PhysicsX, a deep-tech startup based in London, revealed it has completed an oversubscribed Series C funding round, raising $300 million. The funding places the company’s valuation at around $2.4 billion, over doubling it from last year’s Series B, in which the company was valued at just below $1 billion.
Leading the round was Temasek, the sovereign wealth fund owned by the government of Singapore. Also participating were new institutional investors, M&G Investments and Intrepid Growth Partners. An intensely concentrated group of existing technology heavyweights and venture capital-backed firms also participated. Existing investors increasing their stakes include semiconductor manufacturing leader Applied Materials, AI chip powerhouse NVIDIA, Atomico, General Catalyst, Siemens, NGP, Radius, and the July Fund.
Overcoming the Simulation Bottleneck
For years, advanced manufacturing and heavy industry have been constrained by one basic limitation: physics is slow. Whether it is designing a more efficient wind turbine, cleaner jet engine, or high-density semiconductor chip package, engineers need to subject concepts to intensive computer-aided engineering (CAE) numerical simulation. However, high-fidelity simulations for CAE traditionally take hours, days, or weeks to compute across vast compute clusters. Consequently, engineering teams can only examine a handful of options, leaving vast performance gains unexplored.
Source: Unsplash
PhysicsX overcomes this constraint by creating a neural network-driven software stack. Unlike traditional approaches that use incremental numerical techniques to work through complex differential equations, the software leverages deep learning to make instantaneous physical predictions. With lightning-fast predictive performance, it is possible for engineers to simulate thousands of design decisions and ratio variants instantaneously, thereby compressing multi-month engineering project timelines into hours and days.
More News: Pathway Power Secures $150M for Battery Storage Expansion
The Rise of Large Physics Models
The massive capital infusion will be heavily directed toward expanding the company’s core frontier research, specifically the development of what PhysicsX terms “Large Physics Models” (LPMs). Much like the large language models that power conversational software and writing assistants, LPMs are trained on massive datasets of physical systems, thermodynamics, and fluid dynamics.
Rather than predicting the next word in a sentence, an LPM predicts how complex hardware will react under extreme real-world stress, heat, and aerodynamic pressure. This allows the software to augment traditional simulation data with actual real-world telemetry gotten from physical assets, creating highly reactive and highly accurate real-time digital twins for industrial operations.
The technology is also opening up advanced simulation. Historically, high-fidelity physics tools have been restricted to niche groups of specialised simulation scientists. By making the process computational and immediate, PhysicsX enables general designers, line engineers, and machine operators across an enterprise to test their ideas and optimise workflows independently.
Fueling the AI Hardware Infrastructure Boom
PhysicsX has been used to serve conventional deep-tech areas such as aerospace, defence, automotive engineering, and renewable energy, but the recent exponential growth is driven by an unexpected new market: the physical supply chain of the AI boom.
The very infrastructure needed to manufacture, energise and operate these enormous AI data centers has brought with it the largest hardware design problems ever faced. Power grids demand super-efficient step-down transformers; facility operators demand enormous centrifugal compressors and next-generation liquid-chilling modules; and semiconductor foundries demand hyper-optimised fabrication equipment to print the latest AI silicon chips.
Because PhysicsX provides the exact simulation layer needed to design this high-power infrastructure faster, its commercial demand has skyrocketed. Semiconductors are projected to become the company’s single largest industry segment this year. The company is currently experiencing extreme demand, doubling its recognised revenue year-over-year, tripling its booked revenue, and forcing management to carefully handle rollouts to keep pace with customer needs.
Source link