Orbital Industries raises $50M to take AI-discovered materials from model to manufacturing

Why it matters

If Orbital can reliably turn AI simulations into shipped materials, it shortens the loop between discovery and deployment for critical hardware like AI data centers, and sets a template for full-stack AI in the physical economy.

Orbital Industries has raised a $50 million Series B to scale its full-stack approach to AI-driven materials, according to an announcement thread on X and a Fortune report.

Aligned News on X: Orbital Industries raises $50M

The round backs founders @jgodwin_ai, @gin_james, and @Dmiodovnik, who are building Orbital Industries around a simple idea with hard execution: use an in-house foundation model to discover new compounds at the atomic level, then engineer and manufacture real products from those discoveries. The thread says @jgodwin_ai spent years working inside DeepMind's materials research program, and cites Demis Hassabis as a model for ambition emerging from London.

The Fortune piece by Jeremy Kahn also spotlights the raise and Orbital's ambitions.

The bet: full-stack materials, not just software

The team frames their strategy as deliberately more integrated than selling R&D software to incumbents. Operating out of London and San Francisco, they are targeting applications in energy, aerospace, and semiconductors. The thread argues this approach can push novel materials into the world faster by controlling more of the stack: discovery, productization, and manufacturing.

At the core is Orbital's foundation model, Orb, which the thread's author claims outperforms efforts from Meta, Google, and leading academic labs on simulating physical systems, with far less funding behind it. Another post asserts Orb can simulate 100,000 atoms on a single GPU, runs roughly 10x faster than the nearest alternative, and maintains stable predictions over time so scientists can trust results. Those are investor-side claims; independent benchmarks were not detailed in the materials provided.

From model to product: dielectric cooling

Orbital's first commercial product is described as a non-toxic dielectric cooling fluid for data centers, a response to one of the tightest bottlenecks in AI infrastructure as chips push past 2,000W TDP. The thread says Orb was used to find and develop the fluid in a few years, a timeline it contrasts with the near-decade often required for such materials. It also describes partnerships with large customers like AWS for further testing and development, and says Orbital is building modular data centers with the fluid embedded, engineered for Vera Rubin and the next generation of high-power chips.

Founder roots and ambition

The thread specifically highlights DeepMind lineage and the founders' ambition to build from that foundation in London. "We're extremely lucky to have @demishassabis in London as a shining example of ambition for people to take inspiration from," one post reads, before suggesting an even broader horizon: a platform company for materials, akin to Isomorphic Labs but for the physical world.

Who is backing them

The thread lists existing backers including @CompoundVC, @FlyVC, NVentures, @radicalvcfund, and @Toyota_Ventures, with Plural joining the round and publishing a note titled "Why we invested in Orbital Industries." Fortune's profile by @jeremyakahn also spotlights the raise and Orbital's ambitions.

What to watch next: how fast Orbital can translate Orb's simulations into validated, manufacturable materials, whether its dielectric fluid proves out in production-scale deployments, and how its integrated approach competes against pure software plays and incumbent materials giants.