Is the future of infrastructure open?

What it takes to build a data schema that will stand the test of time.

Without a doubt, the most enduring marks we’ve made on the world all take the form of infrastructure. Roads and bridges—not to mention temples, tombs, and towers—that were built centuries ago are still in use today. But when the details of their design and construction have been lost to time, engineers are left with the task of rediscovering those secrets, which can be a very tall order indeed.

To avoid repeating the mistakes of the past, we ought to build infrastructure today that future generations will still be able to understand and maintain. But therein lies the challenge, because the same feature that makes infrastructure such an impressive feat of engineering—its endurance—is also the biggest obstacle to its futureproofing.

How do we ensure that the engineering knowledge that went into building our roads, highways, bridges, dams, and utilities is still accessible a decade or a century from now?

Open infrastructure data

Data is as much a part of modern infrastructure as concrete or rebar, and not just in the initial design. When engineers set out to build a road or a bridge, they build a digital version of that asset first, and those digital counterparts may require their own maintenance throughout the asset’s lifecycle.

“What starts as a building information model then becomes a digital twin of a physical thing made of concrete and steel,” explains Greg Demchak, VP of emerging technologies at Bentley Systems. “That digital representation can be maintained through explicit geometry editing, or reality capture methods—such as photogrammetry, laser scanning, or Gaussian splatting—as well as IoT sensor data, photos, etc.”

All of this data is added on top of what’s initially created in engineering drawings, CAD models, and other tools. The result is an enormous amount of information that needs to be accessed and edited by multiple stakeholders: not only engineers but also surveyors, project managers, and construction crews, both today and over the asset’s entire lifecycle. “You can’t expect that the industry is going to work with a single vendor, a single file type, or a single established export type,” Demchak says. This is why Bentley is pushing for interoperability in infrastructure data through its software.

“Our platform is open to basically every CAD format out there as input or output,” Demchak says. “You can use other vendors, other products, other types of CAD software or different IoT sensors. They can all feed into this platform, and that’s what we mean by open.”

Whence openness?

On the surface, the value of open infrastructure data seems obvious, but there are also reasonable concerns that might be raised by the concept of openness in this context. One might worry, for example, that an open system is less secure than a closed one or that it might be more vulnerable to bad actors looking to damage or disable critical infrastructure. That scenario might be particularly concerning these days, given the availability of AI tools that could potentially make hacking into a digital twin easier than ever.

Demchak was unequivocal in rebuffing this worry, emphasizing the importance of fundamental security measures, such as two-factor authentication, while noting that Bentley’s platform being open doesn’t make it any less secure than its competitors. “Cybersecurity is a risk across the board,” he says. “It’s not specific to Bentley.”

Still, facts aren’t always sufficient to dispel this sort of objection, and so one might naturally wonder where the demand for openness really originates: in the users or the developers. Demchak is uniquely positioned to answer this question, since part of his role at Bentley involves working with users directly. “I really enjoy finding the users who are hungry to push the envelope,” he says. “I’m always working with engineering or construction or architecture firms who are willing to take that leap with us.”

Based on that experience, Demchak says that users don’t want the data they’ve created to be locked into a particular file format or the sort of proprietary framework that carries the risk of being inaccessible in the future due to licensing. “If you adopt open standards, even if you’re using USD, IFC, or glTF, at least those formats are being managed by third-party, vendor-neutral governance,” he explains. This is why an open, extensible schema can offer built-in futureproofing, evolving as new requirements and formats emerge.

There’s also an undeniable competitive advantage to open standards: the more they’re adopted across an industry, the more that industry is incentivized to participate in them and maintain them. “We believe in a governance model with an open approach because it reflects how we as a society want to maintain and manage our infrastructure,” says Demchak. “An open approach is actually an enabler for us as a society to deliver better, more resilient, higher-quality infrastructure.”

The future of infrastructure

Designing, building, and maintaining infrastructure requires a much more long-term perspective compared to most other fields of engineering. It’s one thing to make decisions about an automotive or aerospace product with a lifecycle measured in decades—or even a medical device that needs to last the lifetime of the patient. It’s quite another matter to make decisions on a project that could outlive your grandchildren. Taking that perspective and looking ahead, what should we expect to see in infrastructure in the near- and long-term?

To start with the multi-trillion-dollar elephant in the room, it seems practically inevitable that AI will play a larger role throughout the infrastructure lifecycle, from design and simulation through to construction and even maintenance. Regardless of the application, it’s clear that an open approach is much more conducive to bringing AI into infrastructure data than a closed one.

“If you have the data schema accessible through APIs and this open framework, that means the API can be documented,” Demchak explains. “It means there can be a Model Context Protocol interaction, so the AI can go in and actually understand the data that’s inside of a project, not just to extract information, but also to execute commands.”

What about the more distant future? What’s in store for infrastructure beyond LLMs?

“In the longer term, I think what’s going to be very interesting is when we start bringing in physical AI,” Demchak says. His examples here include drones that can do autonomous inspections, road machinery outfitted with sensors for autonomous roadwork, robots building prefabricated components, and autonomous trucks or tower cranes delivering those components.

What this means for the future of labour in construction, in roadwork, and on factory floors could well be one of the defining questions of our generation, but the full extent (not to mention the timeline) of this sort of radical economic and social change is still very much uncertain.

It’s been written that the past is a foreign country and it’s been said that the future is an undiscovered one. If that’s true, then perhaps we should see open infrastructure data as the map that enables us to navigate fluently between them.

Written by

Ian Wright

Ian is a senior editor at engineering.com, covering additive manufacturing and 3D printing, artificial intelligence, and advanced manufacturing. Ian holds bachelors and masters degrees in philosophy from McMaster University and spent six years pursuing a doctoral degree at York University before withdrawing in good standing.