The race for artificial intelligence is no longer limited to conversational models, software agents or the productivity of knowledge workers. A new generation of companies is now seeking to apply AI to the physical systems that structure the global economy. Behind the fundraising of 255 million euros carried out by PhysicsX, led by Temasek, with the participation of Nvidia, Siemens, Applied Materials, Atomico and General Catalyst, there is a broader battle around strategic infrastructures, industry and material innovation.
For several years, debates around artificial intelligence have mainly focused on the capacity of models to generate text, code or images. However, the economic value of many sectors is still based on physical constraints which have lost none of their complexity. Designing an aircraft engine, optimizing an electronic chip, developing a next-generation battery or improving the energy efficiency of a data center always involves thousands of hours of simulation, analysis and iterations.
It is precisely this problem that PhysicsX intends to solve. Founded in London by former Formula 1 engineers, the company is developing an engineering platform natively designed around artificial intelligence. Its objective is to accelerate the understanding and modeling of physical phenomena in order to drastically reduce the design and validation times of industrial products.
“Almost every complex problem in the physical economy, better planes, better chips, better engines, or better energy systems, depends on how quickly and well engineers and operators can work on the underlying physical phenomena. For decades, this constraint limited material innovation. AI applied to physics removes it,” says Jacomo Corbo, co-founder and CEO of PhysicsX.
The promise is ambitious, where some simulations require several hours or even days of calculation, the models developed by the company aim to produce results in a few seconds. According to PhysicsX, this approach allows engineering teams to explore thousands of design variations where previously they only evaluated a handful.
The issue goes well beyond improving individual productivity, in sectors such as aeronautics, defense, semiconductors, energy or advanced materials, the time needed to go from a concept to an operational product is often the main limiting factor. As systems become more complex, development cycles lengthen while competitive pressure intensifies.
PhysicsX believes that recent advances in model architectures and the relative drop in the cost of GPU computing now make large-scale industrial deployment of this approach possible. The company says its technology is already used in the aerospace, defense, semiconductor, automotive, energy and manufacturing industries.
PhysicsX has extended its Series B funding round, bringing the total raised to more than 133 million euros. The investment comes from NVentures, alongside Atomico, Temasek, Siemens, Applied Materials, July Fund, General Catalyst, NGP and other existing investors
Beyond fundraising, another element deserves attention. PhysicsX announces that it intends to devote part of its investments to the development of new generation models which it calls “Large Physics Models”.
For three years, the sector has become familiar with the Large Language Models which power ChatGPT, Claude or Gemini. PhysicsX now suggests the emergence of an equivalent category no longer applied to language but to the understanding of physical phenomena.
The ambition is to train these models on immense volumes of simulations, industrial data and real-world information so that they can anticipate the behavior of complex systems. Ultimately, these models could become fundamental building blocks of digital engineering, capable of assisting the design of products in fields as varied as aircraft engines, energy reactors, advanced materials or semiconductors.
Robin Tuluie, founder and president of PhysicsX, believes that this development could transform access even to advanced engineering. “High-fidelity physics simulations have always been powerful, but they have remained slow, expensive and reserved for a limited number of specialists. AI applied to physics changes this reality in all its dimensions. »
This desire for democratization constitutes another strategic aspect of the project. In many industrial companies, the most sophisticated simulation tools remain concentrated in the hands of expert teams. PhysicsX seeks to bring these capabilities to a much wider range of users, whether engineers, designers or industrial operators.
The choice of sectors targeted by PhysicsX is just as revealing. Aeronautics, defense, energy, semiconductors, advanced materials and data centers are today among the strategic priorities of major economic powers. These industries concentrate a growing share of public and private investments linked to technological sovereignty, the energy transition and the rise of artificial intelligence.
As the United States, China and Europe increase investments in digital and industrial infrastructure, the ability to more quickly design chips, data centers, energy systems or defense equipment is becoming a major competitive advantage. From this perspective, the next frontier of artificial intelligence may not lie in conversational interfaces but in software that builds the physical world.
The growth claimed by PhysicsX demonstrates the market’s interest in this vision. The company indicates that it has doubled its recognized turnover over one year, tripled its contracted revenue and more than doubled its number of customers. Its workforce now exceeds 300 people, compared to around half a year earlier.
For investors, the bet is to finance not just an industrial software company, but potentially a new technological layer intended to slot into the heart of global engineering processes. After the battle of language models, that of strategic infrastructures could well become the next area of expansion of artificial intelligence.
