Chapter 1

The 800,000-Person Payroll

Accenture is one of the largest companies in the world that owns almost nothing. In the fiscal year ended 31 August 2025 it booked $69.7 billion of revenue, employed approximately 779,000 people, and served roughly 9,000 clients — three-quarters of the Fortune Global 100 and 500 — from an Irish domicile, with its Class A shares listed on the New York Stock Exchange [1]. To produce all of that, it carried $1.57 billion of net property and equipment [2] — roughly one dollar of fixed plant for every forty-four dollars of sales. There is no factory here, no fleet, no store network, no reserve base. What Accenture sells is the time of its people, organized around a client list, and everything else about the business follows from that single fact.

That makes it an unusually clean company to learn, because there is only one thing to understand. This chapter draws the map: what the enterprise physically is, what its people do and for whom, the shelf of rivals it competes against, and why — after a decade of quiet reliability — it arrived in front of investors this summer with the steepest single-session share decline in its recent history.

Sources: FY2025 Annual Report (Form 10-K), Item 1 Business [3] and Note 16 Segment Reporting [4]; capex-to-revenue derived from reported cash flows.

A company that is mostly a payroll

The whole cost structure fits on one line. In its FY2025 segment note, Accenture reconciles revenue to operating income with only three recurring cost categories: payroll of $45.7 billion, non-payroll costs including subcontractors of $11.6 billion, and depreciation and amortization of $1.5 billion [5]. Payroll alone is 65.6% of revenue and 76.8% of total operating expense. The company says as much in plain words: cost of services "is primarily driven by the cost of people serving our clients" [6].

Source: FY2025 Annual Report (Form 10-K), Note 16 Segment Reporting [7].

Read across the three years, the striking feature is how little the ratios move. Payroll has sat at roughly 65-67% of revenue and the other cost lines within a point of themselves, while the operating margin has held near 14-15%. A business with a factory would show operating leverage — fixed costs spread over more units as volume grows. Accenture shows almost none, because its costs are not fixed. They are people, hired and paid roughly in proportion to the work, so the model is less about spreading a cost base than about holding a set of ratios steady while the whole thing scales. Managing this company means managing the gap between what it charges for an hour and what that hour costs.

The other side of "owns almost nothing" is that growth needs almost no capital. Capital expenditure has fallen from about 1.4% of revenue a decade ago to 0.9% in FY2025, and net property and equipment barely changed — $1.53 billion in FY2023, $1.57 billion in FY2025 — while revenue grew 8.7% over the same span [8]. Asia Pacific alone produced $10.0 billion of revenue on $239 million of segment net assets.

Source: derived from reported cash flows, FY2016–FY2025 10-Ks.

The practical consequence is that nothing on the asset side constrains how fast Accenture can grow, and nothing on the asset side is where its money goes. When the company wants to add a capability, it does not build a plant; it hires, or it buys a firm. That is why the reinvestment story, when it comes, is a story about acquisitions and payroll rather than capital spending — a thread the later chapters pick up.

What the people do, and for whom

Accenture organizes the same pool of people two ways. By type of work, it splits into Consulting — advising, designing and building, recognized as the work is delivered — and Managed Services, where it runs a client's process or system on an ongoing basis under multi-year contracts. In FY2025 the two were almost exactly even at $35.1 billion and $34.6 billion, a deliberate drift toward the annuity half of the book from a consulting-weighted mix a few years earlier [9].

By industry group, the same people are pointed at five end-markets. Products — consumer goods, retail, industrial, travel and life sciences — is the largest at $21.2 billion, followed by Health and Public Service, Financial Services, Communications, Media and Technology, and Resources [10].

Source: FY2025 Annual Report (Form 10-K), Note 16 Segment Reporting [11].

Geographically the business runs through three reportable markets — the Americas at 50% of revenue, EMEA at 35%, and Asia Pacific at 14% [12]. Country concentration is milder than the geography split suggests: the United States was 45% of consolidated revenue in each of FY2023, FY2024 and FY2025, and no other country reached 10% [13]. This breadth — five industries, three regions, one large but sub-majority country — is the reason no single vertical shock moves the whole company, and it is the first thing that separates Accenture from a specialist.

There is one exception the company itself carves out. Its U.S. federal work runs through Accenture Federal Services, which represented about 8% of total revenue in FY2025 — but 36% of Health and Public Service and 15% of the Americas [14]. This is the one demand block whose budget is set by government procurement rather than by client spending, and in fiscal 2026 it has behaved differently from the rest. Management has attributed roughly one point of the year's revenue-growth shortfall to the federal business in every quarter — enough that it now guides the whole company two ways: 3-4% local-currency growth including federal, or 4-5% excluding an estimated one-point federal drag [15]. One 8%-of-revenue relationship has, for four straight quarters, been worth a full point of company growth.

A last feature to hold in mind: because revenue is billed hours, it swings with the calendar. Quarterly gross margin moves several points across a year on billable days and holidays alone, so any single quarter has to be read against the annual run-rate rather than the one before it.

The shelf it competes on

Accenture is the largest single revenue pool in its industry by a wide margin. Its $69.7 billion dwarfs the businesses usually named alongside it: IBM's Consulting segment at $21.1 billion and Cognizant at $21.1 billion are each about 30% of Accenture's scale [16] [17].

Sources: Accenture FY2025 10-K [18]; IBM Consulting segment [19]; Cognizant FY2025 10-K [20].

Scale, though, is not the same as a fortress, and the reason is structural. What makes this industry unusual is that its barrier to entry is not capital. When Accenture lists its competitors, it names an entire ecosystem: large multinational IT-service providers, the services arms of technology vendors, offshore providers "particularly in India," accounting firms and consultancies, advertising holding companies, engineering-services firms, technology start-ups — and, tellingly, "in-house IT departments of large corporations that use their own resources rather than engage an outside firm, such as global capability centers" [21]. The company adds that clients "typically retain us on a non-exclusive basis." Two things follow. First, one of the named competitors is the client itself: a global capability center is a captive offshore unit a company builds instead of buying the service — a demand leak, not just a rival. Second, the competitor set is not Accenture's own framing. Cognizant's 10-K opens its direct-competitor list with Accenture and separately flags in-house resources "such as GCCs"; IBM's 10-K names Accenture, Capgemini and India-based providers as its consulting competitors [22]. Three independent filers describe the same crowded field, and nothing physical stops entry into it.

What the incumbents claim instead is a barrier made of relationships and breadth. Accenture has partnered with 195 of its top 200 clients for ten years or more, counts 305 of them as "Diamond" clients, and is the number-one partner to all ten of its largest technology-ecosystem partners [23]. Whether that incumbency is a genuine moat or simply a good position in a competitive industry is a question the next chapter takes up; for now it is enough to see that the defense is a client list, not a cost curve.

And the whole shelf is planning for a slow year. Every named competitor, and Accenture itself, is guiding low-single-digit constant-currency growth for the year ahead — even as the third-party market they draw from is described as growing faster. TCS's annual report, citing Gartner, puts IT services on track to grow 6.8% and surpass $1.87 trillion in 2026; Infosys cites IT services growing 4.6% in calendar 2025 [24] [25]. The vendors themselves plan for less.

Sources: Accenture Q3 FY2026 release [26]; IBM Q2 FY2026 call [27]; Infosys Q1 FY2027 call [28]; HCLTech Q4 FY2026 call [29]; TCS FY2026 as reported.

That the market pool grows mid-single-digit while every large vendor guides low-single-digit — and TCS's own constant-currency revenue fell in FY2026 — is the industry's central unresolved arithmetic. Either spending is flowing somewhere other than the named vendors, or the price of the work is falling faster than the volume is rising. Either way, the growth Accenture reports has to be read against a shelf where the base rate is now roughly flat to slightly up.

Why this is a story now

For most of the last decade, Accenture was the sort of company investors bought precisely because it did not surprise them. That changed in one session. On 18 June 2026 the company reported its fiscal third quarter — and the shares fell 17.97% that day, from $156.01 to $127.98, on about eight times their normal trading volume. It was the steepest single-day decline in the price record available for the company.

What makes the reaction worth an entire report is that the report behind it was, on its face, fine. Accenture beat consensus earnings, held its free-cash-flow guidance and raised its capital-return plan to at least $9.5 billion; the only guidance line it cut was the top of its local-currency revenue range, from 3-5% to 3-4% [30]. What appeared to trouble the market sat lower in the release: new bookings of $19.32 billion, down 3% in local currency, at a book-to-bill of 1.0 — the order book, not the current quarter [31]. The arithmetic of that repricing, and what the recovered price now embeds, is the closing chapter's subject. The point here is only that the market stopped treating this company as reliable, and did so over the durability of demand rather than the level of this year's earnings.

The reason a single soft bookings line could do that is everything the map has just shown. This is not a business with many independent variables. It is a headcount pyramid and a client list: $69.7 billion of revenue produced by about 800,000 people and $1.57 billion of property, in an industry with no capital barrier, at the moment its own peers are quantifying how much cheaper artificial intelligence is about to make the work. And it sits at the center of the report's organizing question. Accenture states in its own 10-K that it cannot measure how much of its revenue growth is price versus volume — the disclosure does not contain the variable that matters most — precisely as named rivals put numbers on AI-driven price deflation in the same book of work. Around that unmeasured core sits a decade-old promise to expand margin ten to thirty basis points a year, kept every year in the company's adjusted presentation but not in its statutory one, on which management is graded and paid; and a fiscal-2026 plan that commits roughly 1.8 times free cash flow to buying growth and buying back stock at the same time. Each of those is a later chapter.

These threads are not separate. Because the entire enterprise is the price and cost of an hour of labour, each one comes back to that same hour — what it sells for, what it costs to produce, and how long the price can hold. That contest — who actually sets the price of the unit of work, and what happens to it as the technology Accenture sells makes the unit cheaper to make — is where the story goes next.