Chapter 6

The AI Contest

Doximity is spending its operating margin to turn a physician-engagement lead into a new pharma AI-search market. Research and development rose 40% to $130.7 million and operating margin fell to 33% from 40% in fiscal 2026 [1], ahead of AI revenue management calls minimal this year [2]. The AI-native rival it must beat, OpenEvidence, is litigating against it over data access [3], and the engagement metrics that evidence the bet are the ones the SEC is pressing Doximity to define [4].

Whether this bet re-expands the pool the growth case needs, or the contest for the same clinical AI audience simply redistributes it, is the swing factor behind the report's central question. This chapter weighs what is spent, what is shown, who is fighting, and how much of it can yet be checked.

The investment year

Management labeled fiscal 2026 its "AI investment year," and the label is literal in the accounts [5]. Research and development spending rose to $130.7 million from $93.0 million a year earlier and $82.0 million the year before — a 40% jump that lifted research and development to 20.3% of revenue from 16.3% [6]. The 380-person engineering team is, in the CEO's words, "all in to win" the race to build the best clinical Scribe and Search for doctors [7].

The spending shows up as a margin the company chose to give back. Operating income fell to $214.9 million from $227.8 million and operating margin dropped to 33.3% from 39.9%, even as revenue grew 13% [8]. Gross margin slipped to a full-year non-GAAP 91% from 92%, and to 89% in the fourth quarter, on a steep ramp in AI compute that is "outgrowing overall workflow engagement" [9].

Source: derived from reported financials, FY2022–FY2026 10-Ks; research and development and operating income per the Consolidated Statements of Operations [10].

The trade continues into the current year by design. Fiscal 2027 guidance carries adjusted EBITDA margin down to 49% from 55%, and stock-based compensation up to the low 20s as a percent of revenue before trending back down from fiscal 2028, largely the Pathway acquisition and performance grants for the growing AI team [11]. Against that spend, the return is deferred: management has forecast "minimal AI revenue contribution this fiscal year," describing a "nascent and regulated market" in which it has closed only its "first few AI search deals" [12]. The prize it is spending toward is a claim, not yet a line item — AI Search alone as "a multibillion-dollar new TAM on top of the existing pharma marketing budgets we serve today," unlocked by the $63 million paid for Pathway AI [13].

What the engagement data shows

The evidence Doximity offers that the bet is working is a set of usage metrics. Benchmark workflow engagement reached over 800,000 unique quarterly active prescribers in the fourth quarter, up roughly 30% year on year — a marked acceleration from the high-single-digit growth of a year earlier — and nearly half of those prescribers used the company's AI tools in the quarter [14]. In the nine months after the Pathway deal, AI Search and Scribe active users tripled, and last month those users averaged 31 queries each, nearly double January's rate [15].

Source: Q4 FY2026 earnings call, May 13 2026 [16]; health-system count corroborated in the FY2026 10-K [17].

The enterprise adoption is the more durable of the signals, because it clears a bar competitors cannot buy their way past quickly. As of the fourth quarter, 140 health systems had purchased the Clinical AI Suite — including 7 of the top 20 hospitals — and over 250,000 prescribers had access to it in a single hospital-approved, HIPAA-compliant workflow [18]. Each of those adoptions passed a hospital's privacy and AI-governance committees, an approval the 10-K frames as "enterprise-grade trust" that takes years to earn [19]. On the monetization side, the first proof points are thin but pointed: the AI Search product launched at a pharma client summit attended by 40 marketing leaders from the largest drug makers, and the company reported closing its first deals with top-20 pharma manufacturers [20].

One benchmark the company cites points straight at the contest. In a side-by-side clinical-search evaluation, 4,700 physician residents preferred Doximity's AI answers over those of its "nearest competitor" by two to one [21]. The competitor goes unnamed, but the identity is not much in doubt.

The rival in court

The rivalry with OpenEvidence — the AI-native clinical-answer tool that reaches physicians and sells pharma advertising the same way Doximity does, set out in Moat and Rivals — is not an abstract market share contest. The two are in active litigation. On June 20, 2025, OpenEvidence sued Doximity, its Chief Technology Officer, and its Director of AI Products in the U.S. District Court for the District of Massachusetts [22].

The subject of the fight is the raw material of clinical AI. In an amended complaint filed in October 2025, OpenEvidence added Doximity's Pathway Medical subsidiary as a defendant and alleged the defendants "gained unauthorized access to its AI medical information platform," asserting claims under the Computer Fraud and Abuse Act, breach of contract, unjust enrichment, and trespass to chattels, and seeking a permanent injunction, actual damages, and disgorgement of the defendants' profits [23]. Doximity has counterclaimed, asserting false advertising under the Lanham Act, Massachusetts consumer-protection law, and common-law defamation [24].

The case is proceeding, not settling. On January 22, 2026, the court dismissed OpenEvidence's trespass-to-chattels claim and two of Doximity's counterclaims, but denied the motions to dismiss "as to all other claims, including all counterclaims" [25]. The company states it "is currently unable to predict the outcome" or estimate any loss [26]. For a reader, the injunction request is the tail risk worth pricing: a bar on Pathway or Ask accessing a rival's data would land directly on the product the "investment year" is funding, and the litigation is a reminder that the physician-query data underpinning clinical AI is contested territory, not settled ground.

Doximity is also staffing the contest from its rival's ranks. Alongside a new CFO drawn from LinkedIn, Atlassian, and DocuSign, it hired Dr. Steve Zatz — who spent 20 years at WebMD Medscape, the last seven as its President and CEO — as President [27]. Recruiting the former chief of the incumbent physician-media franchise to run the company as it pivots to an AI platform is a costly signal that the pivot is the priority, not a side project.

How much of this is verifiable

The traction case rests almost entirely on engagement metrics — active prescribers, query counts, user-growth multiples — because the AI revenue that would settle the argument is, by the company's own guidance, minimal this year and is not disclosed as a separate line [28]. That places unusual weight on metrics that are, at present, defined by management and disclosed selectively on calls. Two facts on the record argue for weighting them with a caveat rather than at face value.

First, the SEC is actively pressing on exactly these disclosures. In a comment letter dated February 23, 2026, the Division of Corporation Finance asked Doximity to define what counts as "use" of its clinical workflow tools, to define and quantify the active-user metrics referenced on earnings calls but absent from its management discussion and analysis, to clarify whether its "over 80% of U.S. physicians" claim reflects verified active users or a broader base, and to disaggregate revenue by module — Newsfeed, Workflow, and Peer — noting the information "appears" to be "readily available" and used internally to run the business [29], [30].

Source: SEC Division of Corporation Finance comment letter to Doximity, February 23 2026 [31], [32].

Second, the metric family at issue has already been litigated. In December 2025 Doximity agreed to settle a securities class action — covering purchasers between June 2021 and August 2023 and centered on "disclosures regarding user count and engagement rates" — for $31 million, an amount fully covered by insurance and carrying no admission of liability, with a final approval hearing set for June 10, 2026 [33]. Six related derivative suits remain outstanding [34]. None of this establishes that today's AI engagement figures are wrong — the settlement was insurance-funded and admitted nothing. It does mean the AI story is being told in precisely the currency — user counts and engagement — that a class action has already contested and that the SEC now wants pinned to a definition.

The read

The AI push is real, self-funded, and a credible option on re-expanding a market that has otherwise stopped growing (Market Trajectory). The company is spending a visible margin — 40% more research and development, six points of operating margin — from a net-cash balance sheet it does not need to protect, and the enterprise adoption of 140 health systems is a genuine, hard-to-replicate asset. The evidence for it, though, is one-sided in a specific way: strong on usage, silent on revenue, and reliant on metrics under simultaneous review by a regulator and a court. The bull and bear split cleanly here — the same 800,000 prescribers are either the leading edge of a new pharma-search franchise or an audience a litigious, comparably-sized rival can rent to the same drug makers.

Three developments would move the read, and each is checkable. AI Search revenue actually appearing — management guides a back-half fiscal 2027 ramp — and, better still, becoming visible in a disaggregated revenue disclosure, would convert engagement into proof. The SEC comments resolving with defined, quantified user metrics in the management discussion and analysis would remove the disclosure caveat. And the OpenEvidence suit resolving without an injunction touching Pathway or Ask would retire the litigation tail. Until those land, the AI contest is best held as an option the price does not pay for (Margin of Safety), not as a growth engine already earning its keep.