BeyondMath closes $18.5M seed round
BeyondMath, a London-based deeptech startup building a generative AI platform for physics simulation, has raised $18.5 million in seed funding, including a $10 million extension. The round was led by Cambridge Innovation Capital with participation from UP.Partners, Insight Partners, and InMotion Ventures.
Alan Patterson, CEO and co-founder, said the company has now raised $20.35 million in total and will use the new capital to scale its technology and team. The company did not disclose its valuation.
Targeting the simulation bottleneck
Engineering teams in aerospace, automotive, electronics, and industrial design face growing pressure to iterate faster while meeting efficiency and sustainability requirements. Traditional simulation workflows can take hours or days per run and often require expensive high-performance computing resources—delays that slow product development.
BeyondMath says its approach relies on a foundational model trained on first-principles physics rather than customer-generated simulation datasets. After training, the model can generate engineering-grade predictions quickly, enabling teams to explore far more design options in the same time window. The company claims performance gains of up to 1,000x on benchmarks such as the DrivAerML dataset.
Industrial validation and expansion plans
A key reference project is the company’s three-year, roughly $19 million STRATA program with Honeywell, aimed at accelerating design iteration for complex aircraft components. BeyondMath argues that faster iteration can support lighter parts, improved fuel efficiency, and reduced emissions.
Founded in 2022 by Alan Patterson and Darren Garvey, the startup plans to expand commercial deployments and broaden the range of physical phenomena its model can simulate, while improving integration with existing design and PLM systems. It also intends to double headcount this year across research, engineering, and customer support, and grow sales across Europe, the US, and Japan.










