Flexcompute Secures Investment from Samsung and Applied Ventures, Pulse 2.0 Reports

Flexcompute Secures Strategic Backing from Samsung and Applied Materials to Expand Physics World Model

Pulse 2.0 reported that Samsung Ventures and Applied Ventures—the venture capital arm of Applied Materials—provided a strategic investment to Flexcompute to speed up the creation of its Physics World Model. Additional undisclosed investors also joined the financing round, which aims to scale the startup’s GPU-native simulation software.

The Tech TL;DR:
• Flexcompute raised new funding from Samsung Ventures and Applied Ventures to expand its Physics World Model.
• The technology combines neural networks with first-principles physics solvers to predict system behaviors without relying solely on historical text or image data.
• Engineering teams across more than 250 companies currently use the platform to simulate semiconductor packaging, aerospace components, and automotive designs.

Developing Neural Networks Grounded in Physical Laws

Founded by researchers from Stanford and MIT, Flexcompute is building AI systems designed to predict how physical products and systems behave by combining neural networks with first-principles physics. According to the company, the approach moves beyond traditional AI models trained primarily on historical text, images, and simulation data. Comprising more than 70 Ph.D.s, the research team has built GPU-native solvers covering computational fluid dynamics, electromagnetics, structural mechanics, and thermal analysis.

The platform uses an approach termed Tokenized Physics, which integrates neural networks with physical laws and conservation principles. This architecture allows the model to make accurate predictions even when evaluating designs it has not previously encountered. The technology addresses engineering problems ranging from semiconductor package warpage caused by heat to the behavior of spacecraft during atmospheric reentry.

Solving Complex Semiconductor Warpage Models in Minutes

Flexcompute demonstrated the scale of its software using an advanced semiconductor packaging example involving more than 800,000 solder joints, an ASIC, and 32 optical engines. Within minutes, the company successfully solved a warpage model featuring 5 billion degrees of freedom, achieving results within 2% of commercial finite-element outputs.

Customers access the Physics World Model through Physics Twins, digital representations of products that allow engineers and AI agents to test, evaluate, and optimize designs before physical manufacturing begins. In this workflow, AI agents handle iterations while engineers set objectives and review outcomes, leaving the underlying physics to verify whether proposed designs are practical. Engineering teams at more than 250 companies across aerospace, automotive, semiconductors, photonics, and advanced manufacturing currently utilize the technology.

Executive Perspectives on AI and Physical Simulation

“Every industry is racing to put AI to work on real engineering problems, and the limit is how quickly each one can be validated against the laws of physics,” said a representative of Samsung Ventures. “Flexcompute has built something unique: simulation fast enough to keep pace with AI and rigorous enough for production decisions.”

Samsung Ventures noted that hardware development requires systems that understand governing laws rather than just training data. “AI is changing how products are designed, but those products still have to obey the laws of physics,” a Samsung Ventures representative stated. “We believe the Physics World Model will shape how the next generation of hardware is developed.”

Vera Yang, Co-Founder and President of Flexcompute, emphasized the verification layer between computational bits and physical atoms. “Language models learned the world from text. Hardware is governed by laws, not text, and no amount of data substitutes for them,” Yang said.

Capital Allocation and Expansion Plans

To scale its Physics World Model across fluids, electromagnetics, structures, thermal systems, and multiphysics interactions, Flexcompute will allocate the new strategic funding toward growing its research staff and boosting GPU computing resources.