Accenture launched Accenture Construct on 23 September 2026, a new global business aimed at helping project owners “plan, deliver and optimize” large infrastructure and capital projects — the release names airports, power grids, data centres, rail networks and advanced manufacturing facilities.
The unit is not a new hire or a new product line. It is a renaming and consolidation of a practice that already exists at scale. Accenture says its capital projects business has grown fourfold over three years across strategic advisory, engineering, project delivery and technology services, and that Accenture Construct brings together more than 5,000 practitioners across the Americas, EMEA and APAC. It puts the addressable market for integrated owner-side capital project services at $260 billion today, growing to $348 billion by 2030 at a rate the release gives as 7.5% a year; the release footnotes that figure to a study it commissioned from an unnamed analyst firm. Adam J. Shaw is named as chief executive of Accenture Construct.
Those are Accenture’s figures, presented as the company presents them.
Eleven Acquisitions, Read as a Map
The most informative part of the announcement is the list it does not dwell on. Accenture says the unit unites capabilities built “through organic growth and a series of acquisitions,” and that its 5,000-plus practitioners come “many from specialist acquisitions including” eleven named firms: Alfa Consulting (Spain), Anser Advisory (US), Arca (Spain), BOSLAN (Spain), Comtech Group (Canada), Fibermind (Italy), Greenfish (Belgium), IQT Group (Italy), Orlade Group (France), Soben (UK) and Verum Partners (Brazil). The wording is worth keeping: “many from” and “including” both leave room, so eleven is a floor rather than the whole account. Separately, the release says Accenture Construct will work with DLB Associates, Accenture’s majority-owned joint venture, which continues to provide data centre lifecycle services to US clients.
Read as a shopping list, that tells you what Accenture thought it was missing, and on our reading of the firms’ own descriptions it is not software. They cluster around project controls, cost management, quantity surveying, owner’s representation, engineering and programme advisory — the disciplines that sit between a client with capital and a contractor with crews.
Two of them show the pattern clearly. Soben, a Glasgow-headquartered construction consultancy of around 250 people that Accenture agreed to buy in March 2025, does project management consultancy, scheduling, project controls, and cost and commercial management, with a stated specialism in data centre development. Anser Advisory, acquired in 2023, is described by Accenture as an advisory and management company for US infrastructure projects, working alongside engineering and construction teams on large programmes. This is the acquisition of delivery capability, in the specific sense of people who can hold a schedule and a budget on someone else’s behalf.
The release’s named engagements point the same way: Vale, the Florida Governmental Utility Authority, and Metrolinx GO Expansion. Mining, water utility, and commuter rail — all owner-side, all programme-scale, and none of them a licence sale. The Vale engagement is described as delivered “through digital engineering, AI and integrated execution”: Accenture is being paid to run the portfolio, not to supply a product for someone else to run it with.
Julie Sweet, Accenture’s chair and chief executive, in the release: “Global demand for critical infrastructure capital programs is accelerating at the same time as AI is fundamentally changing how projects are planned, delivered and operated.”
Shaw, in the same release: “Capital projects haven’t lacked expertise; they’ve lacked alignment on shared success outcomes.”
What AI Is Actually Doing in This Announcement
The AI claim in the release is specific enough to be worth stating precisely, and modest enough to be worth not overstating. Accenture describes AI-enabled workflows that shift project delivery from reactive management toward predictive decision-making, combined with a data foundation giving a single operational view across the delivery ecosystem.
That is a programme-controls argument, not a jobsite-autonomy argument. The announcement does not mention robots, robotics, machine control, drones or computer vision anywhere in its text; the AI it describes is “AI-enabled workflows” that “shift delivery from reactive management to predictive decision-making.” It is about the owner’s view of a portfolio of projects — where the schedule is drifting, where the cost is going, and which of forty concurrent workstreams is about to become the critical path. The unit of analysis is the programme, not the pour.
This matters for reading the competitive set, and here the analysis is ours rather than Accenture’s. A business built to sit next to the owner, holding the data and the predictive model, is not primarily competing with Procore or Autodesk, whose products sit with the contractor and the designer. It is competing with the engineering and programme-management majors — the AECOMs, Jacobses, Turner & Townsends and Mace consultancy arms — and, increasingly, with those firms’ own AI ambitions.
That contest is already visible in this newsroom’s coverage. AECOM paid $390 million for Consigli and an AI that claims to automate a large share of engineering work. Arcadis ran 150 engineers through Nomic’s platform for six months and then invested. The pattern in both cases is a professional services firm concluding that the AI has to be owned rather than licensed, because the thing being automated is the billable work itself.
Accenture is arriving at the same conclusion from the opposite direction. It does not have a century of engineering practice to protect. It has a technology and consulting business with roughly $70 billion in FY25 revenue and around 799,000 people, per the release’s own boilerplate, and it is buying the domain expertise it lacks while carrying the systems-integration capability the engineering firms are trying to build.
The Timing Is the Data Centre Buildout
The phrase in Accenture’s own headline is “global investment supercycle,” and the demand story underneath it is not speculative.
This newsroom has spent much of 2026 covering the consequences of one segment of it. Buildots raised $130 million on the strength of a customer list drawn from the contractors building data centres. OpenSpace has documented more than 1,000 data centre projects, a figure that says a good deal about where construction AI is actually being deployed. The capital going into compute has to become buildings, substations and transmission, and someone has to programme-manage it.
What is not yet established is whether a consultancy is the right shape of organisation to do that better than the incumbents, or whether “fragmented capital project delivery” is a problem that yields to better information at all. Cost and schedule overruns on megaprojects are among the best-documented findings in the empirical literature on construction — the body of work associated with Bent Flyvbjerg is the standard reference — and they have survived several previous generations of improved project controls. Accenture’s argument is that AI changes the economics of prediction. That argument is plausible and, at launch, untested.
The near-term tell will be mundane: whether Accenture Construct wins owner’s-representative mandates on named programmes against the engineering majors, and whether the eleven acquisitions operate as one business or as eleven businesses under a shared logo. The second question is the harder one, and it is the one every consolidation of professional services firms has had to answer.
