The $15 Billion Race to Become Your Doctor’s AI Assistant

Before a prescription is written or a specialist is called, a doctor often needs an answer. Which evidence applies to this patient? Has a guideline changed? Does a complicating condition alter the next step?

The company that becomes the first place a physician turns could occupy one of the most valuable positions in healthcare.

That is the investment thesis coming into focus around OpenEvidence. On September 24, Business Insider reported that the medical AI company had raised $250 million from investors including hospital systems and Andreessen Horowitz, at a $15 billion valuation. The report cited people familiar with the financing.

Our reading of that price is straightforward: investors are betting that a trusted answer engine can become a habit, and that a habit embedded in medical practice can support a substantial business.

Whether those expectations are justified depends on three things that deserve to be examined together: what doctors get, who pays, and how much confidence patients should place in the result.

Why a medical answer engine has an opening

OpenEvidence gives clinicians a conversational way to search and synthesize medical knowledge, with references they can inspect. Its publisher relationships are central to the proposition.

The JAMA Network’s June 2025 agreement brought full-text and multimedia material from its 13 journals into the platform’s answers. Springer Nature announced a further agreement in August 2026, adding its research content and emphasizing attribution and links to original sources. OpenEvidence also has a content agreement with NEJM Group.

That combination addresses a practical problem. A clinician can phrase a question around the circumstances of a case and receive a starting point for evaluating the evidence. The potential benefit is less time searching and more time considering what the evidence means for the person in front of them.

The appeal also helps explain why specialized AI businesses can survive alongside much larger model developers. Medical content rights, a usable clinical interface, and professional trust all require work beyond making a language model available.

But those advantages need to hold up in daily use. A cited answer still has to select the right research, represent it faithfully, and avoid expressing more certainty than the evidence warrants. Publisher partnerships establish access to material; the quality of each synthesis remains a separate test.

There are signs of substantial demand. Fierce Healthcare reported on September 28 that OpenEvidence had 1.12 million verified, medically licensed U.S. clinician users, including physicians, nurses, nurse practitioners, and physician assistants. That is a broader population than doctors alone, and the reported total should not be mistaken for daily active users.

For investors, the more revealing measure will be how consistently clinicians return—and whether the product helps them do their work better.

Free for doctors creates a different kind of customer

OpenEvidence’s advertising model makes the commercial story particularly interesting. Research firm Sacra describes a business funded by pharmaceutical and medical-device advertising. Removing a subscription charge lowers the financial barrier for clinicians. It also means the user and the paying customer can have different priorities.

In a June JAMA Viewpoint, Nina Singh and Michelle Mello examined advertising in clinical AI. They noted that OpenEvidence keeps answer generation and advertisement selection in separate systems and does not permit advertising to directly influence answer content.

That distinction matters. There is no basis here to claim advertisers purchase medical recommendations. The question is how a platform demonstrates that separation as its advertising business grows. Clear labels, meaningful disclosures, and outside scrutiny would help clinicians evaluate it.

For an advertiser, proximity to a clinical question could be commercially attractive. For a physician, the same proximity raises the standard for transparency. The platform has to satisfy advertisers without undermining the confidence that brought clinicians there.

The attraction of a free product is therefore inseparable from the quality of its governance. Trust is part of what the business is selling, even when the clinician pays nothing.

The competition already has a place in the doctor’s day

OpenEvidence is competing against companies with different starting advantages. Some own established medical reference products. Others already support clinical communications or documentation.

Company or product	Starting position	The competitive question
OpenEvidence	Medical search and evidence synthesis supported by publisher agreements	Can habitual use become a durable business while preserving clinical trust?
Doximity	A clinician network with communications, workflow, and clinical AI tools	Can it make medical answers part of an existing daily routine?
Wolters Kluwer’s UpToDate Expert AI	An established reference product with expert-authored medical content	Will clinicians and institutions prefer AI delivered through a familiar source?
Microsoft Dragon Copilot	An assistant embedded in clinical documentation and broader work processes	Will integration make a separate research application less necessary?

Product descriptions are based on the companies’ materials; competitive questions are our analysis. Sources: OpenEvidence/JAMA, Doximity, Wolters Kluwer, Microsoft. These products overlap without performing identical jobs.

UpToDate Expert AI provides conversational answers grounded in Wolters Kluwer’s medical reference content. Dragon Copilot is expanding beyond documentation to bring clinical knowledge, patient information, and partner applications into existing workflows.

OpenEvidence is expanding its own reach, too. Its developer’s app listing includes visit documentation and secure patient calling alongside medical answers. The competitors are increasingly approaching the same work from different directions.

The competitive implication is that accuracy alone may not determine which product gets used. A useful answer available inside the system a clinician already has open can have an advantage over an equally useful answer in another application. Conversely, a specialized tool can earn a separate place if its answers are sufficiently valuable.

This leaves room for several successful businesses. It also creates a pricing problem: if competing products make evidence-based answers readily available, companies must show why their particular combination of content, convenience, and reliability deserves lasting loyalty.

The valuation is racing ahead of the visible economics

OpenEvidence’s ascent has been unusually rapid. Its February 2025 financing announcement put its valuation at $1 billion. By January 2026, a company-confirmed $250 million round valued it at $12 billion.

Private financing prices express expectations. They do not, by themselves, establish the revenue or profits a business will eventually produce.

Sacra estimates that OpenEvidence reached $300 million in annualized revenue in July 2026, up from $150 million at the end of 2025. Those are third-party estimates of an annual revenue pace, not audited revenue earned over a full year. They suggest a business of meaningful scale, while leaving outsiders with much less financial visibility than a public-company filing provides.

OpenEvidence is also investing in expensive capabilities. In January, founder Daniel Nadler told Fierce Healthcare that much of its new funding would support model development and computing, alongside further content licensing. More use can strengthen the product while creating more costs to serve its users.

Doximity provides a useful public-market comparison because its financial statements reveal both the opportunity and the expense. For the fiscal year ended March 31, 2026, it reported revenue of $644.9 million and net income of $196.1 million. Those are company-wide results, covering its broader business rather than AI alone.

Its subsequent quarterly filing offers a sharper caution. Revenue in the June quarter rose 7% to $156.6 million, while gross margin fell from 89% to 85%. Doximity attributed the margin decline primarily to costs supporting its AI initiatives. The filing identifies pharmaceutical manufacturers and health systems as its principal revenue-generating customers.

That is a concrete illustration of the investment challenge: a company can have an established commercial audience and still spend more to defend and expand its position through AI.

For OpenEvidence, the key financial questions are how much revenue recurring use generates, what it costs to deliver dependable answers, and how much additional spending is needed to stay competitive. A large user count cannot settle those questions.

A better-rated answer is an encouraging start

There is evidence that specialization can improve the experience of medical AI, but the scope of that evidence matters.

A June 2026 preprint used 149 practicing physicians to compare OpenEvidence with three general-purpose models. Its principal question set contained 620 queries derived from actual clinical use. In blinded comparisons, physicians rated OpenEvidence highest across the five dimensions assessed, including accuracy and clinical usefulness.

The study disclosed that OpenEvidence helped design and implement data collection and paid survey respondents. The authors reported no affiliation with the company, and the statistical analysis was conducted with model identities concealed.

Those are useful findings with defined limits. The study evaluated answers and physician preferences; it did not measure whether patients recovered faster, suffered fewer complications, or received less unnecessary treatment. It also did not compare every competing clinical product.

For purchasers, the next standard should be evidence from actual deployment: whether clinicians reach appropriate decisions more efficiently, whether errors are detected, and whether benefits persist across specialties and patient populations.

Investors should want that evidence too. A product that earns trust through measurable clinical usefulness has a stronger foundation for retention than one whose appeal depends on novelty.

Expansion abroad raises the stakes

The potential benefit becomes especially clear where access to specialist knowledge is limited. Reuters reported on September 22 that OpenEvidence and Anthropic were bringing a free, regionally adapted version of the platform to clinicians in approximately 100 countries. Financial terms were undisclosed.

Fierce Healthcare’s follow-up described support from Anthropic in computing credits, engineering, and funding, as well as a separate collaboration with Penn Medicine through the Botswana-UPenn Partnership.

Local adaptation will be essential to making such access useful. Advice has to account for which tests, treatments, and clinical resources are available. Broad distribution creates an opportunity to improve access to knowledge; local evaluation determines whether that opportunity translates into better care.

For the business, global reach and near-term revenue should be assessed separately. A free access initiative can deliver public value without immediately supporting the same economics as a commercial market.

What would make the bet hold up?

The strongest case for doctors’ AI assistants rests on a recurring need: clinicians must make decisions with limited time and an expanding body of evidence. Software that reliably reduces that burden has a clear purpose.

The investment case adds further demands. Clinicians must keep returning after the initial excitement. Revenue must grow enough to support computing, content, evaluation, and product development. The business must retain its place as competitors improve. And its commercial incentives must remain compatible with the professional trust on which adoption depends.

Public investors can examine parts of this competition through companies such as Doximity, Wolters Kluwer, and Microsoft, while recognizing that each has a broader business than clinical AI. OpenEvidence itself remains privately held. Exposure to the theme and the attractiveness of a particular investment are separate judgments.

The most consequential outcome may be a change in how medical knowledge reaches the examination room. An assistant that becomes a physician’s habitual starting point gains substantial influence over which evidence is considered first.

That position could justify an important business. Earning it requires answers doctors can trust. Keeping it requires a business model that gives them reason to continue doing so.

---

Reporting current through September 28, 2026. Financing reported by Business Insider is identified as reported; product, usage, and financial claims are attributed to their sources. Research findings are distinguished from patient outcomes. This article is editorial analysis based on published reporting, company disclosures, and research, rather than an independent clinical test of the products.