Self-Help

Self-Help

Salesforce can clearly outlast its problem: net debt of roughly $4.8 billion sits against $14.4 billion of annual free cash flow, and the only near-term maturity is a $4.0 billion acquisition bridge. It is spending heavily on repurchases — $28 billion over three years and a $50 billion authorization — and returned 99% of free cash flow in FY2026. But stock-based compensation absorbs roughly two-thirds of that, so the share count has fallen only about 1.4% a year.

The balance sheet against the problem's duration

The dislocation here is a growth-and-multiple problem, not a solvency one. Salesforce carries $14.44 billion of debt at carrying value against $9.6 billion of cash, equivalents and marketable securities, plus an undrawn $5.0 billion revolving facility that runs to October 2029 [1]. Net debt is roughly $4.8 billion on that liquidity measure, or $7.1 billion counting cash alone — either figure is a fraction of one year's free cash flow. The deterministic feature file leaves the balance-sheet class as unknown because FY2026 EBITDA was missing from its inputs; on the figures in the filings, leverage is well below one turn of cash flow by any reasonable EBITDA estimate.

Source: FY2026 Annual Report (Form 10-K), Note 9 Debt — contractual future principal payments [2].

The maturity wall is benign. The $4.0 billion due in FY2027 is the 364-day tranche of the Informatica acquisition financing, borrowed in November 2025 at a floating 4.42% [3] — repayable from a single quarter of operating cash flow. Nothing falls due in FY2028; the $4.5 billion in FY2029 is two senior-note tranches plus the three-year Informatica facility. Everything from FY2031 out is $6.0 billion of senior notes struck in 2021 at coupons of 1.95% to 3.05%, so cheap that their market value was just $6.7 billion against a higher face [4]. Total cash interest paid in FY2026 was $276 million [5]. The company reported compliance with all debt covenants at year-end [6].

The one thing worth naming is that this balance sheet is a year past its most comfortable point. Through FY2025 Salesforce ran roughly net-cash. In FY2026 it drew $6.0 billion of new debt to help fund the $8.1 billion Informatica purchase while still buying back $12.6 billion of stock and paying $1.6 billion of dividends [7]. It ran all three uses of cash at once and let leverage tick up to do it. That is a choice, not a constraint — but it is the first evidence that repurchases have to share the till with acquisitions, which is the live question for capital allocation, not debt service.

The repurchase record — executed, not just authorized

The buyback is real cash out the door, not a press release. Salesforce spent $7.62 billion (FY2024), $7.83 billion (FY2025) and $12.60 billion (FY2026) on repurchases per the cash-flow statement [8], on top of about $4.0 billion in FY2023 when the program began. The Board lifted the authorization by $20 billion in September 2025 to a $50 billion aggregate, of which $17.9 billion remained at fiscal year-end [9].

Source: FY2026 Annual Report (Form 10-K), Share Repurchase Program table [10]. Current price $170.77 as of 2026-07-17.

The prices paid tell their own story. The FY2024–FY2026 repurchases averaged $210, $260 and $254 a share [11]; the stock trades at $170.77 today. The last three years of buybacks were transacted 23% to 52% above the current quote — capital deployed into a stock the market has since marked down. The mirror image is that the same program is far more accretive at $171 than it was at $254, which is exactly the logic management is now invoking (below).

The harder number is what all that cash bought. Between FY2024 and FY2026 the program retired about 116 million shares gross [12], yet shares outstanding fell only from 997 million to 956 million — a net reduction of 41 million. Roughly two-thirds of the buyback was consumed offsetting stock-based compensation and employee-plan issuance. SBC ran $2.79 billion, $3.18 billion and $3.51 billion across those three years, about 8.4% of revenue [13].

Source: share count from fit_features.share_count_trend, derived from company filings; FY2017–FY2026.

This is the framework's central test for capital allocation, and Salesforce sits on the wrong side of the easy version of it. The deterministic feature file records the share count as rising — a positive five-year change of about 0.55% a year — and over the full decade the count climbed from 700 million to 956 million as stock-funded acquisitions (Slack, Tableau, MuleSoft) and compensation added shares faster than any buyback removed them. A share count that keeps rising on SBC and serial acquisitions is a framework hard-fail, and on the multi-year record that flag is tripped. The two-sided fact is that the trend inflected in FY2024: the count has fallen for three consecutive years as buyback spend finally overtook issuance. But the pace is modest — roughly a 1.4% net reduction per year — not the compounding retirement engine the framework is built around. At a hypothetical 10% adjusted yield the flywheel would retire close to a tenth of the float annually; Salesforce is delivering a small fraction of that, because the buyback is fighting its own compensation bill.

The levered exception does not apply and does not need to: it is reserved for balance sheets carrying a ~25%+ yield to justify the debt, and Salesforce is neither that levered nor that cheap.

Management's intent, from the record

Management has been explicit, and consistent, that the depressed price is the reason to lean in. On the Q4 FY2026 call Benioff tied the enlarged authorization directly to the drawdown: "we're using our remarkable cash flows to take advantage. This is not our first SaaSpocalypse… that's why we are doing this incredible repurchase authorization of $50 billion… because these are some low prices" [14]. He quantified the return: "In fiscal year '26, we returned more than $14 billion or 99% of our free cash flow to shareholders" [15]. Subsequent to that call the company commenced a $25 billion accelerated repurchase — half the entire authorization — explicitly framed against the low stock price, per the curated call record.

The counter-current is that management insists on a balanced allocation policy rather than a buyback-first one. Asked on the same call why not "lean harder" into acquisitions given the $50 billion authorization, the reply was that "there are many uses of cash such as dividends, buybacks, and acquisitions… Our approach focuses on using cash effectively and keeping our balance sheet strong," alongside a 5.8% dividend increase [16]. That is the tension the FY2026 spending pattern already made concrete — an $8 billion acquisition and a record buyback in the same year. For a framework that wants repurchases prioritized precisely when the yield tailwind is maximal, Salesforce's willingness is genuine and currently pointed the right way, but it is shared with an active acquisition appetite that can pull cash the other way. Insiders are not buying the dip in the open market — across the trailing Form 4 record the activity is option exercises, grants and tax-withholding, with just three small open-market purchases (SEC EDGAR Form 4 filings).

The absurdity check

At $170.77 the market values Salesforce equity at about $163.3 billion. Reported FY2026 free cash flow of $14.40 billion would retire the entire float in 11.3 years; net of the $3.51 billion SBC charge, about 15 years [17]. The feature file's float_retirement_years is not_computable because it depends on an adjusted-FCF figure the deterministic pipeline could not derive (see Yield); the arithmetic above uses reported free cash flow and is directional. The point is what the number is not: this is not the ~3-year float retirement that signals a price making an impossible claim. Eleven-plus years is an ordinary, un-dislocated figure — the buyback here is a support, not a coiled spring.

Dividend safety

The dividend is immaterial to the case. Salesforce initiated it in FY2025 and pays $0.44 a quarter after a 5.8% raise in February 2026 [18] — roughly a 1.0% yield, covered about nine times by free cash flow, with no downturn record to test because the payout is barely two years old.

Management credibility

The promise-versus-delivery record is the part that most cuts against the promotional-CEO exclusion. Salesforce runs a founder-CEO with a genuinely promotional register — "one of the best performances in software ever," the "SaaSpocalypse" framing [19] — which is exactly the pattern the framework treats as a red flag. But the exclusion is about big claims paired with repeated misses and no skin in the game, and on the checkable promises Salesforce has delivered.

Sources: Q1 FY2024 call [20]; Q4 FY2024 call [21]; Q4 FY2025 call [22]; Q4 FY2026 call [23].

Take the most material sample. After the January 2023 restructuring, management committed on the Q1 FY2024 call to a 30% non-GAAP operating margin by the first quarter of FY2025 and to "fully offset our stock-based compensation dilution through our share repurchases" [24]. It delivered a 30.5% full-year margin in FY2024, up 800 basis points [25], and 33% in FY2025 [26]; the share count did begin falling that same year. FY2026 revenue guidance of $41.0–41.3 billion closed at $41.5 billion [27]. Against that, one large forward promise is still open and unproven: a return to double-digit revenue growth in the second half of FY2027, hardened from a "possibility" to near-certainty across recent calls even as brokers press that Agentforce's usage metrics have not yet shown up in current remaining performance obligations. That is the credibility item to watch, and it belongs to the growth case rather than the balance sheet.

On skin in the game, Benioff beneficially owned 22,791,882 shares as of March 2026 — about 2.8% of the company, worth roughly $3.9 billion at the current price [28]. The proxy notes his holdings crossed Hart-Scott-Rodino thresholds through the vesting and continued retention of previously vested equity awards [29]. A founder with a multi-billion-dollar economic stake and a record of hitting his margin and revenue marks is not the promotional-executive-without-ownership pattern the framework excludes — the register is promotional, the delivery and the alignment are not.