PhysicsX has established its Asia-Pacific hub in Singapore, where the London-based company aims to use AI models to speed up engineering simulations and industrial work. The expansion puts a substantial AI-for-engineering business closer to manufacturing and semiconductor customers, though its productivity claims still need to be judged by results in real workflows.
Watch Desk analysis
What happened
PhysicsX develops AI models intended to replace some traditional numerical computer simulations, which engineers use to predict how materials, machines and factory processes will behave. The Straits Times reports that the company’s Singapore hub opened on 8 October. CEO Jacomo Corbo told the paper that its platform can integrate with customers’ existing computer-aided design, simulation and product-lifecycle systems, rather than requiring them to abandon established tools.
PhysicsX says its customers span sectors including aerospace, semiconductors, automotive, materials science and energy. Corbo said aerospace and defence are its largest sectors, followed by semiconductors, where the company is working mainly with equipment manufacturers and increasingly deploying into chip fabrication plants.
The company has raised about US$500 million across three funding rounds and has grown to around 400 staff, according to the article. It has more than 20 people in Singapore and plans to reach 100 there within three years. The report also names partnerships with companies including Leonardo, Audi, Rio Tinto, NVIDIA and Siemens.
Why it matters
Engineering simulation is often computationally demanding and embedded in specialist workflows. PhysicsX’s pitch is that trained AI models can produce predictions faster, while fitting into the systems engineers already use. If that works reliably, the payoff could be shorter design cycles and more opportunities to test or optimise industrial processes.
Singapore gives the company a base near customers and a dense cluster of manufacturing, engineering and semiconductor activity. But a new office, a large funding total and impressive partner names do not tell us how much faster or more accurate a particular customer’s work becomes. The practical test is whether engineers can trust the models for the decisions that matter.
Our read
This is a serious industrial-AI expansion, not a chatbot with a hard hat. The compelling idea is AI moving into engineering workflows where quicker simulation could change what teams can test and build. PhysicsX’s own claims about accuracy and productivity are promising, but customers will want evidence tied to specific applications, not just the general promise of faster inference.
What to watch
- Whether PhysicsX names specific customer deployments and measurable results.
- How its models perform against conventional simulation on accuracy as well as speed.
- Whether the Singapore team reaches its planned 100 staff, and what work is based there.
Discussion spark: For engineering AI, what should customers demand before trusting a model in the design workflow: independent accuracy comparisons, or proven savings on a live project?
Sources and evidence
- Source update (11 October 2026, 23:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.