Doximity
Two-sided, advertising-funded, subscription-billed. The clinician side is free: verified US physicians, NPs, PAs, CRNAs, pharmacists and medical students get the profile, newsfeed, Dialer telehealth, AMiON on-call scheduling, digital fax, Scribe and Ask at no charge. The paying side is enterprise subscriptions across three lines the company names in its 10-K — Marketing Solutions (pharmaceutical manufacturers and health systems buying access to clinician attention), Hiring Solutions (health systems and recruiting firms, including the Curative Talent staffing business) and Workflow Solutions. Contracts are billed in advance and recognised ratably; a material share is transacted indirectly through pharma marketing agencies. The clinical AI is therefore a retention and engagement asset priced at zero, not a product line with its own price — which is precisely why the board holds the name.
The thesis on this name
State of AI for Healthcare
The only listed name whose PAYER matches the corrected thesis: Doximity is paid out of pharma's commercial budget — not a health system's IT opex line, and not a CMS fee schedule. It gives its AI Scribe away free to every verified US clinician and still earned FY2026 revenue of $644.9M (+13%) at an 89.1% GAAP gross margin, $196.1M of net income (30.4% net margin) and $317.5M of free cash flow (+19%). The AI is a retention feature funded by an advertising business, not a product it has to price against Epic — which is why a free tier here is the correct price rather than charity, and why >800,000 active prescribers on its workflow tools (nearly half using its clinical AI, a count that bundles AI search and Scribe, not Scribe alone) is a competitive fact rather than an unmonetised one. It is also the board's cleanest refutation of the clearance-stock leg: a name search of all 1,524 rows of FDA's AI-Enabled Medical Device List returns ZERO Doximity authorisations, against GE HealthCare's 130 (8.5% of the list) and a 13.9% Q2 2026 EBIT margin in GE's Advanced Imaging Solutions segment. At $20.91 (31 Jul 2026 close) the stock sits ~73% below its $76.51 52-week high, on a $3.76B market cap. It trades at 21.3x TTM diluted EPS, and at 14.6x forward on a third-party consensus estimate. The de-rate is a growth guide, not a margin break.
State of AI for Healthcare
The board's control case: 89.1% GAAP gross margin, $196.1M FY2026 net income and zero entries on FDA's AI-enabled device list, earned by giving the AI away free and billing pharma's commercial budget — de-rated ~73% from its 52-week high on a growth guide, not on a margin break.
State of AI for Healthcare
The only listed name whose payer is pharma's commercial budget and whose clinical AI is priced at zero. FY2026 (year ended 31 Mar 2026): revenue $644.86M +13%, GAAP gross profit $574.54M = 89.09% gross margin, GAAP operating income $214.92M, net income $196.05M, free cash flow $317.5M (+19% year over year) — earned with zero entries on FDA's AI-enabled device list. ~$20.91 (31 Jul 2026 close), $3.76B market cap, 14.59x forward earnings on a third-party consensus estimate, against a 52-week range of $17.15–$76.51: de-rated ~73% from the high on the FY2027 revenue guide of $664–676M, not on a margin break. It is also the instrument that expresses the ambient finding directly. Doximity gives Scribe away — more than 800,000 active prescribers on its workflow tools in Q4 FY2026, with nearly half using its clinical AI (a count that bundles AI search with Scribe rather than Scribe alone) — while the one controlled measurement of what a buyer gets back from ambient documentation is $167.37 per adopting clinician per month of marginal E/M revenue, 95% CI $86.52–$248.21, which the authors describe as a conservative lower bound rather than a valuation of total benefit. Against a measurement that wide, a $0 tier is not charity; it is the correct price for a vendor whose margin comes from a different payer. Largest single position and the board's single-name cap.
Earnings, margins, COGS & capex
Doximity is the cleanest existence proof of the QAI board's central premise — that the identity of the payer, not the quality of the technology, decides whether software margin exists in healthcare. It holds none of the three assets the sector treats as decisive. Zero FDA AI authorisations, verified across all 1,524 rows of the FDA AI-Enabled Medical Device List (canon REG-FDA-019, list vintage 2026-06-16). No CPT code. No clinical-outcomes claim. It gives its ambient scribe away at $0 to every verified US clinician while roughly sixty ambient vendors (a March 2025 count) compete to charge $1,188–$5,000 a clinician a year for the same function. And it earns the best margin structure on the board anyway — 89.1% GAAP gross margin, 33.3% GAAP operating margin, 30.4% net margin, 55.5% adjusted EBITDA margin and a 49.2% free-cash-flow margin — because the invoice goes to a pharmaceutical manufacturer's commercial budget, which is large, inelastic and already aimed at exactly these clinicians. The bear case is not that this is a bad business. It is that the business has stopped growing and has started paying more for its own AI: revenue growth 20% → 13% → a guided 3–5%; net revenue retention 119% → 109%; cost of revenue +25.9% against +13.1% revenue; Q4 FY2026 GAAP gross margin 86.7% against 89.5%; stock-based compensation at 18.9% of revenue; and an FY2027 adjusted-EBITDA guide of $323–335M that is BELOW FY2026's $357.8M actual. A 55.5% EBITDA margin guided down to about 49% at the midpoint is the company itself telling you the free AI has a cost.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~11¢ is cost of goods and ~56¢ operating expense, leaving ~33¢ of operating profit (~30¢ net).
Revenue trend
Margins
down 110bp on the fiscal year and down 280bp in Q4 alone (86.7% vs 89.5%) — the compute line arriving in cost of revenue
a fiscal-YEAR move, company-attributed to increased AI compute costs (canon CF-DOCS-08 — never quote it unlabelled as a quarterly move)
down — operating income fell in absolute dollars from $227.8M on 13% more revenue, as opex grew 25.4%
up on the year, guided DOWN to roughly 49% at the FY2027 midpoint ($323–335M on $664–676M) — an implied 6–10% decline in absolute adjusted EBITDA
down 12.2% in dollars (canon CF-DOCS-09), driven by opex growth and a $53.95M tax provision against $40.39M
improving — the one headline metric that accelerated in FY2026 (canon CF-DOCS-11)
COGS structure
Cost of revenue was $70,326K in FY2026, 10.9% of revenue, up 25.9% against 13.1% revenue growth (FY2025 $55,874K, +10.3%; FY2024 $50,669K). The 10-K enumerates the line as cloud hosting, customer-success personnel, third-party platform access, IT and software costs 'including generative AI platform usage and inference costs', amortisation of acquired intangibles and internal-use software, editorial and content expense, and allocated overhead. Two structural points follow. First, Doximity does not train frontier models — it buys inference from third-party platforms, so its AI COGS is a pass-through price it does not set, and it falls or rises with someone else's list. Second, the cost is driven by USAGE, and usage is the thing the company is actively trying to grow: management said prompts per user nearly doubled between January and April 2026. A free product whose unit cost scales with engagement and whose revenue does not is the exact shape of the board's falsifier. Against that, stock-based compensation inside cost of revenue was only $11,680K, so 88% of the FY2026 gross-margin base is cash cost, and the reported 89.1% is close to a cash gross margin.
Capex
Immaterial and shrinking as a share of the model. FY2026 investing outflows attributable to capital formation were $8,901K of internal-use software development and $62K of purchased intangibles; the cash-flow statement shows no separate property-and-equipment purchase line. The only sizeable investing item was a $26,528K acquisition (which lifted goodwill from $67.9M to $85.0M and intangibles from $23.1M to $35.3M). Everything else in investing was securities portfolio management. The corollary is that Doximity's AI capability is rented rather than owned: there is no GPU fleet, no proprietary training run capitalised on the balance sheet, and therefore no depreciating asset that a competitor has to match — but equally no owned cost curve to ride down.
Latest earnings
The FY2026 year was fine and the FY2027 guide was not. Revenue of $644.9M (+13%) and record free cash flow of $317.5M (+19%) landed with an FY2027 revenue guide of $664–676M, about 3–5% growth, against a Street number reported around $697M; the shares fell approximately 24% on 14 May 2026. Both the consensus figure and the price reaction are SECONDARY evidence, not issuer disclosures (canon CF-DOCS-13). The under-discussed half of the guide is the adjusted-EBITDA line: $323–335M against $357.8M delivered, an implied 6–10% decline in absolute profit and roughly 640bp of margin compression at the midpoint.
FY2027 (year ending 31 March 2027): revenue $664–676M, adjusted EBITDA $323–335M. Q1 FY2027 (quarter ended 30 June 2026): revenue $151–152M, adjusted EBITDA $68.5–69.5M, a roughly 45–46% adjusted-EBITDA margin. Issued with the FY2026 results, whose release carries a 13 May 2026 dateline; the roughly 24% share fall came in the 14 May 2026 session. No revision was published between then and the 6 August 2026 print.
- FY2026 revenue
- $644.9M, +13% YoY (canon CF-DOCS-06; primary)
- FY2026 GAAP gross margin
- 89.1% — gross profit $574,537K on $644.9M of revenue, fiscal-year GAAP basis (canon CF-DOCS-07)
- FY2026 non-GAAP gross margin
- 91.1%, from 92.1% in FY2025 — a fiscal-YEAR move, company-attributed to increased AI compute costs (canon CF-DOCS-08)
- Q4 FY2026 GAAP gross margin
- 86.7%, from 89.5% a year earlier (Q4/FY2026 release reconciliation) — 170bp above the board's 85% falsifier line
- FY2026 GAAP net income
- $196.051M, −12.2% YoY; 30.4% net margin (canon CF-DOCS-09)
- FY2026 free cash flow
- $317.5M, +19% YoY — NOT to be confused with operating cash flow of $326.5M, also +19% (canon CF-DOCS-11 / CF-DOCS-12)
- FY2026 adjusted EBITDA
- $357.8M, 55.5% margin — guided to $323–335M for FY2027, an absolute decline
- Net revenue retention
- 109% (FY2026), from 119% (FY2025) and 114% (FY2024) — 10-K key-metrics table
- Customers ≥ $500k TTM subscription revenue
- 125 (FY2026), from 118 and 100; approximately 83% of revenue (10-K)
- Single-customer concentration
- One customer was 11% of FY2026 revenue — the first year any customer crossed 10% (10-K concentration note; none did in FY2025 or FY2024)
- Active prescribers on workflow tools
- >800,000 in Q4 FY2026, 'nearly half' using clinical AI (~400,000) — a count that BUNDLES AI search with Scribe (canon CM-MKT-05)
- FDA AI authorisations
- ZERO — a verified zero across all 1,524 rows of the FDA AI-Enabled Medical Device List, list vintage 2026-06-16 (canon REG-FDA-019). No 510(k) clearance, no De Novo grant, no PMA approval, and none needed
- Scribe price
- $0 to every verified US physician, NP, PA and medical student — the zero rung of the board's ambient price ladder (canon PB-055)
- Net cash
- ~$748.6M ($219.2M cash + $529.4M marketable securities), no debt, at 31 March 2026
- Buyback capacity
- $492.5M remaining on the $500M programme authorised 3 February 2026; the prior $500M programme completed with 11,591,950 Class A shares retired
- Stock-based compensation
- $121.6M FY2026, 18.9% of revenue, from $72.4M (12.7%) — diluted shares 199.0M against 187.0M basic
- Trailing P/E
- 21.3x on TTM diluted EPS of $0.98 at the $20.91 close (canon CF-DOCS-04 — quote it on this basis only; market cap divided by TTM net income gives 19.2x because the diluted count exceeds shares outstanding after buybacks, and the two must not be printed side by side)
- Forward P/E
- 14.59x — a THIRD-PARTY CONSENSUS ESTIMATE, not an observable and not an issuer disclosure (canon CF-DOCS-05). The board declines to assert a dollar forward-EPS figure (canon PB-066)
- Dual-class control
- Class B carries ten votes to Class A's one; Class B holders — executives, directors and affiliates — held approximately 79% of voting power at 31 March 2026 (10-K)
Growth drivers
- Engagement conversion, not user acquisition — more than 800,000 active prescribers used Doximity's workflow tools in Q4 FY2026, with 'nearly half' using its clinical AI — approximately 400,000 clinicians. That count BUNDLES AI search (Ask, formerly DoxGPT) with Scribe and is NOT a Scribe-only figure (canon CM-MKT-05, the caveat is mandatory). For scale, the entire PAID US ambient category's implied installed base is 240,000–600,000 clinicians (canon CM-MKT-04, derived from ~$600M of 2025 category revenue at a $1,000–2,500 blended ASP)
- Pharma budget mix shift from field force and print into point-of-care digital, which is the single revenue pool the board found actually funds software margin. Doximity's top-of-funnel is structural: the 10-K states more than 85% of US physicians are members and approximately 90% of graduating US medical students join before earning their degree, via Residency Navigator
- Module attach inside existing accounts rather than new logos — 125 customers contributed at least $500,000 of trailing-twelve-month subscription revenue in FY2026 (from 118 and 100 in the two prior years) and that cohort was approximately 83% of revenue. Growth here is a renewal-size question, not a sales-coverage question
- Distribution partnerships that push the Clinical AI Suite into settings Doximity does not reach directly: Aledade (announced 13 May 2026, independent primary-care practices) and Photon (in-workflow prescribing) — both company-announced, neither with disclosed economics
- Workflow surface expansion — Dialer telehealth, AMiON on-call scheduling, digital fax, PeerCheck and Scribe — which raises the quarterly active-provider count that Marketing Solutions is ultimately priced against. The AI is the funnel, not the SKU
- Capital return as a per-share driver in the absence of revenue growth: $431.7M of FY2026 repurchases retired stock at an average well above the current price, and $492.5M of a fresh authorisation remained at 31 March 2026 — roughly 13% of the 31 July market capitalisation
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-05-19. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’26
Bull & bear
The market is paying a below-market multiple for the highest-margin, most cash-generative, least regulated asset in health AI because it is extrapolating one soft guidance year as terminal. Doximity earns 89.1% gross margins and a 49.2% free-cash-flow margin from a customer — pharma's commercial budget — that no competitor in ambient documentation is even billing, holds ~$748.6M of net cash and no debt, and sits 73% below its 52-week high on 14.6x forward earnings (a third-party consensus estimate). The board's method is explicit: this is a multiple re-rate on an unchanged earnings base, not an earnings model — bull ~22x for a $31 target at 25% probability, base ~17x for $24 at 50%, bear ~10x for $15 at 25%, a probability-weighted ~+12% against the $20.91 spot (canon PB-012).
- The payer test is the whole edge, and Doximity is the only listed name that passes it cleanly. It holds zero FDA authorisations and no CPT code and still earned $196.1M of net income and $317.5M of free cash flow (+19%) on $644.9M of revenue. Compare the alternative structure the same board audits: GE HealthCare holds the largest FDA AI-clearance stock in the world at 130 authorisations and earned a 13.9% segment EBIT margin in Advanced Imaging Solutions in Q2 2026. Clearance stock is a barrier to entry, not a rent
- The de-rate is a growth-guide event, not a margin break. Gross margin is 89.1%, net margin 30.4%, adjusted EBITDA margin 55.5%, and free cash flow ACCELERATED to +19% in the same year the multiple halved. Nothing in the FY2026 statements shows a business breaking; the 52-week low of $17.15 already sits above the board's $15 bear target
- Free Scribe is the correct price, not charity, and it is a weapon. Doximity distributes ambient documentation to a free-tier base of roughly 400,000 clinicians using clinical AI — bundling AI search with Scribe, not Scribe alone (canon CM-MKT-05) — against an entire PAID ambient category whose implied installed base is 240,000–600,000 clinicians. It is competing for the same clinician minute at a price of zero while its rivals must justify $1,188–$5,000 a year against a measured marginal E/M return of $167.37 per adopting clinician per month, 95% CI $86.52–$248.21 ($2,008/yr at the point estimate, $1,038–$2,979 across the interval; JAMA 2026;335(16):1408-1417, an EXPLORATORY outcome the authors call a conservative lower bound)
- Capital return does the work if growth does not. ~$748.6M of net cash on a $3.76B market capitalisation, $317.5M of annual free cash flow, and $492.5M of buyback authorisation left at 31 March 2026 — the company retired 11.6M shares completing its prior programme and is buying at roughly a quarter of last September's price. At the current free-cash-flow run rate the implied enterprise value is under 10x FCF (approximate — it pairs a 31 July market cap with a 31 March balance sheet)
- Cost of revenue is 10.9% of revenue, so even a severe inference-cost shock is arithmetically bounded. Doubling total cost of revenue from here — a far worse outcome than the 25.9% growth actually printed — would still leave gross margin near 78%. The bear mechanism is real; its magnitude is capped by the smallness of the base it operates on
- The FY2027 guide is a management choice made four days into a new CFO's tenure and against a comparison that has already reset. Guidance of $664–676M with adjusted EBITDA of $323–335M is conservative by construction if pharma spend normalises and the Aledade and Photon channels convert; a return to high-single-digit growth on an unchanged 89% gross margin re-rates the multiple to ~22x without a single dollar of new margin
Doximity is a high-margin advertising business that has stopped growing, is guiding profit DOWN in absolute dollars, and is spending an accelerating amount on an AI product it has no intention of charging for. The 89.1% gross margin is the strongest fact about the company and it is also the ceiling — the board's own bull case concedes the margin has nowhere to expand to, so every dollar of return must come from growth or the multiple. Growth is guided to 3–5% and net revenue retention has fallen to 109% from 119%. Meanwhile the payer that funds the whole structure is a single discretionary line in pharma's commercial budget, one customer just crossed 11% of revenue, and the most credible competitor for that budget is a private company whose January-2026 mark exceeds Doximity's market capitalisation more than three times over.
- The FY2027 guide is worse than the headline. Revenue $664–676M is +3–5%, but adjusted EBITDA is guided to $323–335M against $357.8M delivered — a 6–10% ABSOLUTE decline in profit and roughly 640bp of margin compression at the midpoint. A business guiding revenue up and profit down is telling you its cost base is growing faster than its price, and the named driver is AI compute
- The margin erosion is already in the quarterly print, not just the outlook. GAAP gross margin was 86.7% in Q4 FY2026 against 89.5% a year earlier; cost of revenue grew 25.9% against 13.1% revenue growth; the 10-K names generative-AI platform usage and inference costs inside that line; and management attributes the non-GAAP fiscal-year move from 92.1% to 91.1% to increased AI compute cost (canon CF-DOCS-08). The board's falsifier is two consecutive quarters of gross margin below 85% while growth stays under 5% — Q4 printed 86.7% on +5% growth, which is one leg of a two-leg test
- The demand signal is deteriorating at the customer level, which is harder to argue away than a guide. Net revenue retention 114% → 119% → 109%; a single customer at 11% of revenue for the first time in three years; 125 accounts carrying approximately 83% of revenue; and a material share transacted through marketing agencies that represent multiple pharma clients and negotiate on their behalf. That is an advertising business's concentration profile, not a software company's
- OpenEvidence is the same trade with better momentum and it is taking the same budget. Reported ~$300M annualised revenue as of July 2026, roughly double the ~$150M at end-2025 (reported, no primary source — the company publishes nothing), free to verified clinicians, zero FDA authorisations, no CPT code, and a $12B post-money mark from 22 January 2026 that at the 31 July 2026 close exceeded Tempus ($7.92B) plus Doximity ($3.76B) combined by $320M, a 2.7% margin (canon CM-COH-09; the as-of is mandatory and a 3% move flips it). Doximity's own 10-K names OpenEvidence as a competitor for Marketing Solutions. The two companies are now publicly disputing each other's clinical-AI benchmark results — Doximity's 15 July 2026 release claims Doximity Ask ranked first on the ARISE NOHARM real-world clinical sample, OpenEvidence's counter-release claims physicians chose OpenEvidence in 22.3% of responses against 19.8% for all other external AI combined, and STAT and Fortune both reported on 29 July 2026 that the comparison is contested. It is UNRESOLVED and neither side should be read as settled (canon CM-BENCH-08 for the parallel Nature Medicine dispute). What is settled is that benchmark rank is not a payment: health-system leaders report benchmarks have minimal impact on purchasing (canon CM-BENCH-07)
- The economics of the free tier scale the wrong way. Prompts per user nearly doubled from January to April 2026. Every additional prompt is a cost with no matching price, and Doximity does not control the price of inference — it buys it from third-party platforms. Success in engagement is, mechanically, cost growth, and the only monetisation route is to sell more advertising against it, which requires pharma to keep bidding
- Ownership and disclosure both cut against the outside holder. Class B carries ten votes to one and insiders hold approximately 79% of voting power; there is no disclosed revenue split across Marketing, Hiring and Workflow Solutions and no Scribe attach rate, so the AI narrative cannot be separated from the ad business in any published number; stock-based compensation runs at 18.9% of revenue and creates a 12.0M-share dilution wedge; and a shareholder class action is live enough that $4.9M of non-ordinary-course legal expense was excluded from FY2026 non-GAAP results
What it is worth
MULTIPLE RE-RATING ON AN UNCHANGED FORWARD-EARNINGS BASE — explicitly not an earnings model, because the forward multiple that anchors it is a third-party consensus estimate rather than an observable (canon PB-012 / CF-DOCS-05). Every target is expressed as a re-rate on one implied base: bull ~22x, base ~17x, bear ~10x. The board asserts the 14.59x forward MULTIPLE and declines to assert a dollar forward-EPS figure, because that dollar is arithmetic performed on somebody else's estimate (canon PB-066, item 2). The trailing multiple is 21.3x and must be quoted on TTM DILUTED EPS of $0.98 — market capitalisation divided by TTM net income gives 19.2x instead, because the ~199.0M diluted weighted-average share count exceeds the 179.75M outstanding after buybacks, so the two figures must never appear in the same sentence (canon CF-DOCS-04). Two cross-checks that do not depend on a consensus estimate: with ~$748.6M of net cash and no debt at 31 March 2026 against the 31 July 2026 market capitalisation, implied enterprise value is near $3.0B — roughly 4.7x FY2026 revenue, 8.4x FY2026 adjusted EBITDA and 9.5x FY2026 free cash flow. All three pair a July market cap with a March balance sheet and should be read as approximate, not as struck multiples.
$15 (25% probability
~10x forward, through the $17.15 52-week low). Pharma commercial budgets contract further while the free tier stops being free at the margin: cost of revenue keeps compounding at roughly twice the revenue rate, GAAP gross margin follows Q4 FY2026's 86.7% below 85% for two consecutive quarters, and revenue growth goes to zero — the board's stated falsifier, fired. The FY2027 adjusted-EBITDA guide of $323–335M against $357.8M delivered is already the first instalment of that path. Concentration bites at the same time: the 11% customer renegotiates through an agency that holds several competing budgets, net revenue retention slips below 105%, and OpenEvidence — free, growing at a reported doubling rate, and holding the same zero-clearance, no-CPT-code structure — takes the incremental pharma dollar. In that branch Doximity is a no-growth advertising business with a decaying gross margin and 18.9%-of-revenue stock compensation, and 10x forward is not obviously cheap. The honest asymmetry: the bear case does not require the business to break, only for the free AI to keep getting more expensive while the budget that pays for it stops growing.
$24 (50% probability
~17x forward). Three to five per cent growth is the new normal — the market is right about the trajectory and wrong about the multiple. GAAP gross margin holds near 89.1% and net margin near 30.4%; free cash flow of $317.5M (+19%) keeps funding the buyback; adjusted EBITDA compresses toward the guided ~49% but the absolute cash generation is undisturbed; and the stock re-rates modestly from 14.6x to ~17x forward on durability rather than reacceleration. The AI stays free and stays a cost line, the pharma budget stays intact, and the company grinds out per-share value through share count rather than revenue. Note that the 52-week low of $17.15 sits above the bear target, so the base case does not require the market to concede anything except that a 30%-net-margin business with net cash should not trade at a growth-stock discount to its own cash generation.
$31 (25% probability
~22x forward on the consensus base). The FY2027 guide proves to have been conservatism from a new CFO rather than a run-rate, pharma commercial spend normalises, and the ~400,000 clinicians already using the clinical AI convert into higher-priced point-of-care inventory. Growth returns to high single digits on an unchanged 89% gross margin; the Aledade and Photon channels put the Clinical AI Suite in front of independent practices the enterprise scribes never reach; free cash flow compounds above $300M a year while $492.5M of buyback authorisation retires stock at a 73% de-rate; and the multiple recovers toward 22x forward on durability plus reacceleration. In this branch the market simply re-prices the only listed asset in health AI that is both profitable and growing.
Framing, not a recommendation, and not investment advice. The QAI board holds DOCS LONG at 7% of NAV — full core size and the book's single-name cap, $700,000 against an illustrative $10M NAV (canon PB-004) — with a stated action of accumulate-on-weakness and high conviction. Targets are bull $31 at 25%, base $24 at 50%, bear $15 at 25%, a probability-weighted $23.50 or about +12% against the $20.91 spot (canon PB-012). Two things about that sizing are load-bearing. First, the name is held for a NAMED MECHANISM — 89.09% gross margin, $196.05M of FY2026 net income, zero FDA authorisations, AI deliberately priced at zero, pharma's commercial budget paying — and would be SOLD the day that mechanism changes; it is not held because it is a profitable health-tech company. The rejected conventional proxy book held DOCS at 15% for exactly that weaker reason, and any 15% weight is the rejected book (canon PB-041 / PB-045). Second, the position goes on in stages: 4% on day one, completing to 7% only at the 6 August 2026 print and only if GAAP gross margin holds at or above 85% against the 89.09% FY2026 base AND Scribe remains free (canon PB-047). The falsifiers are two consecutive quarters of gross margin below 85% while revenue growth stays under 5%, and — counter-intuitively — Doximity beginning to CHARGE for Scribe, which cancels the second tranche rather than reading as bullish monetisation, because it removes the $0 rung that anchors the board's entire ambient-pricing finding. Portfolio context: DOCS and Waystar are the same analytical bet expressed twice long, on PAYER IDENTITY, with Epic as the unhedgeable shared counterparty (canon PB-031).
SWOT
Strengths
- The payer is right and the margin proves it — 89.1% GAAP gross margin, 30.4% net margin, 55.5% adjusted EBITDA margin and 49.2% free-cash-flow conversion, earned from pharma's commercial budget rather than a health system's IT opex line or a CMS fee schedule (canon CF-DOCS-07/09/11 — the board's control case)
- An identity moat that cannot be bought — more than 85% of US physicians are verified members and approximately 90% of graduating US medical students join before earning their degree (10-K). Verified clinician attention is the scarce input pharma is bidding for, and Doximity aggregates it
- Zero regulatory surface on the AI. No FDA authorisation across all 1,524 rows of the AI-Enabled Medical Device List (canon REG-FDA-019), no CPT code, no payer contract — so no clearance backlog, no reimbursement rate risk, no coverage determination gates growth. The entire regulatory apparatus that constrains every other name on this board simply does not apply
- A fortress balance sheet with an active return programme — ~$748.6M of cash and marketable securities, no debt, $431.7M repurchased in FY2026 and $492.5M of fresh authorisation outstanding — roughly 13% of the market capitalisation, at a price 73% below the 52-week high
- Structural pricing immunity in ambient documentation. Doximity cannot lose a price war it is not in: Scribe is $0, so the roughly sixty ambient vendors on the market as of March 2025, charging $1,188–$5,000 per clinician per year (reported list, no primary source — no ambient vendor publishes a price list; canon CM-PRICE-01), are competing against a free substitute funded by a different customer entirely
Weaknesses
- Growth has stopped — 20% (FY2025) → 13% (FY2026) → a guided 3–5% (FY2027), with Q4 FY2026 already at +5%. Net revenue retention fell 119% → 109%. A 3–5% grower is not a software compounder regardless of its margin, and the board's own bull case concedes the margin 'has nowhere to expand to'
- The free AI has started to cost money in a place investors can see. Cost of revenue grew 25.9% against 13.1% revenue growth, the 10-K names 'generative AI platform usage and inference costs' inside it, Q4 GAAP gross margin printed 86.7% against 89.5%, and the company itself attributes the FY2025-to-FY2026 non-GAAP move (92.1% → 91.1%) to increased AI compute costs (canon CF-DOCS-08)
- Guided profit is going backwards — FY2027 adjusted EBITDA of $323–335M against $357.8M delivered — an absolute decline of 6–10% and roughly 640bp of margin compression at the midpoint, in a year revenue is guided up
- Concentration is rising, not falling. One customer was 11% of FY2026 revenue — the first breach of the 10% line in three years — 125 customers were approximately 83% of revenue, and a material share is transacted indirectly through pharma marketing agencies, which places a negotiating intermediary between Doximity and the budget holder (10-K)
- Stock-based compensation at 18.9% of revenue ($121.6M, up from $72.4M) drives a 12.0M-share gap between basic and diluted counts and a $106.6M gap between GAAP and non-GAAP net income. The buyback is partly offsetting dilution rather than purely returning capital, and the dual-class structure leaves ~79% of voting power with insiders
Opportunities
- Prove the AI is a revenue line, not just a retention tool — but note the board treats this as a FALSIFIER rather than a catalyst: Doximity beginning to CHARGE for Scribe cancels the tranche-2 add outright, because it removes the $0 rung anchoring the entire ambient pricing finding (canon PB-012 / PB-047, the counter-intuitive rule)
- Convert the ~400,000 clinicians already touching clinical AI into higher-value Marketing Solutions inventory. Point-of-care AI placements are a scarcer, more attributable ad unit than a newsfeed impression, and Doximity owns both the identity graph and the moment
- Distribution into settings it does not reach directly — the Aledade partnership into independent primary care (13 May 2026) and Photon for in-workflow prescribing — extending the network beyond the health-system and academic footprint where Epic and the enterprise scribes concentrate
- Buy back stock at a 73% de-rate. With ~$748.6M of net cash, $317.5M of annual free cash flow and $492.5M of authorisation outstanding, Doximity can retire a double-digit percentage of its float without touching the balance sheet's solvency — the highest-certainty per-share lever available to it
- Own the safety narrative in clinical AI. Doximity's 15 July 2026 release claims Doximity Ask ranked first among all systems evaluated on the real-world clinical sample of the ARISE NOHARM benchmark (Stanford and Harvard physician researchers), ahead of OpenEvidence and frontier models — a credibility asset with pharma buyers, though the claim is contested by OpenEvidence and benchmarks are reported to have minimal impact on health-system purchasing (canon CM-BENCH-07)
Threats
- Pharma commercial-budget contraction is the whole thesis in one variable. There is no CPT code, no subscription lock from a health system and no clinical necessity underwriting the revenue — if pharma cuts point-of-care digital promotion, the 89.1% gross margin does not save the top line. The board's primary falsifier keys on exactly this
- OpenEvidence is attacking the same payer with the same free-to-clinician model and more momentum: reported revenue of ~$300M annualised as of July 2026 from ~$150M at end-2025 (trade press, no primary source — the company publishes nothing), against a $12B post-money mark from 22 January 2026 (canon CM-PRIV-OE-01/03). At the 31 July 2026 close that January mark exceeded Tempus ($7.92B) plus Doximity ($3.76B) combined by $320M, a 2.7% margin that a 3% move in either listed name flips (canon CM-COH-09 — the as-of is mandatory). Doximity's own 10-K names OpenEvidence as a Marketing Solutions competitor
- Inference cost scaling faster than engagement monetises. Prompts per user nearly doubled from January to April 2026 on a product that charges nothing; if the third-party model prices Doximity pays do not fall as fast as usage grows, the free tier becomes a cost centre — which is precisely the board's stated failure mode, and the mechanism the company has already conceded in its own margin commentary
- The record owner re-addressing the workflow layer. Epic AI Charting reached general availability on 4 February 2026, licensed inside the EHR relationship at a price Epic does not disclose and which must not be asserted (canon PB-055). Epic's revenue was $6.7B in 2025 (company-stated, unaudited; Epic files nothing, no primary source exists, no margin knowable — canon CM-EPIC-01). Doximity's scribe never needs an Epic contract, which is the defence; but Epic owns the moment of documentation, which is the attack
- Governance and disclosure asymmetry — ~79% of voting power sits with Class B holders, the company discloses no revenue split across Marketing, Hiring and Workflow Solutions, and it discloses no Scribe-specific attach rate — so an outside investor cannot separate the AI story from the advertising business it is embedded in. The company also carries an ongoing shareholder class action, for which non-ordinary-course legal expense of $4.9M was excluded from FY2026 non-GAAP results
Moats, dependencies & bottlenecks
Moats
More than 85% of US physicians are verified members and approximately 90% of graduating US medical students join before earning their degree, largely via Residency Navigator (10-K). Credential verification plus a pre-populated professional profile makes the graph expensive to rebuild and the audience trustworthy to an advertiser who must reach a specific prescriber. This is the asset; everything else is a way of using it.
The board's central finding is that margin sits where the buyer is pharma's commercial budget or the provider's own collected revenue, never the provider's IT opex line and never a CMS fee schedule. Doximity is the listed proof: 89.1% GAAP gross margin with zero FDA authorisations and no CPT code. Durability is capped, not by competition, but by the budget itself — it is discretionary, and one customer is already 11% of revenue.
AMiON on-call scheduling is the stickiest piece — a department's call schedule is a shared operational dependency, not a personal app. Dialer, fax and Scribe raise the quarterly active-provider count that Marketing Solutions is priced against. But these are free tools with no contract and no switching cost to the clinician, so the lock-in is habitual rather than structural, and the record owner sits closer to the documentation moment.
Strong as a competitive weapon, zero as a revenue asset High while the ad business funds it Scribe at $0 to every verified US clinician is the bottom rung of the board's ambient price ladder (canon PB-055) and the free substitute that constrains what the paid ambient tier can charge. The only controlled measurement of that benefit — $2,008/yr at the point estimate, $1,038–$2,979 across the 95% interval — is small relative to enterprise list prices, but the interval is wide enough that overpricing is not demonstrated, and the authors call the estimate a conservative lower bound (canon CM-RATIO-03). The counter-intuitive corollary the board enforces: Doximity starting to charge for Scribe would CANCEL the position's second tranche, not validate it.
Zero entries across all 1,524 rows of the FDA AI-Enabled Medical Device List (canon REG-FDA-019, list vintage 2026-06-16) — no 510(k) clearance, no De Novo grant, no PMA approval, and no CPT code. That is a structural cost and speed advantage over every diagnostic vendor on the board. It is also a live exposure: the 10-K flags enacted and proposed AI-in-healthcare legislation that could change how clinicians may use such tools, and a product that never sought a clearance has no regulatory standing to point at if that changes.
~$748.6M of cash and marketable securities, no debt, 49.2% free-cash-flow margin and $492.5M of live buyback authorisation. Not a competitive moat in the Porter sense, but it is the reason a 3–5% growth year is survivable without dilution, and the mechanism by which a de-rated multiple still compounds per-share value.
Dependencies
The revenue source — effectively the sole payer Marketing Solutions to pharma is the profit engine and it is a discretionary line. One customer reached 11% of FY2026 revenue, the first breach of the 10% threshold in three years, and 125 accounts carry approximately 83% of revenue. There is no CPT code, no clinical necessity and no multi-year health-system contract underwriting any of it. A promotional-spend cycle is the single variable that moves this position most.
Channel / negotiating counterparty The 10-K states some customers purchase indirectly through marketing agencies, some of which represent multiple customers, and that renegotiation of fees is a normal-course activity. An agency holding several pharma budgets negotiates against Doximity with far more leverage than any single brand team, and it also obscures the true end-customer concentration behind the disclosed 11%.
COGS input for the free clinical-AI suite The 10-K names 'generative AI platform usage and inference costs' inside cost of revenue. Doximity does not train frontier models and does not own GPUs, so it buys inference at a price it does not set, for a product it charges nothing for, whose usage it is actively trying to grow (prompts per user nearly doubled January to April 2026). Cost of revenue grew 25.9% against 13.1% revenue growth. This is the mechanical channel through which the board's falsifier would fire.
The inventory being sold Advertising revenue is priced against verified reach and engagement, so the >800,000 quarterly active prescribers on workflow tools is the real leading indicator — not Scribe attach. Attention is finite, and OpenEvidence, Medscape and the frontier chatbots are competing for the same minute. If engagement stalls, the inventory shrinks before the revenue does.
Adjacent platform / workflow gatekeeper Doximity's scribe never needs an Epic contract, which is exactly why the board holds it — but Epic AI Charting reached general availability on 4 February 2026 licensed inside the EHR relationship at an undisclosed price that must not be asserted (canon PB-055), and Epic owns the documentation moment and the chart the note lands in. Epic reported $6.7B of 2025 revenue (company-stated, unaudited; Epic files nothing, no primary source exists, no margin knowable — canon CM-EPIC-01).
Regulatory / compliance The 10-K flags numerous passed and pending laws on the use of AI in healthcare that could change how clinicians engage with these tools, plus AI-specific regimes raising compliance cost, and HIPAA / PCI-DSS / SOC 2 obligations across the platform. Separately, any tightening of rules governing pharmaceutical promotion to healthcare professionals lands directly on the revenue line — there is no diversification behind it.
Class B carries ten votes per share against Class A's one, and Class B holders — executive officers, directors and affiliates — held approximately 79% of voting power at 31 March 2026. Outside holders cannot force a capital-allocation or strategy change, including on the question the whole thesis turns on: whether the clinical AI ever gets a price.
Advantages
- The only listed name on the QAI board whose payer is pharma's commercial budget — the one buyer the board found actually funds software margin — held at full core size, 7% of NAV, the book's single-name cap (canon PB-004 / PB-012)
- 89.1% GAAP gross margin, 30.4% net margin and 49.2% free-cash-flow margin, all primary-sourced from the FY2026 filings — the highest verified margin structure on the board
- Verified membership covering more than 85% of US physicians, with roughly 90% of graduating US medical students joining before they qualify — a top-of-funnel no competitor can buy
- Zero FDA AI authorisations and no CPT code (canon REG-FDA-019) — no clearance queue, no coverage determination, no administratively-set price, and no reimbursement rate risk
- ~$748.6M net cash, no debt, and $492.5M of remaining buyback authorisation against a market capitalisation 73% below the 52-week high
- A free ambient scribe distributed to roughly 400,000 clinicians using its clinical AI (a bundle of AI search and Scribe, not Scribe alone) — a substitute that structurally caps what the entire paid ambient category can charge
Weaknesses
- Revenue growth guided to 3–5% for FY2027 after 20% and 13%, with Q4 FY2026 already printing +5%
- Adjusted EBITDA guided DOWN in absolute dollars — $323–335M against $357.8M delivered — roughly 640bp of margin compression at the midpoint
- Q4 FY2026 GAAP gross margin of 86.7% against 89.5%, with cost of revenue up 25.9% on 13.1% revenue growth and generative-AI inference named inside that line
- Net revenue retention down to 109% from 119%, and one customer at 11% of revenue for the first time in three years
- Stock-based compensation at 18.9% of revenue, a 12.0M-share basic-to-diluted wedge, and ~79% of voting power held by Class B insiders
- No disclosed revenue split by solution line and no Scribe attach rate — the AI story is unverifiable from the outside in any published figure
Bottlenecks
- Revenue growth, full stop — 3–5% guided against an 89.1% gross margin that the board's own bull case says has nowhere left to expand, so the return has to come from the multiple or from buyback
- Pharma promotional-budget size and cycle — the single pool the entire profit engine draws on, with one customer already at 11% of revenue
- Inference cost per prompt, which Doximity buys rather than sets, on a product priced at zero whose usage it is deliberately growing
- Monetisable clinician attention — >800,000 quarterly active prescribers is close to the ceiling of the verified US prescriber population, so incremental inventory has to come from more engagement per clinician rather than more clinicians
- Disclosure opacity — no revenue split across Marketing, Hiring and Workflow Solutions and no Scribe-specific attach rate, so the AI thesis cannot be validated or falsified in any published number until the company chooses to break it out
- Sales cycle dependence on agency intermediaries who hold multiple competing pharma budgets and renegotiate fees as normal course
- The board's own tranche gate — the second half of the position is event-gated on gross margin holding at or above 85% AND Scribe remaining free (canon PB-047), so the position cannot be completed on price alone
Top signals & trends
Top signals
The board's control case, primary-sourced (canon CF-DOCS-06/07/09/11, REG-FDA-019). It is the cleanest refutation on the board of the idea that clearance stock or a reimbursement code is what earns margin in health AI.
The guide is an issuer disclosure (canon CF-DOCS-13); the consensus comparison and the price reaction are secondary. This single event is the entire de-rate — the stock now sits ~73% below its 52-week high while the margin structure is intact.
An absolute profit decline of 6–10% and roughly 640bp of margin compression at the midpoint, in a year revenue is guided up. This is the guidance line that gets least attention and says the most about the cost of the free AI.
The quarterly print, computed from the FY2026 release reconciliation. The board's falsifier requires two consecutive quarters below 85% while growth stays under 5%; Q4 is 170bp above the line with growth already at 5%. This is the number to watch first.
10-K key-metrics and concentration disclosures. Retention decay plus rising single-customer concentration is a demand-side signal that is harder to explain away as conservatism than a guidance range is.
The 'nearly half' figure BUNDLES AI search with Scribe and is not a Scribe-only count (canon CM-MKT-05 — the caveat is mandatory). Bullish as inventory and as a free substitute at scale against a paid ambient installed base of 240,000–600,000 clinicians; bearish as a cost driver, because every extra prompt is inference Doximity pays for and does not bill.
Capital return running ahead of free cash flow, funded by letting the securities portfolio run down. At the 31 July market capitalisation the remaining authorisation is roughly 13% of the equity.
Both releases are partisan. OpenEvidence's counter-claim is that physicians chose it in 22.3% of responses against 19.8% for all other external AI combined; STAT and Fortune both reported on 29 July 2026 that the comparison is contested. Read alongside the unresolved Nature Medicine dispute over specialised clinical AI versus frontier models (canon CM-BENCH-08) and the finding that benchmarks have minimal impact on health-system purchasing (canon CM-BENCH-07). Neither side is settled.
The dated catalyst (canon CF-DOCS-14 / PB-015): the first read on whether the $664–676M guide was conservatism or the new run-rate, and the gate on the position's second tranche (complete DOCS 4% → 7% only if gross margin holds at or above 85% against the 89.09% FY2026 base AND Scribe remains free — canon PB-047). No result from this print is carried anywhere in this dossier.
A new CFO issuing the first guide is a standard reason to expect conservatism, which is the bull case's mechanism. The partnerships extend distribution into independent primary care and prescribing workflow, but neither has disclosed economics, so neither is a model input.
Trends
The secular driver behind the whole model and the one pool the board found actually funds software margin. It is also the concentration risk: it is one discretionary budget line, and Doximity has no second payer.
Doximity is on the correct side of this. The zero and bundled rungs — Doximity Scribe at $0, athenahealth's athenaAmbient at no additional cost, Epic AI Charting licensed inside the EHR relationship at an undisclosed price — are re-addressing the standalone ambient dollar rather than destroying it (canon PB-055). A vendor charging nothing cannot lose a price war; it can only lose the attention.
The clearest company-specific instance of a sector-wide effect: cost of revenue +25.9% on +13.1% revenue, generative-AI platform usage named in the 10-K's own definition, and a company-attributed non-GAAP gross-margin move from 92.1% (FY2025) to 91.1% (FY2026). Free products with usage-linked COGS are where this lands first.
It is the pattern that works — and it is now crowded. OpenEvidence runs the identical model (free to verified clinicians, monetised entirely by pharma and device advertising, zero FDA authorisations, no CPT code) with reported revenue roughly doubling to ~$300M annualised in the seven months to July 2026 (reported, no primary source). Validation of the mechanism and competition for the same budget are the same event.
Purpose-built clinical benchmarks are unsaturated and the leaderboards disagree; the ARISE NOHARM results are now being claimed by two competitors simultaneously, and the parallel Nature Medicine dispute over frontier models versus dedicated clinical products is unresolved (canon CM-BENCH-08). Health-system leaders report benchmarks have minimal impact on purchasing (canon CM-BENCH-07) — but pharma buyers of clinician attention may weight safety credibility differently, which is the only channel by which this matters to Doximity's revenue.
Epic AI Charting generally available 4 February 2026 and Epic Penny live at 200+ organisations. Epic is a private counterparty with no security to hedge against, and it is simultaneously re-addressing the ambient layer Doximity prices at zero and the revenue-cycle pool the board's other long sits in (canon PB-031 — the second shared factor across the book and the one that cannot be hedged).
Doximity is 73% below its 52-week high and Waystar roughly 49% below its own; the board flags further health-IT multiple compression as a factor hitting both longs regardless of payer identity (canon PB-031). The 17-name pure-play health-AI cohort — 16 US-exchange-listed names plus one OTC-quoted (Veradigm, Nasdaq-delisted 2024), and two telehealth platforms rather than health-AI pure plays (Teladoc and Amwell) — was ~$33.90B at the 31 July 2026 close, a board-computed aggregate of aggregator-sourced component quotes, against roughly $33.05B of six private marks that are last-round figures, not uniformly post-money, dated across June 2025 to May 2026, not marks-to-market (canon CM-COH-01 to CM-COH-04). That comparison is the investability gap the board's 89% cash position exists to answer.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Doximity buys inference rather than training frontier models; the 10-K names 'generative AI platform usage and inference costs' inside cost of revenue. Doximity does not disclose which model providers it uses, so no vendor is named here — but the dependency is a price it does not set for a product it does not charge for.
Cloud hosting is the first-named component of cost of revenue. HIPAA-compliant hosting for Dialer telehealth, Scribe audio processing (no audio retained, per the 10-K product description) and the member platform. Azure = MSFT, GCP = GOOGL — note Microsoft is simultaneously a supplier and a competitor.
Indirect — the GPU layer under whichever inference platform Doximity buys from. Falling inference prices flow straight to Doximity's gross margin because the sell-side price of the AI is fixed at zero; rising usage flows the other way.
The verification layer that makes the network sellable to pharma. Profiles are pre-populated from publicly and commercially available third-party data before a member refines them (10-K) — the input that turns an audience into a targetable prescriber list.
Medical content publishers and clinical editors Newsfeed content is drawn from in-house editorial and third-party sources; editorial and content expense sits in cost of revenue. The content is what brings the clinician back between visits, which is what makes the ad inventory recurring.
The profit engine. Marketing Solutions sold to brand teams at the top pharmaceutical manufacturers, billed as advance-paid subscriptions recognised ratably. One customer reached 11% of FY2026 revenue — the first breach of the 10% line in three years — and 125 accounts at ≥$500k of TTM subscription revenue were approximately 83% of total revenue.
Buy both Marketing Solutions (service-line and physician-referral marketing) and Hiring Solutions (recruiting into the network's passive candidate pool). This is the buyer the board's payer test says does NOT fund software margin — which is why Doximity's exposure to it being the smaller share of revenue is a feature, not a gap.
Hiring Solutions plus Curative Talent, the owned staffing business. The 10-K states revenue from temporary and permanent medical recruiting services was not significant to total revenue in any of FY2024–FY2026.
The same point-of-care inventory sold to a second promotional budget with the same verified-prescriber targeting logic. Not separately disclosed, and therefore not sized here.
PRIVATE, and the single most direct threat — the same business model aimed at the same payer. Free to verified US clinicians, monetised entirely by pharma and device advertising, zero FDA authorisations across all 1,524 rows of the FDA list, no CPT code. Reported ~$300M annualised revenue as of July 2026 from ~$150M at end-2025 (trade press, no primary source; the company publishes nothing), 757,000 clinicians and more than 20M clinical conversations a month. Last disclosed mark $12B post-money on a $250M Series D co-led by Thrive Capital and DST Global, 22 January 2026 — a last-round post-money, not a mark-to-market, and seven months stale at 6 August 2026. At the 31 July 2026 close that mark exceeded Tempus ($7.92B) plus Doximity ($3.76B) combined by $320M, a 2.7% margin (canon CM-COH-09; a 3% move in either listed name flips it). Named by Doximity's own 10-K as a Marketing Solutions competitor, and its public sparring partner over the ARISE NOHARM results.
Two separate attacks from one balance sheet. LinkedIn competes for medical professional networking and directly against Hiring Solutions — Doximity's 10-K names LinkedIn, Facebook, Google and X as competitors for members, and Microsoft Teams as a Workflow Solutions competitor. Separately, Microsoft/Nuance holds the largest share of the paid US ambient-scribe category (roughly a third of the ~$600M 2025 pool) with Dragon Copilot, marketed on '13 additional appointment slots per provider per month' — a claim the board measures at roughly 6x the controlled JAMA estimate. Microsoft cannot easily attack the pharma advertising budget, which is the profit engine, so the overlap is on attention and workflow rather than revenue.
PRIVATE. The incumbent HCP media property and the closest structural analogue on the sell side — named verbatim in Doximity's own 10-K as a Marketing Solutions competitor ('online outlets such as health-related websites and mobile apps, like WebMD's Medscape'). It sells the same thing to the same pharma brand teams: reach into clinician attention. Older franchise, weaker clinical-workflow surface, no free ambient scribe, but long-standing agency relationships and a deep content library. Ownership sits inside a private group and no current financials are public, so none are asserted here.
PRIVATE. Not a competitor for pharma dollars at all — a competitor for the clinician's documentation moment, which is Doximity's engagement moat. Epic AI Charting reached general availability 4 February 2026, licensed inside the EHR relationship at a price Epic does not disclose and which must not be asserted (canon PB-055). Epic reported $6.7B of 2025 revenue (company-stated, unaudited; Epic files nothing, no primary source exists, no margin knowable — canon CM-EPIC-01) and held 43.7% of US acute-care hospitals in 2025 per KLAS, as relayed by the trade press — the KLAS report itself is paywalled (canon CM-KLAS-01). Doximity's scribe never needs an Epic contract; that is the defence, and it is also the limit of the reach.
The largest incumbent in pharma commercial services — HCP data, prescriber-level analytics, media planning and measurement. It competes for the same commercial budget line rather than for the clinician's attention, and it is the party that measures whether a Doximity placement worked. A shift of pharma spend toward measurement-integrated, closed-loop channels is a threat to any pure-attention property; IQVIA's scale in the analytics layer makes it the most credible aggregator of that shift.
Owns the pharma commercial system of record — Vault CRM plus content management — and, through Veeva Crossix, the HCP campaign-measurement layer that grades whether a Doximity placement worked (Crossix HCP Digital measures outbound online media including third-party programmes and eNewsletters, per Veeva's own product pages). Not a media property, so it does not compete for the clinician's attention; it competes for control of the budget and the scorecard, which is the same logic by which Epic threatens the ambient layer.
PRIVATE. The paid ambient-scribe leader — approximately 30% of the ~$600M US category in 2025, 250+ health systems reported by mid-2026, $5.3B post-money on a June 2025 Series E (fourteen months stale at 6 August 2026) plus a reported April 2026 extension at an undisclosed price. It is not competing for Doximity's revenue: its payer is the health system's budget, which is the side of the payer test the board is not long. It competes for the same clinician minute at a price Doximity has set to zero — which is why the board's watchlist lean on Abridge is 'watch, do not chase the mark'.
PRIVATE (Bain Capital and Hellman & Friedman). Ships athenaAmbient at no additional cost inside the core athenaOne fee — the only confirmed zero-incremental-fee EHR-native ambient bundle. It reaches the ambulatory and independent-practice tier Doximity's Aledade partnership is aimed at, and it reaches it through a contract the practice already holds. No overlap on pharma revenue; direct overlap on the free-scribe rung.
Tokyo-listed (TSE 2413). The closest international analogue to the whole Doximity model — a members-only physician network monetised by pharmaceutical marketing, operating m3.com in Japan, MDLinx in the United States (acquired 2006) and Doctors.net.uk in the UK (acquired 2011). Same two-sided structure, same payer. Its direct head-to-head with Doximity inside US pharma brand budgets is limited today, so it matters less as a share threat than as the read-across on how a mature version of this exact business model ages — and as a plausible consolidator of US HCP-media assets.
Doximity's 10-K names the Hiring Solutions competitive set generically — 'large and regional staffing companies, job boards, self-service recruiting tools, and medical recruiting firms' — and no individual firm is identified by the issuer, so none is named here. This is the smallest and least strategically important of the three lines: the 10-K states revenue from temporary and permanent medical recruiting services was not significant to total revenue in FY2024, FY2025 or FY2026. It is, however, the line most directly exposed to health-system labour-budget cycles rather than to pharma promotional spend.