The Buyback Bet

The Buyback Bet

The one promise Gartner kept while the operating model slipped was the promise to shrink the share count, and in FY2025 it kept that promise harder than ever — a record $1,991.1 million of stock repurchased, more than the company earned in cash that year, part-funded by a fresh $800 million bond, struck as the stock sat about three-quarters below its high [1]. To judge whether that was cheap capital well spent or a levered position in a still-falling asset, a reader first has to separate what the machine actually earns from what the income statement reports — because for three straight years those two numbers have diverged by hundreds of millions.

One channel for all the cash

Gartner does one thing with its surplus: it buys its own stock. There is no dividend. Acquisition spending over the last three years totals under $10 million — nothing in FY2025, $2.0 million in FY2024, $3.8 million in FY2023 — the only deal a $7.9 million tuck-in of a Czech sales agent, against which the company was actually a net seller, disposing of TalentNeuron in 2023 for about $161.1 million [2]. Capital expenditure runs near $115 million a year, under 2% of revenue [3]. The customer-funded, asset-light model leaves almost every dollar of operating cash flow discretionary, and management sends essentially all of it to the same place.

The scale of that single channel is easiest to see against the cash the business throws off. Across FY2021–FY2025 Gartner generated roughly $6.34 billion of operating cash flow and spent about $6.03 billion buying back stock — a payout of roughly 95%, with the balance sheet, not retained cash, absorbing the difference in the heaviest years.

Sources: FY2025 10-K, Consolidated Statements of Cash Flows [4]; operating cash flow FY2021–FY2022 per company filings, as reported.

The authority behind the program is standing and refreshed continuously: a $1.2 billion authorization in May 2015, an incremental $5.8 billion added between February 2021 and September 2025, and a further $500 million in January 2026, with $745.3 million still available at year-end. The 10-K states plainly that repurchases "will be funded by cash on hand and borrowings" [5]. Borrowing is not a backstop here; it is a named funding source, and in FY2025 the company used it.

The record year

FY2025 was the year the policy went to the balance sheet. Repurchases of $1,991.1 million exceeded the $1,290.4 million of operating cash flow, and after $115.1 million of capex the free cash flow the business actually produced was about $1,175 million — so the buyback ran roughly $816 million ahead of internally generated cash [6]. The gap was bridged by issuing $799.9 million of new long-term debt — the only such issuance in the five-year window — which lifted net debt from $526.8 million to about $1.26 billion [7].

The residual landed in the equity account. Because repurchases are booked into treasury stock against retained earnings, a record buyback into a shrinking net-income year drained book equity: total stockholders' equity fell from $1,359.2 million to $319.9 million in a single year, against a balance sheet carrying $6.5 billion of revenue and $3.0 billion of long-term debt [8].

Sources: FY2025 10-K, Consolidated Statements of Cash Flows [9] and Consolidated Balance Sheets [10]; free cash flow derived from reported operating cash flow less capex.

What makes this the year's defining allocation decision is the price. The open-market repurchases in the fourth quarter of 2025 were executed at an average of $239.06 per share [11] — itself less than half the three-year high — and by mid-2026 the stock changed hands near $140, roughly 40% below what management paid. The same decision carries two readings: capital deployed at a price less than half its recent peak, or capital committed to an asset that kept falling after the purchase. Which reading holds depends on whether the drawdown was mispricing — the matter Priced For Impairment takes up, not one this chapter can settle. What this chapter can settle is what the shares are a claim on, and that is where the reported numbers mislead.

What the cash actually is

The reported bottom line is a poor guide to Gartner's earning power, because a large non-operating item lands in each of the last three years. FY2023 pre-tax income carried a $135.4 million gain on the TalentNeuron divestiture; FY2024 carried a $300.0 million gain on event-cancellation insurance claims and, below the operating line, roughly $161.9 million of net tax benefits from a December-2024 intercompany transfer of intellectual property — a benefit the cash-flow statement shows was almost entirely non-cash; and FY2025 absorbed a $150.0 million goodwill impairment [12] [13]. The impairment fell entirely on the Digital Markets reporting unit — the same business carried, then sold as non-core in early 2026 — after its long-term forecast was cut, and it is non-cash [14].

Stacked together, those items whipsaw the EPS line in a way the operations do not justify. Diluted earnings per share ran $11.08, then $16.00, then $9.65 — and the headline story of FY2025 is a 42% net-income collapse, from $1,253.7 million to $729.2 million [15]. Strip the two items that sit inside operating income — the FY2023 divestiture gain and the FY2025 impairment — and operating income was roughly flat to modestly higher across the three years, near $1,100 million, $1,156 million and $1,176 million. The engine did not fall 42%; the reported total did, because FY2024 was flattered by non-recurring gains and FY2025 was charged with a non-cash write-down.

Source: FY2025 10-K, Consolidated Statements of Operations [16]; MD and A Results of Operations for the tax benefit [17]. Operating income adjusted for the divestiture gain and impairment that sit within it; derived from reported figures.

Cash tells the cleaner story. Operating cash flow exceeded net income in every year of the window and fell only 13% in FY2025 — from $1,484.9 million to $1,290.4 million — against the 42% drop in reported profit [18]. The collection side is clean: fees receivable of $1,684.5 million actually fell year over year and were a $38.0 million source of cash, so there is no receivables build-up flattering revenue ahead of collections [19]. One honest caveat sits on the conversion ratio itself: operating cash flow at 1.77 times net income looks pristine partly because the $150 million impairment depressed the net-income denominator without touching cash. The quality signal is real, but the flattering arithmetic should be named.

The one accounting position worth flagging is the tax line the FY2024 benefit came from. The auditor's sole critical audit matter was the company's unrecognized tax benefits on transfer pricing, which had grown to $301.6 million by year-end from $257.5 million and $148.4 million in the two prior years [20]. The same intercompany-IP structure that lifted FY2024 earnings is the one the auditor singled out — a reminder that a chunk of the reported tax advantage remains contestable.

The engine's own tell

The cleanest forward signal in the statements is not on the income line at all — it is in how much cash the subscription base prepaid. Gartner's clients pay in advance, so the year-over-year change in deferred revenue measures whether forward billings are still growing faster than the revenue being recognized. For years that delta was a large tailwind, adding cash on top of earnings. In FY2025 it reversed.

Sources: FY2025 10-K, Consolidated Statements of Cash Flows for FY2023–FY2025 [21]; FY2023 10-K, Consolidated Statements of Cash Flows for FY2021–FY2022 [22].

The tailwind did not simply switch off; it decayed in step with the deceleration the earlier chapters traced. Deferred revenue added $324.1 million of cash in FY2021, when contract value was compounding in the mid-teens, then $297.1 million, then $169.9 million and $181.2 million — before turning to a $41.9 million drag in FY2025 [23] [24]. Total deferred revenues on the balance sheet rose only to about $2.81 billion, up a slim 1.8%, while recognized revenue grew nearly 4% — billings no longer outrunning the book [25]. This is the same flat contract value showing up in the cash statement one line at a time.

The timing matters for the bet. The working-capital engine subtracted about $223 million from cash generation in FY2025 versus what it contributed the year before — in the very year management chose to repurchase a record amount of stock and borrow to do it. The prepayment cushion that historically let cash conversion run above earnings is, for now, gone; forward cash generation leans on the recurring base and on contract value re-accelerating, not on the tailwind that padded the last several years.

Where the bet stands

Assembled in one place, the decision has a clear shape. Gartner returns essentially all its free cash through one instrument, at no dividend and with M&A dormant. In FY2025 it stretched that instrument past internally generated cash and onto the balance sheet, thinning equity to $320 million and lifting net debt while buying at an average price the market subsequently marked down by roughly 40%. The reported earnings the shares are a claim on are cleaner than the −42% headline suggests — operating income roughly flat, cash conversion still strong — but the customer-funded tailwind that made that cash conversion exceptional has reversed, so the margin of comfort funding future repurchases is thinner than the multi-year record implies.

Whether the record buyback compounds value per share or merely concentrates a still-impairing position depends on the price management paid, which this chapter deliberately leaves open. Management has already committed its capital to one answer — a leveraged, all-in position taken on the view that the drawdown was mispricing. What the market is actually paying, what that price implies about the business, and the concrete tests that would prove the view right or wrong are where the story goes next.