“Optimization Belongs To The Machine, Trade-Offs To The Engineer”: Takeaways From Bentley Year in Infrastructure 2026
Bentley Systems (Bentley) hosted its 2026 Year in Infrastructure (YII) Awards in Singapore in October, recognizing innovative projects spanning infrastructure, energy, buildings and cities. Competition this year was fierce: of more than 300 submissions from 53 countries, just 36 projects were shortlisted and 12 took home awards.
What can we learn from this year’s finalists?
The award presentations demonstrated how organizations are applying Bentley’s solutions to real-world challenges, with Verdantix finding that the most significant value came from connected toolchains. The handover from design to construction is where infrastructure projects typically lose time and data, and this risk grows with scale. To tackle this, the team of five specialist subcontractors working on Jakarta International Stadium kept every team – with up to 5,000 workers on site – aligned by working from a federated model as a single source of truth. Structural analysis in STAAD.Pro informed fabrication data and 4D sequencing in SYNCHRO, enabling the team to digitally rehearse the lift of a 3,900-tonne retractable roof and deliver the stadium within 36 months.
Connected models are also reaching beyond the project team. SUDOP PRAHA, a Czech design and engineering firm working on the high-speed railway (HSR) development in Central Bohemia, used ProjectWise- and OpenRail-enabled design partners to collaborate around a shared model. This was then integrated with GIS data in OpenCities Planner and published online, allowing municipalities to explore the proposed route’s impact on water-protection areas, land access and noise in its real geographic context.
Echoing Singapore’s ambitions for climate-resilient infrastructure, several finalist projects focused on mitigating environmental risks. Aboitiz Renewables’s GeoShield, for example, uses Bentley’s iTwin platform alongside third-party geotechnical applications to predict landslides in the Mindanao area, avoiding potential losses of approximately $6.8 million. iTwin was the critical enabler, bringing lidar, drone photogrammetry, subsurface geophysics, GIS analysis and field observations into a single digital twin, so engineers could validate AI-flagged risk areas against site-specific ground conditions before failure occurred.
Engineers keep the final call
One of the headline statistics from the event was that 69% of finalist projects used AI in some form. More telling than the level of adoption, however, was how AI is being applied. Continuing a trend seen at YII 2025, clients are building their own AI solutions to automate workflows, and Bentley’s engineering software is serving as the core data foundation. For example, in its Porto Metro expansion project – centred on the design of 5km of tunnels beneath a historic city – Ayesa deployed AI agents to validate geotechnical simulations against design criteria, testing over 100 scenarios rather than just a handful. This cut design costs by 30% and saved 4,500m³ of concrete.
Yet automating analysis does not remove the need for engineering judgement. Bentley CEO Nicholas Cumins captured this distinction: “Optimization belongs to the machine. Navigating the trade-offs belongs to the engineer”. While AI excels at iterating and generating options, responsibility for design decisions remains firmly with engineers. With simulations now fast-tracked, jury members stressed that critically evaluating results is becoming a core skill for junior engineers. On infrastructure projects that cannot afford to fail, trust in AI is still being earned and the engineer remains the essential layer of assurance, bringing the context each project requires and confirming outputs are accurate and fit for purpose.
The pace of AI progress is already reshaping what engineers can achieve. Bentley illustrated this by revisiting the design of London’s ‘Gherkin’, which won at the inaugural YII Awards in 2004, and recognized Foster + Partners’s use of MicroStation to generate its complex geometric models. When Bentley asked OpenAI’s Astra to recreate the building’s shell in the same application using only reference images, the model grasped the 3D space and modelled its details with minimal human prompting.
This shift also helps explain why Bentley is prioritizing openness in AI development. Rather than competing to ship the largest number of agents, the vendor sees value in connecting its ecosystem with the tools customers already use. To that end, the vendor announced continued support for the Model Context Protocol (MCP), the open standard that lets AI agents connect to external data and applications. For engineering firms already building in-house AI tooling, this lowers the barrier to plugging Bentley data into their own agents and workflows – positioning Bentley as an open data foundation rather than a closed AI platform.
For further insight on the solutions highlighted at Bentley’s YII 2026 – as well as a broader view of offerings available on the market – please read:
About The Author
Annemarie Briggs
Industry Analyst
Sayanh Alam
Industry Analyst
