‘Digital twins’ could revolutionize planes, cars and hearts

As the conflict in Ukraine underscores, the blistering speed of software and data, not industrial-era hardware, now drives the battlefield cadence. The U.S. military can’t currently simulate, much less master, such interconnected hyperwar. The need to do so is escalating. Aside from improving military decision making, such integrated digital environments are the means to train battlefield artificial intelligence. As important as it will be for our security, it will have important civilian applications as well.

The rise of virtual technology, including digital twins, hasn’t been as flashy as videogaming’s virtual realities, but it’s quietly reshaping how industries conceive physical technology, from race cars to artificial hearts. As accurate simulations become stand-ins for physical testing, innovation becomes faster and less expensive. According to a recent Allied Market Research report, the global digital-twin industry is projected to grow 20-fold, from $6.5 billion in 2021 to $125.7 billion by 2030. Some pundits predict it will spark a new industrial revolution.

Leading the charge is Formula 1. When sportwide cost caps were first introduced in 2021, they inadvertently triggered a digital-engineering race before the races we watch on television. Teams began designing more than 1,000 digital twins per race, each iteration shedding make-or-break milliseconds. By season’s end, cars evolve 85% from their initial designs, and those initial designs no longer qualify—all thanks to this digital-first approach. But there are challenges. Mercedes’s W13 car struggled early in 2022 because of a single simulation error. Crafting an accurate racing metaverse is no easy achievement, but it is a crowning one for teams that succeed.

With innovation velocity improving by orders of magnitude in Formula 1, it’s no surprise other industries are in hot pursuit. As the “Internet of Things" has expanded, models and simulations across fields have improved exponentially. From agriculture to medicine, other industries are accurately creating technology digitally before making it physically. This doesn’t only build things faster, cheaper and greener; it unleashes AI’s potential to master physical industries. And the time required might surprise you.

Google’s watershed Go-playing AI took years to develop. But within five weeks of retraining, the same game-playing algorithm successfully co-piloted an Air Force U-2 spy plane in 2020. Warfare may be gamified, and so may many industrial tasks currently performed by humans, putting them all within AI’s retraining grasp. All that’s needed is accurate simulations to learn.

Safely unleashing AI from the internet and into the physical world makes virtual engineering environments important to get right. But there’s a rub: Integrating disparate data sources is challenging. Computer limitations make it impossible to capture the universe’s complexity in a single simulation. So we slice up the complexity into thousands of models that capture intricate details like regulations, structures, manufacturing processes, performance and logistics. Stitching them together accurately into what is called a “digital thread" is slow and expensive. No one wants a digitally designed artificial heart that skips a beat.

Model One is a much needed attempt to revolutionize digital threading, automating it in a way all industries can use. Following another Air Force initiative to create the world’s first digitally certified airplane, the key to both is connecting disparate data securely and accurately. Laying the groundwork for digital twinning across disparate networks, while securely protecting proprietary and classified data, is next-level data meshing. But if successful, such “digital trust" infrastructure could simplify and accelerate virtual technology globally.

Imagine billions of future technologies being digitally designed, tested, even certified each year—like Formula 1 cars—all without the time, cost, and environmental degradation of physical innovation. Now, picture AI learning to improve these designs, operate them, and even train humans. Then the hoped-for future of personalized medicine, sustainable energy, abundant agriculture and space colonization may become attainable.

At the same time, so will the dark side of this future, with new cyber and AI risks to overcome. Suppose today’s state-of-the-art farming algorithms on harvesters around the world all suddenly began classifying crops as weeds due to some malign digital-twin virus. A catastrophic famine could ensue. When the physical world is controlled by a digital one, better security measures will become more valuable than data.

From the racetrack to the skies, that future is starting to take shape. As humanity navigates new horizons, we hope for the wisdom to avoid dark clouds.

Mr. Roper is founder and CEO of Istari Digital. He served as an assistant Air Force secretary, 2018-21. Mr. Schmidt is a former CEO and executive chairman of Google.