OpenEvidence Hits $15 Billion and Starts Making Cancer Drugs
The physician AI search engine raised $250 million at a $15 billion valuation and said it will put its first oncology drug into clinical trials before the end of the year.
OpenEvidence built a business by answering doctors' questions. Now it wants to answer one itself: what the next cancer drug should be. On September 24, Business Insider reported that the Miami company had raised $250 million at a $15 billion valuation, up from $12 billion in January. The same week it disclosed plans to develop its own oncology therapies, starting with rare cancers.
The round was led by Andreessen Horowitz, with hospital systems participating, according to Business Insider. Axios and Dealroom also named Byers Capital as a lead. OpenEvidence has now raised more than $1 billion in the past year from a roster that includes Thrive Capital, DST, GV, Kleiner Perkins, Sequoia and Nvidia.
A lower number than it floated
The $15 billion mark is a 25% step up in nine months. It is also below where the company was aiming. In July, The Information reported that OpenEvidence was weighing a $200 million raise at a $20 billion valuation, as Digital Health Wire noted in its coverage of the new round. The round that closed was larger in dollars and smaller in price.
That gap is not a distress signal. It is a sign that investors are pricing OpenEvidence as an advertising-driven search business rather than a frontier lab. Research firm Sacra estimates the company reached roughly $300 million in annualized revenue by July 2026, up from $150 million at the end of 2025 and $7.9 million at the end of 2024. By Sacra's figures, most of that comes from pharmaceutical and medical device advertising shown to verified clinicians, who use the product free. At $15 billion, the company trades near 50 times that estimate.
AI Weekly, summarizing the Business Insider report, said the founders are also fielding acquisition interest, which would make a sale a possible alternative to an eventual IPO.
Usage is the moat
The engagement numbers explain why investors keep paying. OpenEvidence was consulted 42 million times by US clinicians in August alone, according to Reuters. Coverage of the round puts its reach at roughly 40% of US physicians. Founded in 2022 by Daniel Nadler and Zachary Ziegler, it reached that scale without selling through hospital procurement: doctors sign up individually, verify their credentials and start asking questions.
September has been a run of announcements built on that base. On September 3, the company released a family of four medical models. Three of them, Osler, Sackett and Snow, are free to verified clinicians and differ mainly in how long they reason before answering. The fourth, Darwin, scored 100% on the MedQA benchmark, which OpenEvidence says is a first, along with 72.8% on MedXpertQA and 87.2% on NOHARM. Darwin is available only by application, and the company cited dual-use risk in areas such as virology, immunology and genetics as the reason.
On September 16, Memorial Sloan Kettering said it would integrate OpenEvidence into its Epic electronic health record workflow, while OncoKB, MSK's precision oncology knowledge base, would be built into OpenEvidence for clinicians outside MSK. On September 22, Anthropic and OpenEvidence announced a free version of the tool for clinicians in about 100 low- and middle-income countries, including Uganda, Haiti and Mongolia, with Anthropic's models providing the back end. "Access to medical knowledge shouldn't depend on geography," Nadler told Reuters.
From search to therapeutics
The drug program is the real departure. Per Business Insider's reporting, relayed by Refresh Miami and Digital Health Wire, OpenEvidence expects its first drug to enter clinical trials before year end and plans three to five more candidates next year, focused first on rare cancers. The stated logic is that rare-disease trials are ones large pharmaceutical companies often do not run, and that a platform used by a large share of American physicians could help find eligible patients.
The MSK deal fits that direction. OncoKB catalogs which genetic alterations drive which cancers and which therapies target them, exactly the evidence a rare-oncology developer needs. A partnership that looks like a clinical search upgrade is also a data asset for a drug pipeline.
The company has not disclosed which molecules it is developing, whether they are licensed or discovered in-house, or how the program will be funded and staffed.
The tension in the model
Here is our read on the part that needs watching. OpenEvidence's revenue, by Sacra's estimates, comes largely from pharmaceutical advertisers. Its trust with doctors comes from being a neutral reader of the evidence. A company that sells ads to drug makers, answers prescribing questions and now develops its own cancer drugs will have to show that those three roles do not bleed into each other. Rare cancers, where few competing therapies exist, soften that conflict at first. They do not remove it.
Competitors are close. Doximity sells its own AI tools to the same physicians, and incumbent reference publishers Wolters Kluwer and Elsevier are defending their clinical products. Clinical search can be copied. A drug that works cannot.
Why it matters
OpenEvidence is testing whether distribution to doctors is worth more than the search product that earned it. If a free clinical tool used by a large share of US physicians can also recruit patients, surface targets and move therapies into trials, a vertical AI company becomes a biotech with a customer base no startup drug developer has. If the drug bets stall, the company still has one of the most used AI products in medicine. Investors paid $15 billion for both outcomes. The first trial start, promised before year end, is the date that shows which one they are closer to.