PhysicsX has raised $300 million at a $2.4 billion valuation, a sign that investors are moving beyond chatbots and into AI that can redesign the physical economy.
PhysicsX is not selling another assistant that writes emails faster. The London startup is building AI models that help engineers simulate aircraft parts, semiconductor equipment, vehicles and energy systems in seconds, instead of waiting hours for conventional computer-aided engineering runs. That difference is why Temasek led a new $300 million funding round that values the company at about $2.4 billion.
The number matters because it gives industrial AI a place in the same funding conversation that has mostly been dominated by large language model companies. For much of this AI cycle, the highest private valuations have gone to frontier model labs, coding agents and cloud infrastructure. PhysicsX points to a quieter but increasingly important layer: software that applies AI to the real constraints of manufacturing, materials and engineering.
According to Bloomberg, the round more than doubles PhysicsX's previous valuation, which was below $1 billion. New investors include Intrepid Growth Partners and M&G Catalyst, while Nvidia and Applied Materials are among the existing backers continuing to support the business. That investor mix is not accidental. If chips, data centers and aerospace systems are becoming more complex, the tools used to design and validate them also have to change.
PhysicsX sits in a market where speed is not just a nice feature. In aerospace, one design decision can ripple through fuel efficiency, certification timelines and production cost. In semiconductors, tiny changes in process equipment can affect yield and factory output. In automotive and energy, companies are under pressure to build lighter, cleaner and more efficient systems while shortening development cycles.
Traditional simulation software has been one of the main ways engineers manage that complexity. The problem is that high-fidelity simulation can be slow, expensive and tied to specialist teams. PhysicsX is trying to compress that process by training AI models on physical systems, letting engineers explore more options faster. The promise is not that AI replaces physics. It is that AI makes physics usable at a different pace.
The company has been preparing the ground for this moment. Its June 2025 Series B raised $135 million and was led by Atomico, with Temasek, Siemens and Applied Materials also participating. At the time, PhysicsX said it had grown to more than 150 employees and had more than quadrupled revenue since its Series A. The latest briefed figures show another step up, with headcount now above 300, revenue doubling year on year and booked revenue tripling.
Nvidia And Temasek Are Betting On The Same Layer
Nvidia's involvement is especially telling. The company is already the central supplier of accelerated computing for AI, but its strategic interest goes beyond selling more GPUs to model labs. PhysicsX uses accelerated computing to train and deploy physics-informed models, and Nvidia has been pushing hard into physical AI, where models interact with engineering, robotics, digital twins and industrial systems.
Applied Materials also has a clear reason to care. Semiconductor manufacturing is one of the most demanding engineering environments in the world. Better simulation can help improve equipment design, materials work and process optimization. For a company tied to chipmaking infrastructure, backing software that can make advanced manufacturing more efficient is more than a venture bet. It is close to the core market.
Temasek's lead role adds another angle. Sovereign and institutional investors are looking for AI exposure that is not only about consumer applications or advertising-driven software. Manufacturing, energy and chips are tied to national competitiveness, supply-chain resilience and infrastructure spending. PhysicsX gives investors a way to back AI that lives underneath those sectors, where the output is measured in better machines, faster design loops and stronger industrial capacity.
The company is also moving toward what it calls Large Physics Models. The idea is familiar from large language models, but the job is different. Instead of predicting the next word, these models aim to understand physical behavior across engineering domains, from aerodynamics to thermal systems. If that works at scale, an engineer could test many more designs before a prototype is built, and companies could carry learning from one project into the next with far less friction.
There is still a hard test ahead. Industrial customers will not buy vague AI promises when safety, cost and performance are on the line. PhysicsX has to prove that its models are accurate enough, secure enough and practical enough to sit inside serious engineering workflows. A backlog of roughly six months of customer demand and projected revenue close to $50 million this year suggest there is pull, but the next phase will be about execution.
That is what makes this funding round worth watching. The AI market is starting to separate companies that generate demos from companies that change how expensive work gets done. PhysicsX is now valued like one of the latter. If seconds-long simulation becomes a normal part of advanced manufacturing, the bigger story will not be the $2.4 billion valuation. It will be how quickly industrial companies decide they can no longer compete without this layer.
Also read: Aviva shows how AI is now fighting and fueling insurance fraud • Texas grid tests put AI data center growth on notice this summer • The stronger dollar is turning Fed risk into a startup financing problem
