Capital Allocation

Capital Allocation

Doximity returns surplus cash one way — buybacks. It has spent roughly $921 million repurchasing stock since the 2021 IPO, paid no dividend, made almost no capital expenditure, and closed one small acquisition, all funded from cash flow with no debt. The structure is disciplined; the timing is uneven. Its largest program, $500 million at a $43 average, was spent into the 2025 peak weeks before the shares fell to about $22. And a rising stock-based-pay bill means gross spend buys only a modest net reduction in the share count.

One lever, self-funded

With near-zero capital intensity and no dividend, Doximity's only real capital-allocation decisions are how much of its cash to return through buybacks and how much to hold. Cumulative repurchases since fiscal 2022 total about $921 million, every dollar drawn from operating cash flow — the company carries no debt and still ended fiscal 2026 with $748.6 million of cash and marketable securities [1]. It has never declared a dividend and does not intend to [2]. The retained cash is not idle: it earned $35.1 million of other income in fiscal 2026 [3], almost entirely interest on cash and securities [4] — roughly a 4.7% yield on the balance and about 18% of net income.

Sources: cumulative buybacks derived from Consolidated Statements of Cash Flows, FY2022–FY2026 10-Ks [5]; net cash [6]; dividend policy [7]; interest income [8].

Repurchases have run every year, scaling with cash generation. They stepped up sharply in fiscal 2026, when the $431.7 million spent was nearly half of everything the company has ever returned [9].

Source: Repurchase of common stock, Consolidated Statements of Cash Flows, FY2022–FY2026 10-Ks [10].

The timing: trough, then peak, then trough

Read against the share price, the record splits into three phases. The first, from the IPO through early 2024, bought the post-IPO trough: successive programs authorizing $410 million retired 16.5 million shares, completed by April 2024 [11], at a blended cost near $25 a share while the stock traded between roughly $23 and $36. The earliest tranche, $340 million for 13.8 million shares, was complete by October 2023 [12]. Those were well-timed purchases.

The second phase is where the money went out. The $500 million program authorized in May 2024 was completed in full, retiring 11,591,950 shares [13] — a blended $43.13 a share. Only $76.0 million of it had been spent by March 2025 (1.9 million shares) [14]; the remaining $424 million was spent through fiscal 2026, at roughly $44 a share, as the stock climbed from the low $50s to a September 2025 peak above $70. Doximity was buying its most stock at its highest prices, immediately before the fiscal-2027 guidance cut sent the shares to about $22.

The third phase began in the fall. On February 3, 2026 the board authorized a fresh $500 million program; by fiscal year-end it had bought just 321,080 shares for $7.5 million — a $23.36 average — leaving $492.5 million authorized and unspent [15]. Management has reverted to buying near the trough, with most of the authorization still in reserve.

Sources: shares and program status per Stock Repurchase Program notes, FY2024–FY2026 10-Ks [16] [17]; average prices derived from disclosed spend divided by shares retired. The IPO-through-2024 average is estimated from cumulative cash-flow spend.

The completed $500 million program is where the timing cost concentrates. At $43.13 a share it cost the company about twice the current price. The same $500 million deployed at $22 would have retired roughly 22.7 million shares rather than 11.6 million — about 11 million more, or 6% of the shares outstanding. That gap is the price of buying into the run-up, and it is visible only in hindsight: the stock spent the entire program period well below its 2021 high, and the collapse followed an unforeseen guidance cut, not a slow bleed the company could have front-run.

Sources: $500M program average [18]; $492.5M remaining and $7.5M spent under the Feb 2026 program [19]; FY2026 spend [20]; share price as reported.

Gross buybacks, modest net shrinkage

The buyback's effect on ownership is smaller than the headline dollars suggest, because stock-based compensation keeps refilling the share pool. That charge has nearly tripled in two years — from $47.4 million in fiscal 2024 to $72.4 million in fiscal 2025 to $121.6 million in fiscal 2026 [21]. In fiscal 2026 it equalled 28% of the $431.7 million spent on repurchases.

Sources: repurchases [22]; stock-based compensation [23].

The result is real but slow deleveraging of the share count. Diluted weighted-average shares fell from 205.7 million in fiscal 2024 to 201.2 million in fiscal 2025 to 199.0 million in fiscal 2026 [24] — a 3.3% reduction over two years against about $552 million of repurchases. Buybacks are winning the contest against dilution, but by a narrow margin: the gross-to-net gap is stock-based pay. To the extent the compensation bill keeps climbing, a growing share of each buyback dollar mops up new issuance rather than shrinking the float — the same charge that separates reported free cash flow from owner earnings (Margin of Safety). As of March 31, 2026 the company had 132,163,035 Class A and 50,896,611 Class B shares outstanding [25].

The one deal

Acquisitions have not been a use of capital in any material sense. The only transaction in the record is Pathway Medical Inc., an AI medical-knowledge platform bought in July 2025 for $36.3 million of consideration, including $26.7 million of cash, plus up to $23.9 million of restricted stock granted to retained employees vesting over five years [26]. This is an acqui-hire — technology and an engineering team folded into the clinical-AI effort — not a growth-by-M&A strategy, and its return is inseparable from the AI investment case (The AI Contest) rather than a standalone capital-allocation bet.

Reading the record

Doximity's capital allocation is well-constructed and unevenly executed. On construction it scores high: every buyback was self-funded, no debt was taken, the retained cash earns a market yield, and the float has genuinely shrunk. On execution it is mixed — the company bought its own trough twice, but committed its single largest tranche, the completed $500 million program, at a $43 average into what proved to be a cyclical peak. The strongest fact for management is what came next: it pivoted to buying at $23 and holds $492.5 million of authorization to deploy at today's depressed price [27]. Whether the multi-year record nets to accretive depends on how aggressively that reserve is spent now: deployed near $22 it retires roughly twice the shares per dollar it did at $43, while left to the same drift that carried the last program up to $70 it would repeat the mistimed 2025 tranche.

Two developments would clarify the read. The first is the pace of the Feb 2026 program: a large fraction of $492.5 million retired near current prices would more than offset the mistimed 2025 tranche; a trickle would signal the buyback is a dilution valve, not a conviction purchase. The second is whether net share count starts falling faster than stock-based compensation refills it — the point at which repurchases stop merely absorbing issuance and begin compounding per-share value for the owners who stay.