Cancer Research (AACR) Annual Meeting: 6290.

20 Data on file.

36

Table of Contents

Clinical

ENLYGHT is a Phase 1, open-label, multicenter dose escalation study designed to evaluate the safety, tolerability, and preliminary efficacy of REC-4539 monotherapy in patients with select solid tumors. The FDA cleared an IND application in January 2025; the program was then placed on strategic pause in May 2025. While the broader field has faced safety challenges, REC-4539 remains highly differentiated by its optimized profile, and we now expect to initiate a Phase 1 trial in the first half of 2026. The Phase 1 study will evaluate REC-4539 as a monotherapy in patients with SCLC and select other solid tumors, to assess safety and tolerability, PK/PD, to establish the MTD, and to evaluate preliminary efficacy as per the study design diagram below. We expect to share early safety and PK monotherapy data in the second half of 2027.

Figure 29. ENLYGHT study design. Phase 1 trial design to assess the safety, tolerability, and preliminary efficacy of REC-4539 monotherapy.

Competitors

The LSD1 inhibitor competitor landscape is limited to a small number of programs currently advancing in active clinical development. These programs vary by indication of interest, combination strategy, and development priority.

Active LSD1i clinical development programs:

•Bomedemstat (Merck) - Merck is focusing on myeloproliferative neoplasms (MPNs) as a route to market for bomedemstat. Bomedemstat is being investigated in a Phase 3 trial in essential thrombocythemia (ET), a Phase 2 study in myelofibrosis (MF) and polycythemia vera (PV), and a Phase 1 IIT study in AML (in combination with venetoclax). Merck terminated a Phase 1/2 SCLC trial in combination with PD-L1 maintenance in 2024 due to low accrual rates.

•Iadademstat (Oryzon) - Iadademstat is being investigated in AML and SCLC patient populations. It is in an ongoing Ph1b study for R/R AML with FLT3 mutation in combination with gilteritinib and in a Ph1b study for newly-diagnosed unfit AML in combination with venetoclax and azacitidine. It is also in a Phase 2 IIT for relapsed/refractory (R/R) SCLC and extrapulmonary high-grade NETs (in combination with paclitaxel), as well as a Phase 1b/2 IIT in first-line extensive-stage SCLC (ES-SCLC) in combination with a checkpoint inhibitor.

•JBI-802 (Jubilant Life Sciences) - JBI-802 is in a Phase 1/2 basket study, with expansion cohorts planned in SCLC, neuroendocrine prostate cancer (NEPC), and other NETs, as well as a Phase 2 study for patients with advanced NSCLC tumors harboring an STK11 Mutation in combination with pembrolizumab.

•TAS1440 (Benz Sciences/ Taiho Pharmaceuticals) -Benz Sciences are preparing for a Phase 1b/2 study for MPN patients, after in-licensing TAS1440 from Taiho Pharmaceuticals in June 2025. Taiho recently completed a Phase 1 trial targeting AML in the US population.

37

Table of Contents

Partnered Discovery

Driving Innovation Across Multiple Diseases

At the core of Recursion’s mission is the pursuit of breakthrough therapeutics for patients. Recursion collaborates with leading pharma partners to identify novel targets and therapeutic candidates across a wide range of disease areas, including neuroscience, oncology, immunology, and inflammation.

Each partnership is designed to advance therapeutic development, with multiple pathways to success:

•Novel Targets: Combining our multi-modal (phenomics, RNA sequencing) maps of human biology with real-world clinical-genomic data, we can identify novel druggable targets with potential therapeutic benefit. Validated targets may be optioned by our partners or advance within the collaborations as a therapeutic program.

•Novel Therapeutics: Using our precision chemistry platform, we can design differentiated molecules across a wide variety of targets. Resulting molecules may be optioned by our partners and advanced for further clinical development.

To date, Recursion has secured over $500 million in upfront milestone payments, with the potential to unlock over $20 billion in additional milestones before royalties. These high-impact collaborations not only generate near-term financial value but also leverage Recursion’s combined capabilities in biology and precision chemistry to accelerate the development of transformative therapies. By collaborating with top-tier biopharmaceutical companies, Recursion gains access to invaluable knowledge from some of the most experienced teams in the industry. Together, we continuously drive innovation and have the potential to expand patient impact, and revolutionize the treatment of complex diseases. Below are some of the latest milestones reflecting this exciting momentum.

Figure 30. Recursion leveraging its OS across its partnerships. 1. Milestones: Potential Roche and Genentech and Sanofi milestones per small molecule program. Royalties: Recursion is eligible for tiered royalties up to high single digits (Roche and Genentech) and up to double digits (Sanofi)

38

Table of Contents

Roche and Genentech

In December 2021, we entered a multi-year, strategic collaboration with Roche and Genentech in the field of neuroscience and a single oncology indication. Through the partnership, we are leveraging the Recursion OS and extensive single-cell perturbation screening data from Roche and Genentech, to rapidly identify novel biological relationships and advance therapeutic programs. Together, we may initiate up to 40 small molecule programs over a decade or longer. As part of this agreement, we received an upfront cash payment of $150 million, with the potential to receive milestones of more than $300 million per small molecule program plus tiered royalties.

Figure 31. Under our collaboration with Roche and Genentech, we are creating multimodal maps of cellular biology to elucidate novel targets and starting points.

Sanofi

In January 2022, we entered a strategic research collaboration with Sanofi to develop an AI-driven pipeline of precision-engineered, small molecule medicines. Through this collaboration, we are using our end-to-end integrated platform to discover and advance up to 15 novel targets in the oncology and immunology therapeutic areas. As part of this agreement, we received an upfront cash payment of $100 million, with the potential to receive up to $5.2 billion in total aggregate milestone payments plus tiered royalties.

Figure 32. Leveraging the Recursion OS with Sanofi to design small molecules against challenging targets using AI

39

Table of Contents

Bayer

In November 2023, we announced an amended and restated collaboration with Bayer. We are using the Recursion OS to identify and advance up to 7 therapeutic targets for challenging oncology indications with high unmet need. Under the terms of the agreement, Recursion is eligible to receive potential, success-based, future payments of up to $1.5 billion plus royalties on net sales.

Merck KGaA (Darmstadt, Germany)

In September 2023, we entered into a collaboration with Merck KGaA, Darmstadt, Germany. This multi-year collaboration utilizes our AI-driven precision drug design and discovery capabilities while leveraging Merck KGaA, Darmstadt, Germany’s disease expertise in oncology and immunology, clinical development capabilities, and global footprint.

Case Study 1: Delivering the World’s First Neuromap

Figure 33. Recursion launched a transformational collaboration with Roche and Genentech, delivering the world’s first whole-genome neuromap in 2024.

Overview

In 2021, Recursion launched a transformational collaboration with Roche and Genentech to create the world’s first neuromap—a comprehensive and scalable neuronal data model powered by machine learning. This effort aimed to uncover novel insights into neurodegenerative diseases, a category of illnesses that has long been difficult to tackle using traditional drug discovery methods. With a dedicated team of 50 people, Recursion set out to overcome numerous technical and biological challenges, all with the goal of driving innovation in neuroscience.

The Challenge

When the project began, the team faced significant uncertainty. The goal was ambitious: produce enough neurons, knock out genes, and generate a reliable signal from machine learning models to guide the development of potential drug programs. This challenge was particularly daunting given the complexity of neuronal cells, which do not divide or proliferate like other cell types. Unlike other cell types, such as human umbilical vein endothelial cells (HUVECs), which Recursion had previously worked with to create large-scale disease maps, neuronal cells posed a unique set of hurdles due to their limited ability to be produced at scale.

40

Table of Contents

Neurodegenerative diseases had long been a difficult area for drug development. Traditional approaches had yielded limited breakthroughs, and the complexity of the biological system presented a higher bar for success. Recursion needed to develop new technologies and methodologies to produce and analyze neuronal data on a scale not attempted before in drug discovery.

Recursion had already proven its ability to create large-scale cell maps in other disease areas, notably in gastrointestinal oncology, as part of its partnership with Roche and Genentech. The success of this collaboration demonstrated the power of Recursion’s phenotypic screening platform, which uses high-throughput technologies to produce vast amounts of biological data. However, creating a neuromap would require more than just expanding on previous work - it required adapting the process to handle the unique challenges posed by neuronal biology.

Execution

To tackle this challenge, Recursion collaborated with Roche and Genentech to develop and refine a model using human-induced pluripotent stem cells (hiPSCs), which could be differentiated into neurons. This protocol enabled Recursion to produce large quantities of neurons, ultimately generating over 1 trillion hiPSC-derived neuronal cells. These neurons served as the foundation for the neuromap, a data-rich resource that Recursion, Roche and Genentech could use to gain deeper insights into the genetic underpinnings of neurodegenerative diseases. In addition to the joint development of the neuronal cell context, Recursion’s machine learning team played a pivotal role in developing algorithms capable of processing the massive amounts of data generated by the neuromap. The combination of scalable cell production and cutting-edge computational models allowed Recursion to generate the first whole-genome neuronal phenomap that can be utilized by the partnership to uncover new relationships between genes and the phenotypes associated with neurodegeneration.

Outcome

Our work led to the exercise of a $30 million option by Roche and Genentech in August 2024 with the neuromap offering an unbiased view of the genetic relationships related to neurodegenerative diseases and providing insights that could pave the way for development of novel therapies in neuroscience. Unlike traditional approaches that are often guided by pre-existing hypotheses, researchers in the collaboration can now explore new biological pathways and identify potential therapeutic targets that may not have been considered before. Together, Recursion, Roche and Genentech have identified a number of biological insights from this first neuroscience-focused phenomap, that could become novel targets of interest.

Case Study 2: Delivering the World’s First Microglia Map

Figure 34. Recursion launched a transformational collaboration with Roche and Genentech, delivering the first-of-its-kind microglia map in 2025.

Overview

In 2025, building on the success of the original neuromap, Recursion reached another milestone in its collaboration with Roche and Genentech by launching the world’s first microglia map. Microglia are the resident immune cells of the brain and play a central role in neuroinflammation and the progression of neurodegenerative diseases. By mapping the whole genome within these complex cells, the microglia map represents a completely new approach to explore the cellular mechanisms underlying neurodegenerative diseases, offering a new approach to explore novel targets and pathways.

41

Table of Contents

The Challenge

Despite decades of research, FDA approvals for neuroscience drugs are less than half of those for other therapeutic areas, and drugs targeting the CNS have some of the highest failure rates in medicine. Attrition rates are high, and traditional approaches can be biased to dominating hypotheses e.g. the "Amyloid Hypothesis" in Alzheimer’s disease. This is the idea that a build-up of the amyloid protein is a major contributor to the disease, and while this hypothesis has historically guided research and led to deep insights into the condition, the underlying biology is both extremely complex and still poorly understood.

To provide an unbiased approach to explore novel targets and pathways, the team studied microglia. However, microglia are very difficult to work with, proving difficult to grow and keep alive and stable outside of the body, in a laboratory setting. Because microglia are the resident immune cells of the brain, they can be highly sensitive to their environment, changing states from relatively stable to becoming more reactive and inflammatory, and they are also highly variable from batch to batch. To create a reliable map, Recursion had to overcome the technical hurdle of producing these sensitive immune cells at a massive, industrial scale while ensuring they remained stable enough to provide a clear biological signal for machine learning models.

Execution

Tackling this effort was a multi-year collaboration that required Recursion and the Roche and Genentech microglia team to develop new protocols for the manufacture of microglial cells. Starting with human-induced pluripotent stem cells (hiPSCs), the team developed a protocol that allowed the most phenotypically active cell Recursion has ever tried to map to be grown at massive scale. Over 100 billion microglial cells were grown in a standardized way and confirmed to be as stable as possible for the start of the mapping process. The collaborative team also worked together to determine the most interesting disease-like perturbations to the microglia from a neuroscience perspective, to generate a rich dataset containing some novel knockdowns and overexpressions not previously tried in other maps. This resulted in 100,000 single guide RNA being used spanning more than 17,000 genes, 46 million microglial cell images, and thousands of chemical compound perturbations. Recursion foundation models, powered by the supercomputer BioHive-2, extracted insights to generate the first-of-its-kind microglia map, allowing scientists to use AI to systematically explore how different genes and compounds may be implicated in a wide range of neurological diseases.

Outcome

The successful completion of the microglia map led to a $30 million milestone payment from Roche and Genentech in October 2025. The microglia map provides a holistic, unbiased approach to drug discovery for neurodegenerative diseases compared with the slow traditional approach, which has yielded very few new therapeutic targets. The map allows for the systematic, unbiased evaluation of thousands of gene targets at once, allowing AI to uncover novel biological connections that humans might miss. Overall, this offers a new approach to exploring novel targets and pathways, addressing a major challenge in neuroscience drug discovery. Following on from development of the map, it will be mined for novel biological insights, which will move forward to robust experimental validation from Recursion in partnership with Roche and Genentech. This could lead to program selection and development, and potential new therapeutic approaches in neurological diseases.

Our Platform

A Unified, AI-Native Platform for Drug Discovery & Development

Recursion is leading the evolution of how medicines are discovered and developed with the Recursion OS: a unified, AI-native intelligence platform designed to translate complex science into medicines that matter — faster, better, and at scale for the patients who are waiting.

Our approach combines proprietary experimental data, purpose-built computational models, and scaled compute infrastructure to support decision-making across the full lifecycle of drug discovery and development. Rather than optimizing isolated tools or workflows, we organize our platform around three tightly connected stages of value creation: novel biological discoveries, precision design, and next-generation clinical development. Together, these stages enable us to initiate programs with stronger biological grounding, design differentiated molecules more efficiently, and advance medicines into the clinic with improved patient relevance.

Across all three stages, Recursion integrates automated wet-lab experimentation with in-house computational analysis. Our laboratories generate large volumes of standardized, high-quality biological and chemical data, while our dry-lab capabilities apply machine learning, physics-based modeling, and statistical inference to extract actionable insights from those data. This tight integration allows experimental results to directly inform computational models, and model outputs to guide subsequent experiments, enabling faster iteration and more consistent decision-making across programs. Additionally, the integration of agentic and automated systems underpins our approach at Recursion and enables us to accelerate learning and decision-making throughout the R&D process.

42

Table of Contents

The Recursion OS

Figure 35. The Recursion OS. The Recursion OS is designed to use AI to advance and accelerate decision-making and insight across the entire R&D value chain with the end goal to make novel medicines that matter. The Recursion OS generates value by advancing a pipeline of differentiated investigational medicines, in addition to building pipelines for our partners.

By leveraging the Recursion OS to explore and advance our programs, we have shown leading indicators of improvement when compared to the traditional drug discovery process, particularly with respect to cost and time. We also use AI/ML tools to better understand which molecules to make and test and ultimately design better quality molecules that can solve complex problems – on average, the industry synthesizes about 2,500 molecules to candidate, compared to the approximately 330 we synthesize per program to advanced candidate.

Biology to Insight

The first stage of the Recursion OS focuses on translating complex biological signals into actionable insights that support program initiation. Recursion combines large-scale cellular phenomics, high-throughput transcriptomics, in silico binding affinity predictions, and real-world patient data to identify disease-relevant mechanisms, validate targets, and prioritize opportunities with stronger biological grounding and patient relevance. These capabilities allow us to systematically interrogate biology at scale and reduce biological uncertainty earlier in the discovery process.

A core differentiator of this approach is the integration of complementary data modalities. Patient-derived data are the most directly relevant to human disease but are often noisy, heterogeneous, and limited in scale. In contrast, cellular phenomics data can be generated reproducibly, at scale, with high completeness and consistency. By integrating these and other modalities through joint forward- and reverse-genetics approaches, Recursion can connect robust experimental signals with patient biology, enabling more confident identification and prioritization of translatable targets.

•Generated and aggregated >50 petabytes of high-quality, multimodal data

•Over 100 novel insights triaged into ~10 actionable and translatable targets for experimental validation within a matter of weeks

•~1.9B-parameter phenomics foundation model delivers ~25–30% gains in biological signal accuracy

•New transcriptional foundation model delivers a 70% improvement in operational efficiency

Deep Dive: Phenomics-based Discovery

Phenomics is Recursion’s large-scale, image-based cellular profiling capability that measures functional cellular responses to genetic and chemical perturbations and serves as a foundational input to biological discovery and program initiation. Using high-content microscopy, automated experiment design and execution, and purpose-built machine learning models, we generate rich, high-dimensional phenotypic data that capture cellular behavior across diverse biological contexts. Our platform is differentiated by its scale, precision, and breadth, operating both Cell Painting and live-cell brightfield microscopy across nearly 50 distinct cell types, including differentiated iPSC-derived neuronal and microglial cells. These experimental capabilities directly enhance our advanced computer vision and foundation models, including our Phenom-2 model series, by enabling our models to learn true biology as opposed to experimental design patterns. The impact of this capability is reflected in both our internal pipeline and strategic partnerships, including the development of two first-of-their-kind whole-genome neuronal and microglia phenotypic maps as part of our collaboration with Roche and Genentech. More broadly, phenomics has enabled Recursion to initiate and advance multiple internal and partnered programs by supporting unbiased discovery, rapid hypothesis triage, and identification of novel biological insights that may translate into new therapeutic opportunities.

43

Table of Contents

Deep Dive: Transcriptional Foundation Model (TxFM)

TxFM is Recursion’s self-supervised transcriptomics foundation model designed for representation learning of complex gene expression data. Built on a transformer-based architecture optimized for biological structure rather than natural language analogies, TxFM harmonizes diverse transcriptomic datasets—including bulk and single-cell RNA sequencing across multiple assays, cell types, and translational systems—into a unified embedding space.

Many transcriptomic models rely on large, heterogeneous public atlases that limit cross-experimental comparability and translational consistency. TxFM leverages Recursion’s proprietary data strategy and architecture decisions to improve cross-sample and cross-experiment relatability, enabling consistent integration of in vitro experiments, in vivo models, and patient-derived samples. This not only captures a more universal biological grammar but also allows us to outperform larger models trained on datasets up to 50x larger in size. By representing experimental perturbations and patient transcriptomes within the same high-dimensional space, we can begin to perform in silico perturbations on digital patient representations, serving as a practical translational bridge between laboratory biology and human disease.

TxFM has demonstrated state-of-the-art performance across multiple zero-shot benchmarks, outperforming existing foundation models and classical baselines. Within the Recursion OS, TxFM improves batch correction and multi-dataset integration, enhances signal recovery from low-read-depth transcriptomic data, and enables consistent mapping of gene–gene and gene–compound relationships. By improving data consistency and reducing the need for experimental re-runs, TxFM has delivered an approximate 70% improvement in operational efficiency for transcriptomics-driven workflows. Furthermore, the model’s learned gene-specific parameters recover known protein complexes and pathways without supervision, providing a non-perturbational gene-gene map for over 40,000 genes, including non-coding RNAs. These capabilities strengthen target discovery, accelerate hypothesis validation, and improve the biological and patient relevance of programs entering the pipeline.

Insight to Molecule

Once we have nominated a program, either through insights derived from our phenomics and multiomics platforms or through the careful selection and validation of a high-potential target, we transition from biological discovery to precision molecular design. Our precision design platform, anchored by Centaur Chemist, represents a transformative shift from traditional trial-and-error drug discovery to a fully integrated, AI-first industrialized process. By fusing massive, high-dimensional biological and chemical datasets with advanced generative AI and automated synthesis, the platform enables the rapid design, prioritization, and physical testing of novel small molecules. This modular end-to-end engine is designed to navigate trillions of biological and chemical relationships with unprecedented speed and efficiency, aiming to deliver higher-quality drug candidates to the clinic while significantly reducing development timelines and costs. Centaur Chemist serves as a critical component of our unified operating system, driving a continuous "design-make-test-learn" or DMTL cycle that refines its predictive capabilities with every successive iteration.

◦100 million+ molecules generated virtually using synthetically aware design in 2025

◦~90% of synthesized molecules are AI-generated, scored, and prioritized – all patentable

◦On average, only ~330 compounds are synthesized per program to achieve an advanced candidate in ~17 months, compared to industry average of over 2,500 compounds and 42 months, respectively,

◦To date, the platform has designed >10 development candidates that address a wide variety of previously unsolved biological or chemistry problems

Deep Dive: Centaur Chemist Methods and Models

At its core, Centaur Chemist is an AI-first learning system that automates the design, prioritization, and optimization of novel small molecules. It is not a single piece of software but an integrated platform within the Recursion OS that enables the development and deployment of a vast number of design tools that have been developed in-house, including the following proprietary models:

•Generative AI and Evolutionary Models: These create novel chemical structures based on target profiles

•Synthesis-Aware Methods: The platform employs an expansive synthesis-driven design toolkit that couples advanced billion-scale search algorithms (e.g. SALSA) and accurate chemo- and regioselectivity models with up-to-date vendor logistics to rigorously generate synthetically feasible compounds and expedite their reduction to practice

•Protein and Target Tractability predictions. These methods predict protein structures with and without the presence of ligands, enabling ligandability and druggability of targets.

44

Table of Contents

•Property Prediction and Scoring: Our industry leading advanced models, including our proprietary version of Boltz-2 that allows us to fine-tune the model on internal program data, predict potency (see figure below), selectivity, and ADMET (absorption, distribution, metabolism, excretion, and toxicity).

•Physics Methods: A toolkit of physics-based methods that apply molecular dynamics and quantum mechanics, which enables us to more accurately predict target-ligand interactions and properties, fully integrated with our generative design capabilities.

The platform not only incorporates our in-house proprietary methods and algorithms but makes it simple to deploy open-source and licensed software, ensuring that we are using state-of-the-art methods developed by the community as well.

Figure 36. Recursion potency models MolE and MolGPS outperform (lower is better on the Y-axis in figure) all other entrants in the 2025 blind challenge for potency predictions21

Deep Dive: DMTL (Design-Make-Test-Learn) and Automation

With these tools in place, our DMTL procedure begins by defining program objectives via a Target Product Profile (TPP), which is encoded using a multi-parameter optimization (MPO) function. This ensures that potency and affinity are balanced with other critical ADMET properties, such as clearance, solubility, stability, and permeability. Every generated molecule is scored by an integrated function called Merit, allowing us to monitor the quality of chemical matter across the program lifecycle in an unbiased, holistic manner. This allows us to focus our efforts from the start on high-quality molecules that meet the needs of our drug discovery programs, optimizing for both cost and operational efficiency.

Once the Design phase is complete, synthesis (Make) is triggered through the platform, which can be routed through to our in-house chemistry automation studio or dedicated CROs. Because our generative tools prioritize synthesizability from the outset, the platform utilizes property prediction models to guide design toward compounds that are both biologically optimized and amenable to efficient synthesis. To achieve this "synthesis awareness," we account for building-block availability and logistics across different CROs, ensuring the system suggests the most cost- and time-effective synthetic routes. A critical enabler of this physical execution is our in-house chemistry automation studio located in Milton Park, shown in the figure below, designed to support high-throughput DMTL cycles with minimal manual intervention (see Figure 40). The modular, automated lab is designed to accelerate DMTL cycles by utilizing automation flexibly across the full Make workflow, with over 1000 compounds having now been made and tested at Milton Park. The core mission of this DMTL procedure is to automate processes that reduce costs, optimize efficiency, and deliver high quality results to drive program success, not just making compounds that can easily be synthesized with automation. Through 2025 there was a 4-fold increase in reaction classes executed with automation, including 14 reactions that are not typically automated.

21 Polaris hub competition results. Source: https://polarishub.io/competitions/asap-discovery/antiviral-drug-discovery-2025#competiton-results 12 February 2026

45

Table of Contents

Figure 37. Recursion's automated chemistry wet lab, a modular system for chemical synthesis preparation, execution, analysis, work-up, and purification.

Following the Make phase, an assay cascade is initiated (Test). Depending on program needs, this includes ADMET, affinity, and phenomic responses via our proprietary -omics platform in either Salt Lake City or Milton Park facilities. Once an assay is completed, results are ingested back into the platform for analysis alongside all other relevant program data. This data automatically updates our ML models, ensuring the most accurate information informs the next cycle and closes the Learning loop. An example of this improvement is shown in Figure 38, which demonstrates that iterative learning has led to an approximately 50% improvement in a property prediction model. Our models improve over time, not by chance, but by learning.

Figure 38. LogD model improvement over time as DMTL cycles progress, showing an approximately 50% increase in accuracy over time.

By tracking all data centrally we can monitor Design, Make, and Test durations, as well as lag times between phases. Streamlining these workflows, both in-house and with CROs, allows us to identify and resolve bottlenecks in real time. This continuous monitoring has driven significant productivity gains: in 2025, our Design-to-Test times improved by 20% over 2024 benchmarks.

46

Table of Contents

To further support our learning system, we also employ our RADME-01 platform. This platform performs high-throughput ADME evaluations through intelligent software prioritization and automation. The system generated over 35,000 total data points across >15,000 novel compounds, including >13,500 human and >5,000 mouse microsomal stability results, alongside thousands of points for PAMPA, protein binding, and rat stability. RADME data is integrated into our proprietary datasets and automatically fed into our suite of ML models for comprehensive ADMET property prediction, which drives drug discovery during Nominations and Design. We employed Active Learning with RADME to generate thousands of datapoints, expanding model training sets and boosting accuracy across design programs. Additionally, new multi-task models optimally combine RADME and CRO assay data for related properties, delivering enhanced predictions to our design programs.

Molecule to Patient

The final stage of the Recursion OS focuses on translating molecules into clinical impact. Recursion utilizes large scale multi-modal datasets to rigorously design and execute clinical trials. As an example, by applying causal AI on human genomics, we select the indications and patient sub-populations most likely to benefit from our investigational therapies. Our AI-driven study planning algorithms applied to our operational data recommend site selection to enhance enrollment in minutes. Our clinical trials incorporate advanced statistical methods across several design elements from selecting optimal doses to forming external control arms to expanding the eligible patient population for our studies.

These capabilities form Recursion’s Clinical Development Technology (ClinTech) platform, an AI-enabled clinical development system that unifies molecule-to-patient decisions within an integrated workflow. It combines global site intelligence data, real-world patient data—including ~300 million real world lives and 1 million molecularly profiled lives—with causal inference, simulation, and agentic automation to continuously inform trial strategy and execution. By embedding patient relevance and operational feasibility earlier in development, ClinTech supports more disciplined trial design and execution at portfolio scale.

◦~1 million molecularly profiled lives between Tempus, Helix, and UK Biobank. Used across our clinical and preclinical portfolio for target validation and patient selection. Impact includes but not limited to:

▪Expansion into ovarian cancer for the CDK7 program

◦De-identified records covering ~300 million real-world lives, including electronic health records, diagnostics, and medical & pharmacy, leading to.

▪10-40% increase in eligible population

▪~1.5X improvement in enrollment rates

◦Global clinical trial site intelligence database covering a wide swath of historical clinical trials

▪Data driven country & site selection in hours vs. months

In practice, ClinTech shortens the cycle from protocol to site activation and enrollment by turning fragmented clinical operations data into actionable, real-time recommendations. For example, the platform can shorten the time to prioritize countries and sites from months (industry standard) to hours using multimodal site intelligence data. It can forecast enrollment trajectories as criteria evolve—enabling benefit-risk tradeoffs that can meaningfully expand the eligible population and improve enrollment performance. It can reinforce upstream portfolio decisions through genetically informed target validation and patient selection. Across the trial lifecycle, these capabilities translate into increased probability of success and timely execution of our clinical studies.

47

Table of Contents

Figure 39. AI-driven clinical development platform for trial design and execution

Deep Dive: Natural History Data in Familial Adenomatous Polyposis (FAP)

Recursion applied its ClinTech platform to generate high-quality natural history evidence to contextualize the single-arm efficacy of REC-4881 in the TUPELO study. Natural history data are increasingly important to regulators in rare diseases, yet for FAP there has historically been limited quantified evidence describing disease progression outside of clinical trials. To address this gap, Recursion leveraged real-world evidence analytics and AI-enabled data extraction to build a comprehensive view of the lived FAP patient experience across routine clinical practice and long-term registry data.

Using ClinTech’s real-world data capabilities, Recursion analyzed records from more than 1,000 US patients with FAP, including over 250,000 unstructured physician notes processed using a custom large language model–based workflow. This analysis enabled systematic characterization of disease burden, intervention frequency, and the progressive nature of polyp growth in everyday clinical care. Recursion extended this work through an academic collaboration with Amsterdam UMC, analyzing nearly 20 years of longitudinal follow-up data from approximately 200 patients enrolled in one of the largest and longest running FAP registries globally. Together, these datasets enabled a level of real-world disease characterization that is rarely available in rare disease development.

Across both real-world and registry datasets, the findings were consistent and clinically meaningful: untreated FAP is characterized by predictable, year-over-year progression of polyp burden in 87% of the trial-relevant patient populations, with a mean annualized increase of 60%. These insights provided a robust, data-driven benchmark for contextualizing therapeutic impact and directly informed the clinical development strategy for REC-4881, including support for a single-arm study design aligned with regulatory expectations. More broadly, this work illustrates how ClinTech augments clinical development with unbiased real-world insight, strengthens the translational path from molecule to patient, and enables more confident engagement with regulators in rare disease programs.

Processing and Data Storage Infrastructure

The need to understand pathways, targets, compounds, and mechanisms of action requires obtaining, synthesizing, or predicting large volumes of data. To store this data in an efficient and low-risk way, Recursion makes use of a combination of cloud storage, and on-premises storage. To process this data efficiently, we bring it close to where the compute will run – either in our HPC datacenter (BioHive) or to our cloud (partnering with Google Cloud). To make this more seamless for our scientists, we have invested in a hybrid storage and compute platform, which enables replication of data and locality of compute to allow us to use these resources as efficiently as possible.

48

Table of Contents

People and Culture

Essential to leading and defining TechBio is our team of close to 600 employees, comprising of life scientists, such as chemists and biologists (approximately 35% of employees) and computational and technical experts such as data scientists and software engineers (approximately 43% of employees). Together our team creates an environment where empirical data, statistical rigor and creative thinking is brought to bear on the problems we address. While we are united in a common mission, Decoding Biology to Radically Improve Lives, our strength lies in our differences: expertise, gender, race, disciplines, experience and perspectives. Deliberately building and cultivating this culture is critical to achieving our audacious goals.

In 2025, we made significant progress to deploy agentic and automated AI solutions across the workforce to enhance enterprise productivity and efficiency. In our technology organization, approximately 91% of employees are actively using AI coding tools to accelerate their work. As a result, 35% of our code is authored by AI, allowing our teams to focus on solving the problems and trusting AI to accelerate creating the solution. This saves an average of 4.3 hours of work per week, per employee.

Employee Recruitment, Development and Training

At Recursion, we believe a diversity of experiences, backgrounds, ideas and expertise will create high performing teams. We are intentional about the employee experience at Recursion, with a fit-for-purpose system that finds, grows and retains top talent to deliver our mission. We employ a targeted approach to identify, attract and hire diverse employees across highly technical scientific disciplines including biology, chemistry, data science, machine learning, engineering, robotics, clinical development and more. People stay at Recursion because of the opportunity to impact the world and grow in a place where they feel challenged, supported, and connected.

Employee Health and Safety

We have dedicated Standard Operating Procedures to manage occupational health and safety, safety training and injury, and illness and incident reporting. Every employee is responsible to ensure these procedures and policies are followed. We offer extensive training to ensure understanding and compliance. Compliance is mandatory for all laboratory employees per requirements of the Occupational Safety and Health Administration standard on Hazardous Chemicals in Laboratories. Our Chemical Hygiene Officer and Lab Manager oversees the day-to-day management of institutional chemical hygiene.

Read more about how we invest in and motivate our people to achieve our mission in Recursion’s latest Environmental, Social and Governance Report, available at our corporate website.

Facilities

Recursion’s global footprint is architected around two primary pillars: industrialized wet lab infrastructure that powers platform operations and offices that allow us to attract and retain top talent.

Industrialized Wet Lab & Platform Operations Hubs

Our two heavy-infrastructure sites in Salt Lake City, Utah and Milton Park, Oxfordshire are designed for the massive-scale data generation that drives our platform. These facilities house our robotics, automation, biology, and chemistry capabilities. Our Salt Lake City site has capacity to generate phenomics and transcriptomics data, and at Milton Park, the automated lab assembles the vision of autonomous DTML loops.

•Salt Lake City, Utah: We utilize 140,000 square feet of laboratory and office space in downtown Salt Lake City. This campus houses our high throughput screening labs, generating large volumes of standardized high quality biological data across phenomic and transcriptomic endpoints.

•Milton Park, Oxfordshire: This 20,151 square foot laboratory and office space serves as our primary European wet lab, focused on DMTL, quantitative pharmacology, complex bioassays, target validation, and automation engineering.

Strategic Talent and Office Hub

Our office locations in New York City, Montreal, and London are positioned in global centers for AI, scientific innovation, clinical development, and executive leadership.

•New York City: In January 2025, we opened a new 11,655 square foot office in New York City’s Hudson Yards neighborhood. This office serves as a key location for our executive leadership, strategy and clinical development teams,

49

Table of Contents

•Montreal: We have a 8,367 square foot site in Montréal that houses our semi-autonomous artificial intelligence research engine, Valence Labs

•London: Located in the knowledge quarter and heart of King’s Cross neighborhood, our 6,792 square foot office serves as a magnet for Europe’s top AI and computational talent.

Commercialization

We retain significant development and commercial rights to some of our drug candidates. If marketing approval is obtained, we may commercialize our drug candidates on our own, or potentially with a partner, in the US and other geographies. We currently have no sales, marketing, or commercial product distribution capabilities. Decisions to create this infrastructure and capability will be made following further advancement of our drug candidates and based on our assessment of our ability to build the necessary capabilities and infrastructure with competitive advantage. Clinical data, the size of the addressable patient population, the size of the commercial infrastructure, manufacturing needs and major trends as to how value is accrued in the industry may all influence or alter our commercialization plans.

Manufacturing

We currently utilize contract development and manufacturing organizations to produce drug substance and investigational drug product in support of the assets within our pipeline. To date, we have obtained drug substance and drug product for our drug candidates from third party contract manufacturers.

Strategic Partnership and Collaboration Agreements

To achieve our mission, we may partner with leading biotechnology companies, pharmaceutical companies and academic research institutions to access datasets, molecules, or other intellectual property.

Roche and Genentech Collaboration and License Agreement