Bentley Systems (Nasdaq: BSY) announced the winners of its 2026 Year in Infrastructure Awards on 7 October, from its Year in Infrastructure Event in Singapore, held 6–7 October. The scale is the easy part to report: more than 300 projects nominated by organisations in 53 countries, winners chosen by independent judges across 12 categories, and 16 further projects given Founders’ Honors selected by the company’s founders and executives. The release notes the programme has honoured infrastructure professionals for more than 20 years.
The release’s subheading sets the frame: “New cycle of AI-enabled innovation is helping engineers solve harder problems faster”. Awards coverage is usually where a trade publication stops reading. It is worth reading further, because the list underneath that subheading is a reasonably rare artefact — a field of several hundred real infrastructure projects, self-selected by organisations willing to be judged, with the winning ones named, categorised and titled by the organisations that delivered them.
Three of 28 name AI on the record
Across the 12 category winners and the 16 Founders’ Honors — 28 recognised projects in total — three carry AI, or a specifically named AI technique, in the project title or the award category:
- APD Global (Australia) won the Transmission and Distribution category for Clean Energy Link-North: AI-Driven Digital Delivery for Renewables.
- WSP Global Inc. (India) took a Founders’ Honor in Connected Data for its iTwin AI Project Lifecycle Intelligence Platform.
- Universidad de las Fuerzas Armadas ESPE (Ecuador) took a Founders’ Honor for Intelligent Leak Management with Autoencoders, in an education category Bentley frames around hands-on teaching using AI.
That count needs its method stated, because it is easy to over-read. It counts the appearance of AI or a named AI method in a project’s or award’s title — not how much AI any project used. Two further entries carry the word “Intelligence” without naming AI or a technique — the Asset Intelligence category under which Empresas Públicas de Medellín’s water digital twin sits, and CAPBP’s Urban Data Intelligence: A CDE for Climate Resilient Infrastructure, a common data environment — and this publication did not count either. A project can be built on machine learning and never say so in its name, and most of the 28 titles describe the asset or the delivery method rather than the toolchain. The number is therefore a measure of what the awards put on the record about AI, which is a useful thing for a reader tracking the category to be able to check, and not a measure of adoption.
The winners release claims no proportion. Bentley’s finalists release of 19 August did: “In a review of the 36 finalist projects, 24 finalists reported using AI in some form, while 22 showed strong or supporting evidence of a connected digital foundation.” That is a broader and more generous measure than a title count — self-reported use of any kind, across the full finalist field rather than the winners — and the two numbers should not be read as competing. A reader who wants Bentley’s own figure for how much AI is in this programme should use 24 of 36, not ours.
What the three do offer is specificity, which is scarcer in this category than enthusiasm. A transmission project in renewables, an owner-side lifecycle intelligence platform at a global engineering firm, and an autoencoder-based leak-detection teaching project are three distinct answers to where the technique is currently load-bearing. Two of the three sit in transmission and water.
The argument underneath was about openness, not models
Cate Lochead, Bentley’s chief marketing officer, opened with the conventional line: “Every year, our users impress us with incredible projects that advance the world’s infrastructure.” The part worth quoting is where she went next. This year’s finalists, she said, had gone “well beyond digitalizing traditional workflows.” Then the mechanism: “By connecting data across lifecycle phases and engineering disciplines, they used AI to unlock potential that was previously out of reach.” And then the argument: “That new cycle of AI-enabled innovation requires an open ecosystem, where infrastructure professionals can connect their data and preferred software to make better, faster decisions.”
That is a vendor making an interoperability argument from a position of considerable incumbency in infrastructure engineering software. It is also, read plainly, the same argument several unrelated parts of this market made in the last quarter. Revizto put it as pointing outside models at live project data on purpose. Arcadis and Autodesk put it as an MCP layer between a consultancy’s own tooling and a vendor’s assistant. The convergence is not coordination; it is what happens when every participant discovers that the useful models are general and the valuable data is not.
Note what the awards themselves do and do not establish. They recognise projects, judged on submissions, by panels Bentley convenes and — for the 16 Founders’ Honors — by Bentley’s own founders and executives. That is a legitimate and well-established format, and it is not a measurement of cost, schedule or asset-performance outcomes. Readers looking for the latter will not find it in an awards release, from Bentley or anyone else, and the release does not offer it.
Where the recognised work actually was
The geographic spread is the quieter finding. The 12 category winners sit in Portugal, India, Ireland, China, the Philippines, Poland, the Czech Republic, Australia, Indonesia, Brazil and the Netherlands. The Founders’ Honors add Colombia, the United States, Ecuador, the United Kingdom, Singapore, Italy, Norway and France. No organisation Bentley files under a United States country label took one of the 12 categories — though the Roads and highways winner, filed under Australia, is the “Torrens to Darlington Alliance (John Holland, Bouygues, Arcadis, Jacobs, Ventia)”, and Jacobs Solutions is a Dallas-headquartered, NYSE-listed company. American firms are inside the winning work; they are not the entities Bentley lists as winning it. Two of the 16 Founders’ Honors did go to United States organisations — Ampirical, for powering what Bentley’s list calls the Western Hemisphere’s largest data centre, and WSP USA for I-35 Capital Express Central – Downtown.
For a sector whose AI narrative is written largely from San Francisco, a list on which no US-labelled organisation takes one of the 12 judged categories is a reminder of where large civil programmes are actually being delivered — and it tracks the pattern this publication has seen in the deal flow, from Nemetschek’s heavy-civil acquisition of HCSS to Trendspek taking Australian condition-record software to North America and Europe.
A second pattern: water networks are over-represented for an asset class with one category to itself. Evides in the Netherlands won Water and Wastewater for Water Distribution Operational Digital Twins at Scale; Empresas Públicas de Medellín took a Founders’ Honor for its EPM Water Secondary Distribution System Digital Twin; Maynilad Water Services took one for scaling leak detection from district metered areas to system-wide analysis. That is three of the 28, in networks nobody can see without instrumenting them — four, if a reader also counts ESPE’s autoencoder leak project, which is filed here under education because that is the category Bentley awarded it in.
Digital twins are not confined to water on this list — Pinnacle Future Build won Cities and Facilities for a campus twin, and Digital Twin at Scale is a Founders’ Honors category of its own, taken by Almondz Global Infra Consultant for NAKSHA. The point is narrower: the places where the technique already has three separate recognised deployments are buried pipes and the grid, and that is a more checkable version of the “AI-enabled” claim than a subheading can carry.
What to watch
The useful question the list raises is one the awards cannot answer: whether the WSP entry — an owner-side, lifecycle-wide intelligence platform built on a vendor’s twin infrastructure — is the shape the next few years take at large engineering firms, or a well-executed one-off. Bentley has every commercial reason to want the former, which is precisely why its awards release is not the place to settle it.
What the release does settle is the framing. A long-established incumbent in infrastructure engineering software has decided its AI story is an openness story. Everyone selling into the same accounts should expect the data-portability question to be asked of them next.
