Orbital Materials
London AI-materials startup (now Orbital Industries) — AI-discovered carbon-capture sorbents and data-center cooling hardware; $50M Series B [1][2].
1. Core Product / Service
Orbital Materials (founded 2022, London + Princeton, NJ) discovers cleantech materials with AI, pairing two in-house models: LINUS, a generative model that proposes material formulas from desired properties, and Orb (open-sourced, Apache-2.0), a foundation model that simulates atoms' quantum-mechanical behavior — reportedly able to simulate ~100,000 atoms on a single GPU and ~5× faster than competing models [3][4].
Its first outputs were in carbon capture (a sorbent that "captures CO₂ like a sponge at the atomic level"), catalysts, and water treatment. In May 2026 it rebranded to Orbital Industries and expanded toward AI-designed data-center infrastructure via a commercial arm, Orbital IT: a PFAS-free dielectric cooling fluid for next-generation high-density GPUs, and modular data centers built off-site (SLB as manufacturing partner) targeting deployment in as little as six months [1][2].
2. Target Users & Pain Points
- Data-center operators facing GPU power-density and cooling bottlenecks — PFAS "forever chemicals" face tightening EPA/EU restrictions after 3M exited Novec coolant production in 2024 [2].
- Industrial partners (AWS, energy and chemical firms) wanting AI-discovered materials for decarbonization.
Pain solved: non-PFAS coolants and cheap carbon-capture sorbents that incumbent chemistry has not produced, plus faster data-center buildout amid the AI compute buildout [1][2].
3. Competitive Landscape
| Player | Approach | Vs. Orbital |
|---|---|---|
| Orbital | AI materials → data-center hardware (Orbital IT) | Direct hardware commercialization, AWS partnership |
| cuspai | AI materials, multi-sector + Foundry | Software/simulation-first; broader sectors |
| radical-ai | Autonomous robotic lab, alloys | Inorganic alloys; wet-lab-first |
| periodic-labs | Autonomous labs (superconductors) | Physics/chemistry discovery; no hardware product yet |
Orbital's distinction is commercializing its own hardware (coolant, modular DCs) rather than selling discovery as a service — the closest AI4S play to a physical product company [1][2].
4. Unique Observations
- The pivot to data centers is the story. Orbital started in carbon capture and steered into the AI-compute physical layer, monetizing its Orb model where demand is hottest (cooling next-gen GPUs). It is a case study of an AI4S company following revenue to the AI buildout rather than waiting on cleantech economics [1][2].
- AWS is both customer and channel. The December 2024 multiyear AWS collaboration (decarbonization, cooling, water) and Orb's availability on AWS Marketplace / SageMaker JumpStart give it distribution most materials startups lack [3].
- "Copilot, not replacement" thesis. CEO Jonathan Godwin (ex-DeepMind) frames AI as amplifying chemists rather than replacing them — a more conservative position than the fully-autonomous-lab rhetoric at periodic-labs and lila-sciences [4].
5. Financials / Funding
- Founded: 2022 by Jonathan Godwin (ex-DeepMind), James Gin-Pollock (CTO), Daniel Miodovnik (COO); ~50 staff, offices in London and San Francisco [2].
- Early rounds: $16M (Feb 2024, incl. Toyota Ventures); ~$21M total from Radical Ventures, Sequoia Scout Fund, Toyota Ventures; an undisclosed Nvidia NVentures investment [3][4].
- Series B (May 2026): $50M (£37M), led by Plural, with Nvidia's NVentures, Radical Ventures, Compound, and Fly Ventures [1][2].
- Total raised: ~$121M (PitchBook) [2].
6. People & Relationships
- Co-founder / CEO: Jonathan Godwin (former DeepMind researcher) [2].
- Co-founders: James Gin-Pollock (CTO), Daniel Miodovnik (COO) [2].
- Key partners: AWS (multiyear, data-center decarbonization); SLB (manufacturing) [2][3].
- Peers: cuspai, radical-ai, periodic-labs.