Cash and Buybacks

Doximity converts a large share of revenue into cash and carries no debt, and its diluted share count has fallen even as stock compensation doubled. But the two facts sit in tension. Stock-based compensation reached $121.6M in FY2026 — a real cost that free cash flow adds back — and the flagship $500M repurchase program retired 11.6M shares at an average near $43 while the stock now trades close to $22. The cash is real; the largest tranche of buyback was expensive.

The cash engine

The business throws off a lot of cash. In FY2026 Doximity generated $326.5M of operating cash flow on $644.9M of revenue, spent almost nothing on property, and ended the year with $748.6M of cash and marketable securities and no debt [1] [2]. Operating cash flow ran at 1.7x net income of $196.1M [3] — the kind of conversion that ordinarily signals high earnings quality.

Source: FY2026 Annual Report (Form 10-K), Consolidated Statements of Cash Flows and Results of Operations [4] [5]; cash flow after SBC ($326.5M − $121.6M) is a calculation.

Where the gap between cash and earnings comes from

Most of the distance between $326.5M of operating cash flow and $196.1M of net income is a single add-back: stock-based compensation of $121.6M [6]. Cash flow treats it as non-cash and adds it back; a shareholder cannot, because the shares issued to employees are a claim on the same equity. Netting it out, the cash the business produces for owners was closer to $205M — about 32% of revenue rather than the ~51% an operating-cash-flow margin implies.

The cost is also rising, and rising fastest in the part of the business the report has already flagged as the reinvestment engine. Total SBC climbed from $51.1M in FY2024 to $72.4M in FY2025 to $121.6M in FY2026 — from 10.7% of revenue to 18.9%. Within that, research-and-development stock comp nearly quadrupled over two years, from $12.0M to $46.2M, as Doximity granted equity to build out the AI Pivot team [7].

Source: FY2026 Annual Report (Form 10-K), Results of Operations — stock-based compensation by function [8]. "Rest" = total SBC less the R&D component.

That is not, by itself, a mark against the company — paying an AI team in equity is a defensible way to fund the build without touching the cash pile. But it changes how the cash story should be read: the reported cash margin flatters the underlying economics by roughly 13 points, and the flattering item is growing.

The buyback record

Doximity pays no dividend; its entire return of capital runs through repurchases, and it has been active. Since FY2022 it has spent roughly $921M of cash retiring stock [9], across four authorizations. The prices it paid, program by program, are what determine whether that spending created value.

Sources: FY2024 10-K (programs through April 2024) [10]; Q2 FY2025 10-Q (16,480,514 shares completed by April 2024) [11]; FY2026 10-K (May 2024 and February 2026 programs) [12]. Average prices are calculated as dollars authorized/spent divided by shares retired.

The first $410M — spent across programs completed by April 2024 — retired 16.48M shares at an average near $25 [13] [14]. Against a stock near $22 today, that early buying looks roughly break-even.

The $500M program authorized in May 2024 is the largest of the four. Doximity retired 11.59M shares under it and completed it by March 31, 2026 — an average of about $43 per share [15]. Most of that half-billion was deployed through calendar 2025, when the stock spent the year between roughly $50 and $70 before falling sharply into early 2026. On February 3, 2026, after the drawdown, the board authorized a fourth $500M program; by March 31 it had bought just 321,080 shares for $7.5M — about $23 each — leaving $492.5M unspent [16].

What the buyback bought

Two readings are both true, and they pull in opposite directions.

On the supportive side, the buyback has done its arithmetic job: it has held the share count down against heavy dilution. Diluted weighted shares peaked at 213.4M in FY2023 and fell to 199.0M in FY2026, a 6.7% reduction, even as stock comp doubled; actual shares outstanding at year-end were 183.1M [17]. The company funded all of it from cash generation without touching its $748.6M balance-sheet cushion or issuing debt [18]. In units, per-share value is being concentrated, not diluted away.

Source: FY2026 Annual Report (Form 10-K), Consolidated Statements of Operations (weighted-average diluted shares) [19]. FY2021–FY2022 reflect the June 2021 IPO.

On the critical side, price paid decides whether reducing the count actually created value, and the biggest program paid up. The 11.59M shares bought at ~$43 are worth about $252M at a $22 quote — roughly $250M below cost on a mark-to-market basis. Put differently, had that $500M been spent at today's price it would have retired about 23M shares rather than 11.6M; the timing cost the company on the order of 11M shares of potential retirement, close to 6% of the share base. The pattern is the uncomfortable part: repurchase spending peaked in FY2026 at $432M, the same window in which the stock was most expensive [20], and the fresh, cheaper authorization has barely been used.

Two caveats keep this from being a verdict. A buyback is properly judged against intrinsic value, not the current tape — if $22 understates the business more than $43 overstated it, the flagship program looks less like an error and more like an unlucky entry point. And the decisions belong to a control structure worth naming: Class B holders, led by the founders and executives, command about 79% of the voting power, so capital-allocation calls are effectively insider calls rather than the board's to be pressured on [21]. What would change the read is observable: whether the $492.5M still authorized is deployed near current prices and the count falls materially in FY2027, or whether it sits idle while the average cost stays anchored at $43.

The cash generation underneath all of this is genuine, and it is what lets Doximity self-fund the reinvestment year the rest of the report describes. The question this chapter leaves is narrower than the thesis but adjacent to it: a company that buys back its most expensive shares and slows as they cheapen is returning cash without reliably creating per-share value — and, so far, has left most of the cheaper authorization unspent.