The Radar: Accenture

Accenture is not defending the billable hour. It is trying to get out from under it before the price of an hour falls. Read as a set, the 28 disclosed moves between Q1 2024 and Q3 FY2026 describe one manoeuvre: convert a labour arbitrage business into a platform and intellectual property business while the balance sheet is still strong enough to pay for it.

Section one

What did Accenture buy, build and kill

The ledger below records every disclosed acquisition, investment, partnership, platform launch and wind-down Global AI Forum could date between Q1 2024 and Q3 FY2026. Undisclosed moves do not appear, so the count is a floor and not a total. Six of the 28 carry a disclosed value. The rest were structural, and the structural ones are the more revealing half.

Scroll the table sideways to see every column.

Source: Accenture SEC filings, newsroom releases and earnings materials, Q1 2024 to Q3 FY2026, compiled by Global AI Forum 7 September 2026. Value is disclosed transaction value, disclosed investment commitment or disclosed restructuring charge, as applicable, and the three figures are not additive. Undisclosed moves are excluded.

Section two

Where the 28 moves point

Grouped by the capability each move buys, the 28 fall into six vectors. Two of the six, the agent platform and model layer neutrality, account for 12 of the 28. Neither of those vectors adds a single billable head.

Section three

How the org chart changed, and when

Accenture publishes little about its intent that is as informative as its reorganisation releases. Two of them, thirteen months apart, produced the current shape. The naming is the tell. The seven client-facing units are named after the buyer's departments. The three internal units are named after Accenture's own capability build.

1. March 2024Udacity acquired, LearnVantage announced

   The learning platform is backed by a stated $1 billion investment. Source: Accenture newsroom, 2024.

2. October 2024AI Refinery launched with NVIDIA

   Built on NVIDIA AI Enterprise. Accenture had filed 55 patent applications across 10 countries on it as of mid 2025. Source: Accenture newsroom, October 2024.

3. September 2025Five service lines collapsed into Reinvention Services

   Strategy, Consulting, Song, Technology and Operations were merged into one unit under a first Chief Services Officer. Source: Accenture, Fiscal 2025 fourth quarter results, 25 September 2025.

4. February 2026Internal AI tool use tied to promotion

   Associate directors and senior managers were told regular use of internal AI tools is a requirement for promotion into leadership roles, and usage began to be tracked. Source: reported coverage, February 2026.

5. March 2026Seven Reinvention Partners and three Reinvention Engines

   Effective 31 March 2026, with a Client Success wrapper covering commercial terms, offerings and delivery quality. Source: Accenture, Reinvention Services Leadership, 13 March 2026.

6. June 2026Dragos, runZero and NetRise agreed, Accenture Edge launched

   The security transaction and the mid-market business were announced five days apart, on 18 and 23 June 2026. Source: Accenture newsroom, June 2026.

7. August 2026McCoy, Comware and General Robotics

   Two acquisitions to feed Accenture Edge, on 25 and 27 August 2026, and one robotics investment. Source: Accenture newsroom and reported coverage, August 2026.

Reinvention Partner

One of seven client-facing Accenture units created in March 2026: Cybersecurity, Digital Core, Finance, Industry and Enterprise, Song, Supply Chain and Engineering, and Talent. Each maps to a department inside the buyer rather than to a consulting discipline.

Reinvention Engine

One of three internal Accenture units created in March 2026: AI and Data, Industry and Process, and Technology. Engines build the skills, methods and tooling that the seven Partners then sell.

Advanced AI, as Accenture defines it

Generative AI, agentic AI and physical AI only. The definition excludes data work, classical AI, robotic process automation and AI used in the delivery of Accenture's own services, so the reported figure understates AI-attached revenue.

Section four

Why a services company bought security software

In June 2026 Accenture agreed to acquire a majority stake in Dragos, valued at about $3.25 billion, plus all of runZero and NetRise, at a combined enterprise value of about $4.18 billion. Accenture is buying a majority of Dragos rather than all of it, which preserves the vendor-neutral market position that gives the asset its value.

Source: Accenture, Strengthening Critical Infrastructure Defense, 18 June 2026, and reported coverage of the transaction, June 2026.

The arithmetic Accenture quotes is a move from a services market it already leads into a software market several times larger: about $7 billion in operational technology cybersecurity services against about $27 billion in operational technology security software in 2026, projected toward $59 billion by 2031 at roughly 16 per cent compound annual growth. Read alongside ServiceNow's $7.75 billion acquisition of Armis, this is two firms building asset-centric security platforms across information technology, operational technology and connected devices.

The strategic content is not the security market. It is the revenue type. A services firm that buys recurring revenue software is telling you what it thinks of its own revenue type.

Section five

Is AI the business, or the door into the business

Most commentary treats Accenture as an AI company whose AI revenue is being disrupted. It is not yet an AI company by revenue. Advanced AI was $2.7 billion of $69.67 billion in FY2025. Cumulatively, from the metric's launch in Q3 FY2023 to Q1 FY2026, Accenture reported roughly $11.5 billion of advanced AI bookings across about 11,000 projects and about $4.8 billion of revenue across roughly 1,300 clients.

The commercially important number is not the AI line. It is the pull-through. Accenture has stated that one in every two projects in generative, agentic or physical AI drags significant data work behind it, drawn from more than 6,000 advanced AI projects in FY2025 alone. That makes AI a demand generator for the technology and data core rather than a line item competing with it. It also explains why Accenture is comfortable with AI at under 4 per cent of revenue.

Section six

Which number breaks first

In Q3 FY2026 the two halves of the business moved in opposite directions in the same quarter. That divergence, not the AI line, is what repriced the stock.

AI is disrupting demand across consulting and managed service.

Bloomberg Intelligence, following Accenture's Q3 FY2026 results, June 2026.

Managed services is the recurring, visible, headcount-priced half of the book. A 15 per cent decline in its bookings in Q3 FY2026 is exactly the shape you would expect if agentic delivery had started to reduce demand for outsourced operations. It is also exactly the shape you would expect from contract timing and a soft macro cycle. Nothing in the current disclosure separates the two, and Accenture reports Q4 and full-year FY2026 in late September 2026.

The market has already taken a view. Accenture shares entered 2026 near $259 and traded in the $128 to $130 range by late June 2026, including an 18 per cent single-day fall on 18 June 2026. The forward price to earnings multiple compressed from roughly 24.5 times in May 2025 to about 12.7 times by late August 2026. Earnings held through that period, so the multiple did nearly all of the damage.

Section seven

What a buyer should do before the next renewal

The practical consequence for a buyer is an incentive change, not a technology change. A supplier moving toward platform revenue is a supplier that wins by removing hours. Most existing contracts were written when it won by adding them.

• Established: your integrator's structure now rewards platform and product revenue over delivery hours, on the evidence of 28 disclosed moves between Q1 2024 and Q3 FY2026.
• Ask for the data pull-through ratio on your own AI programmes over the last four quarters. If AI work is not dragging data work behind it, the foundation problem is not being addressed on your estate.
• Ask which managed services towers in your current contract year are priced on headcount and which on outcome, tower by tower, with the number for each.
• Ask what happens to your rate card if delivery hours on a tower fall by 30 per cent. The answer is the whole negotiation.
• Where your supplier now owns software rather than reselling it, price the engagement as a product roadmap with a version cadence, not as a service.
• Ask for advanced AI revenue attributable to your account. No large integrator currently discloses this, and Accenture stopped disclosing it at company level after Q1 FY2026.

Section eight

What we expect next

Four expectations follow from the ledger. Each is stated so that it can be checked.

1. Other large integrators buy recurring revenue software

   Accenture has established that a services firm can buy an ARR line without losing the neutrality that made the asset valuable. Global AI Forum expects at least two further platform-scale acquisitions by large systems integrators to be announced before the end of calendar 2027.

2. Outcome pricing moves from exception to renewal default

   A Chief Commercial Officer whose stated mandate includes win-win commercial models exists because time and materials pricing becomes a liability once hours fall. Watch renewal terms rather than press releases.

3. Separate AI disclosure disappears across the sector

   Accenture reported advanced AI as a defined category for eleven quarters and then stopped. The justification, that AI is now embedded everywhere, is credible and unfalsifiable. Expect peers to adopt the same position.

4. Revenue per person replaces headcount as the growth signal

   Accenture employed about 799,000 people as of Q3 FY2026, up from about 779,000 in August 2025, while stating an intent to reshape the mix rather than simply grow it. Headcount growth and revenue growth are decoupling, and the ratio is the number to track.

Global AI Forum · GAIF-RA-001 · v1.0 Research current as of 7 September 2026 gaiforum.com