A racing car can be moving very quickly while the people responsible for it are running out of time. Should it stay out for another lap? Should it stop now? What will the other team do if it does? At the pit wall, an answer delivered after the decision has passed is an elegant souvenir.
Jacomo Corbo made a career out of that awkward relationship between knowledge and timing. He worked in race strategy at Renault, became a founder of QuantumBlack, and now leads PhysicsX, a company building AI tools for engineering and manufacturing. The settings have changed. The recurring problem is how to make useful calculations arrive while someone can still act on them.
He grew up in Canada watching Formula One. Years later, while studying computer science at Harvard, his research into game theory attracted Renault’s attention. He took a year away from his doctorate to work with the team. A sport he had followed became a place where the mathematics had somewhere urgent to go.
An early professional biography dates his Renault stint from January 2006 to June 2008, first as race strategist and then chief race strategist. In 2006, Renault won both the constructors’ and drivers’ championships. Corbo’s work belonged to that collective effort: a team of people trying to make one car, one driver and one decision add up at the right moment.
An engineer, even when the job title changes
“I still think of myself as an engineer,” Corbo has said. Before Harvard, he studied electrical engineering at McGill. He also spent time at Bosch in Germany building steer-by-wire systems. Control theory, computer science and a racing team make a less peculiar career sequence when the common interest is getting a complicated system to behave.
His account of arriving in Formula One contains a revealing observation. He found people with deep knowledge of aerodynamics, vehicle dynamics and materials, working with ways of handling data that did not use techniques familiar to computer scientists. Expertise was plentiful. The methods available to put it to work had room to improve.
That observation helped inform QuantumBlack. Founded in 2009, the company applied data science to difficult performance problems. Its early work included Formula One, but its client list extended to organizations such as Boeing and Arup. The proposition could travel: an organization with valuable data and difficult decisions might benefit from a different way of connecting them.
Corbo’s professional life also included research roles at the University of Ottawa and Wharton. The academic work and commercial work overlapped in their subject matter: information, performance and the way organizations make decisions. It is a useful background for someone selling software into companies where one department’s promising idea can become several other departments’ scheduling problem.
“I still think of myself as an engineer.”
Jacomo Corbo
The second company begins with the waiting
McKinsey acquired QuantumBlack in 2015. Corbo became a partner at the consulting firm while serving as QuantumBlack’s chief scientist. He helped build its research and product engineering efforts, including QB/Labs. By the time PhysicsX announced his arrival as co-CEO in February 2023, QuantumBlack had grown to more than 1,500 technical practitioners globally.
The new job concentrated his attention on industrial engineering. PhysicsX paired him with Robin Tuluie, whose career included theoretical physics, Renault, Mercedes and Bentley. They shared experience of a world in which a design improvement has to survive contact with airflow, materials, manufacturing and the calendar. A lovely rendering offers little protection against an uncooperative fluid.
The question underlying PhysicsX is straightforward. Before building a component, engineers need to know how it will perform. Numerical simulation lets them investigate heat, flow, loads and other behavior without making every candidate physically. But detailed calculations can consume hours or days. That waiting limits how many possibilities a team can seriously investigate.
PhysicsX trains AI models on simulation data and, where available, experimental observations. The trained model can then predict behavior much faster for the application it has learned. In his writing, Corbo describes a shift in where simulation and physical testing sit: they supply data and help evaluate and validate models, while inference handles more of the repeated work inside engineering workflows.
The brake duct is allowed to be interesting
To understand the appeal, consider a brake cooling duct. It has to direct air where it is needed without creating unacceptable drag or exceeding the space available inside a vehicle. A change that improves one measurement can upset another. Even a small component can keep a team busy with a rather large argument.
In a project described by PhysicsX in August 2025, ten design parameters defined a brake duct, air guide and connecting path. Its approach selected simulations where uncertainty or potential gains justified the effort, then used the results to improve the model. The company reported 50 simulation runs for the first vehicle platform and 25 for a second, using knowledge transferred from the first.
It is a modest-looking object with a consequential lesson. The value lies in making a constrained search more practical, including carrying what has been learned into another design. The work still has packaging limits and manufacturing requirements. The computer gets more chances to search within them.
In another project, PhysicsX worked with an aerospace manufacturer on casting. Changes in temperature, flow and other conditions can affect whether a component emerges usable. The company described models that forecast defects during production and give operators feedback, alongside tools for improving mold and gating geometry. A prediction becomes more valuable when it arrives before an expensive mistake has finished taking shape.
What he wants a team to know about failure
Corbo’s description of hiring offers a quieter view of the person behind the engineering argument. At Harvard he worked with David Parkes, in an environment he valued for its collaboration across disciplines. He has described seeking doctoral graduates who can break unfamiliar problems into smaller parts, teach themselves what they need and persist when attempts fail.
There is a practical affection in that account of research. A doctorate offers repeated experience of approaching a problem without a promised result. Corbo values the habits acquired along the way: independence, collaboration and tolerance for unsuccessful attempts. Those are useful qualities in a business asking teams to work on problems without a ready-made solution.
He has also connected his time in Formula One with learning to remain calm under pressure. A racing background lends itself to theatrical business metaphors; the more useful inheritance may be the discipline of making a decision with incomplete information, then paying close attention to what happens next.
That discipline matters as the organization grows. PhysicsX’s July 2026 writing on adoption acknowledged that training and maintaining models costs money and engineering effort. Some numerical solvers are already fast enough that a dedicated AI model needs a careful business case. Faster prediction has to improve the larger working process, including its handoffs and disconnected tools. An impressive demonstration alone does not make a useful deployment.
The company gets bigger. The objects stay physical.
In June 2025, PhysicsX announced a $135 million Series B led by Atomico, with investors including Temasek, Siemens and Applied Materials. The mix brought together financial backing and companies familiar with the engineering environments Corbo wanted to serve. The industrial setting was becoming part of the financing story as well as the product story.
A year later, on June 8, 2026, the company announced a $300 million Series C led by Temasek at an approximately $2.4 billion valuation. PhysicsX said its team had passed 300 people and its customer count had more than doubled over the preceding year. Those are company-reported measures of expansion, with an accompanying obligation to turn more deployments into repeatable work.
One application brings a pleasing circularity to the business. PhysicsX and Siemens are using physics AI to investigate power distribution in AI data centers. Busbars carrying electricity also produce heat; the design must cope with changing loads. The collaboration aims to give engineers thermal predictions in seconds for analyses that could otherwise take days. AI is helping design the physical infrastructure required to run AI.
In October 2026, PhysicsX established an Asia-Pacific hub in Singapore. It announced a five-year partnership with A*STAR involving a joint S$40 million commitment and plans to grow its Singapore team beyond 100 over three years. Corbo’s stated ambition includes aircraft, chips, energy systems and industrial infrastructure. Reaching those applications requires people close to the organizations making them.
Financing led by Temasek.
Approximately $2.4bn company valuation.
Another race, a different kind of hull
On October 21, 2026, the racing thread returned in a new form. PhysicsX and GB1, Britain’s America’s Cup team, presented work on yacht hull design for the 2029 competition. Geometry models generate candidate shapes; physics models predict performance; the results guide the next round of exploration. The aim is to investigate a wider set of possibilities while there is still time to do something with them.
The partners reported early results including 100 times faster evaluation overall, measured against their previous approach. That figure belongs to this yacht-design work, with its particular comparisons and conditions. The more interesting continuity is the working habit: generate possibilities, judge them, learn, repeat. The expensive decision gets more rehearsals before it becomes a physical object.
Corbo’s path has moved through an electrical engineering degree, a computer science doctorate, a racing team, an analytics business and a company working across advanced manufacturing. The connections are less mysterious when viewed from the engineer’s desk. Each stage concerns what people can discover before money, machinery and time commit them to a choice.
A race strategist asks whether another lap is worth the risk. An engineer asks whether another design is worth the wait. Corbo has spent much of his working life trying to improve the terms of that second question. At PhysicsX, his ambition is to give the engineer enough time to ask it again.