Physics AI - Thermal-Mechanical Reliability

Flexcompute collaborates with Starlab and Voyager on gpu-native simulation of critical Space Station Life Support System. Oct 06, 2026, 08:49 ET Leading CFD platform provider helps Starlab's life support team model airflow inside the commercial successor to the International Space Station BOSTON, Oct. 6, 2026 /PRNewswire/ - Flexcompute, the physics company, today announced it is collaborating with Starlab Space and Voyager Technologies (NYSE: VOYG) to simulate cabin airflow inside Starlab, the commercial space station designed to succeed the International Space Station (ISS). On a space station, air does not move on its own. In microgravity, warm air does not rise, and the carbon dioxide crew members exhale can linger around them unless the ventilation system keeps it moving. Starlab's Environmental Control and Life Support System (ECLSS) has to circulate fresh air through every part of the station and keep the crew comfortable. Because Starlab is designed to launch in a single flight, that design has to be right before it leaves the ground. Flexcompute used Flow360, its GPU-native computational fluid dynamics (CFD) platform, and GeometryAI, its automated geometry preparation engine, to model airflow across the full station interior. The first study closely matched Starlab's existing models in a fraction of the time traditional CFD tools take. The teams are now extending the work to model how the carbon dioxide crew members exhale moves through the cabin and how the life support system removes it. "Building a commercial space station takes partners who move at our pace," said Jonathan O'Neill, Atmosphere Management System Lead at Starlab Space. "Flexcompute's engineers paved new ground in cabin ventilation CFD. That level of expert support gives us more confidence as Starlab moves from design into production." "Life support airflow can't be fully tested on the ground, so simulation carries much of the design work," said Vera Yang, Co-founder and President of Flexcompute. "Working with Starlab's ECLSS team shows what GPU-native physics makes possible for commercial space: faster answers, earlier in design, so engineers can move forward with confidence." About Starlab Starlab Space is a U.S.-led, global joint venture among Voyager Technologies (NYSE: VOYG), Airbus, Mitsubishi Corporation, MDA Space, Palantir Technologies, and Space Applications Services, with strategic partners including Hilton, Journey, Northrop Grumman, and The Ohio State University. Starlab is developing a next-generation, AI-enabled commercial space station to ensure continued human presence in low Earth orbit and a seamless transition of microgravity science and research alongside the retirement of the International Space Station. Starlab's advanced, user-driven design and robust capabilities make it a premier platform for scientific discovery and technological advancement in space. For more information, visit starlab-space.com. About Flexcompute Flexcompute is a frontier AI lab building a Physics World Model that bridges data-driven AI and the laws of physics. Founded by researchers from Stanford and MIT, its team of more than 70 PhDs develop GPU-native solvers for fluid dynamics, electromagnetics, structures and heat, and integrate physics-based computation into AI agents through tokenized physics. This creates a verification layer that grounds AI predictions in trusted physics and helps engineers and AI agents evaluate how designs will perform before they are built. Engineering teams at more than 250 companies across aerospace, automotive, semiconductors, photonics and advanced manufacturing use Flexcompute to evaluate, explore and optimize designs, allowing them to innovate with confidence. Learn more at flexcompute.com. | Media Contacts | / | | Flexcompute | Starlab Space | | Silvie Casanova | Melissa Price | | [email protected] | [email protected] | | Voyager Technologies | / | | Nora Ellish | / | | [email protected] | / | SOURCE FlexCompute Inc.

Flexcompute has introduced the first fully autonomous agent-driven system for end-to-end photonic chip design. The system uses AI agents to propose designs, run physics simulations, verify fabrication constraints and iterate independently, producing tapeout-ready layouts in hours rather than weeks. The technology combines Tidy3D Multiphysics solver, PhotonForge foundry process design kits, and GPU clusters to enable rapid iteration across optical, electrical and thermal physics. Agents now complete designs that previously required senior photonic engineers, finishing them in overnight runs. Flexcompute is among the first companies deploying autonomous AI engineers with NVIDIA NemoClaw, announced at COMPUTEX 2026. The system works across platforms including Anthropic's Claude and OpenAI's Codex. The architecture will extend to electromagnetics, heat transfer and computational fluid dynamics applications.

Flexcompute and Northrop Grumman have developed an AI Physics model using NVIDIA technology that reduces space mission preparation time by 100 times. The system accurately predicts thruster impingement effects during space docking in real time, a process that traditionally required months of simulation. The breakthrough addresses plume impingement, where rocket thruster exhaust interacts with nearby spacecraft structures in space's vacuum. Built on NVIDIA Physics NeMo, the model delivers predictions in seconds rather than months whilst providing uncertainty estimates for mission-critical decisions. The technology enables more efficient spacecraft operations through improved plume interaction modelling, allowing for more confident control strategies, lighter structural margins and more efficient fuel usage. This supports longer mission lifetimes and more sustainable space operations, advancing capabilities in station keeping, rendezvous and space robotics.