This AI Startup Just Built a Liquid Coolant That Could Save Every Data Center on Earth — Orbital Industries Raises $50M

What Is Orbital Industries?

While the AI industry obsesses over language models and chatbots, a small company in London just solved one of the most urgent infrastructure problems threatening the entire AI revolution — and almost nobody noticed. Orbital Industries, a startup that uses AI to design entirely new materials at the atomic level, has raised $50 million in Series B funding to scale its first commercial products. And the company’s breakthrough could not be more timely: they have designed a revolutionary liquid coolant for data centers that works better than existing solutions and does not contain the toxic “forever chemicals” that regulators are racing to ban.

This is not another AI wrapper or chatbot startup. Orbital Industries is using artificial intelligence to do something that was physically impossible before: simulate the quantum mechanical behavior of atoms at scale, then use those simulations to design materials with specific properties that have never existed in nature. The implications stretch far beyond data center cooling into semiconductors, aerospace, critical minerals, and energy. But the cooling product is where the money is right now, because every tech company on earth is desperately trying to cool the GPU farms powering the AI boom.

The $50 Million Series B and Nvidia Backing

The $50 million Series B round was led by venture firm Plural, with notable participation from Nventures — the venture arm of Nvidia. When Nvidia invests in a company that is building cooling solutions for GPU-heavy data centers, pay attention. Nvidia knows better than anyone how much heat their chips generate and how critical cooling is to the performance and economics of AI infrastructure. The company’s willingness to back Orbital validates the technology in a way that no benchmark or white paper could.

Previous investors Radical Ventures, Compound, and Fly Ventures also participated in the round. The company, which has offices in London and San Francisco, recently rebranded from Orbital Materials — a name change that reflects its expansion from pure materials research into integrated industrial hardware and infrastructure systems.

The $50 million will fund commercial deployment of Orbital’s first two products — both targeting the data center industry — and grow its 50-person team. For a company that is essentially reinventing materials science from the ground up, a 50-person team is remarkably lean. The efficiency comes from the AI platform doing the heavy lifting that would traditionally require massive laboratory operations and years of experimental trial-and-error.

The Orb Model: Simulating 100,000 Atoms on a Single GPU

Orbital’s core technology is an AI model called Orb, which can predict and simulate the quantum mechanical behavior of atoms. What makes Orb exceptional is its scale: the model can simulate 100,000 atoms on a single GPU, running roughly 10 times faster than competing models released by Meta and Microsoft. That speed advantage is not incremental — it is the difference between screening hundreds of candidate materials and screening hundreds of thousands.

Traditional materials science is painfully slow. Discovering a new material typically involves years of laboratory experimentation, testing thousands of candidates one at a time. Orb compresses that process to days or weeks by simulating how atoms will behave in different configurations, predicting material properties before anything is synthesized in a lab. The company screens hundreds of thousands of candidates computationally, then only synthesizes the most promising ones — a process that is orders of magnitude more efficient than brute-force experimentation.

This approach — using AI to design materials at the atomic level — represents one of the most genuine applications of artificial intelligence in the physical world. Unlike chatbots and image generators that operate purely in the digital realm, Orbital is using AI to create things that exist in the real world and solve tangible engineering problems. It is the kind of AI application that gets lost in the noise of consumer-facing AI products but arguably has far more transformative potential.

Solving the Data Center Cooling Crisis

The AI boom has created an infrastructure crisis that few people outside the industry fully appreciate. Training and running large AI models requires massive clusters of GPUs that generate extreme amounts of heat. A single Nvidia H100 GPU can draw 700 watts of power, and modern AI training runs use thousands of them packed tightly together. The heat generated by these GPU farms is pushing traditional air cooling systems past their limits.

Liquid cooling is widely regarded as the solution, but existing liquid coolants have significant drawbacks. Many of the most effective coolants contain PFAS (per- and polyfluoroalkyl substances) — synthetic chemicals that are incredibly persistent in the environment and are increasingly subject to regulatory restrictions in the U.S. and Europe. PFAS are often called “forever chemicals” because they do not break down naturally and accumulate in water, soil, and living organisms.

Orbital used its Orb model to screen hundreds of thousands of candidate formulations and design a new liquid coolant that matches or exceeds the thermal performance of PFAS-based coolants without using any PFAS compounds. This is a genuinely elegant solution to a genuine problem: the AI industry needs liquid cooling to grow, but the best existing coolants are being banned. Orbital’s product threads the needle perfectly.

The PFAS Problem and Why It Matters

The PFAS issue is not hypothetical — it is a regulatory and legal tsunami bearing down on the tech industry. The EPA has set enforceable maximum contaminant levels for several PFAS compounds in drinking water. Multiple states have enacted PFAS bans covering specific product categories. And the EU is considering one of the broadest PFAS restrictions in the world, which could affect thousands of industrial applications.

For data center operators, PFAS regulations create a potential nightmare scenario. If the coolant you are using gets banned, you do not just switch to a different product — you may need to drain, flush, and refill your entire cooling infrastructure. For a hyperscale data center processing billions of dollars in AI workloads, that kind of disruption is unacceptable. A PFAS-free coolant that performs as well as PFAS-based alternatives is not just nice to have — it is a strategic imperative.

Orbital’s timing is impeccable. By bringing a PFAS-free coolant to market while the regulations are still being finalized, they are positioning themselves as the safe choice for data center operators who want to invest in liquid cooling without the risk of regulatory disruption down the road. The competitive moat is the AI platform itself — reproducing Orbital’s results would require building an equivalent materials simulation capability, which is not something you can do overnight.

Beyond Coolants: Semiconductors, Aerospace, and Energy

While data center cooling is the immediate commercial opportunity, Orbital’s platform has applications across virtually every industry that depends on advanced materials. The company has identified semiconductors, critical minerals, aerospace, and energy as key expansion areas. In each case, the fundamental value proposition is the same: use AI to design materials with specific properties faster and more cheaply than traditional experimental methods.

In semiconductors, new materials are needed for next-generation chip architectures, advanced packaging, and thermal management. In aerospace, lighter and stronger materials can improve fuel efficiency and enable new vehicle designs. In energy, materials innovation is critical for better batteries, more efficient solar cells, and next-generation nuclear reactors. The total addressable market across these industries is measured in hundreds of billions of dollars.

Why This Matters for the AI Industry

Orbital Industries represents a category of AI company that does not get enough attention: companies using AI to solve physical-world problems rather than digital ones. While the AI industry has been dominated by foundation model companies and consumer applications, the long-term economic value of AI may be greatest in exactly the kind of applied science that Orbital is doing — using computational intelligence to design better materials, drugs, chemicals, and engineered systems.

The data center cooling angle is particularly poignant. The AI industry’s growth is constrained by its own energy consumption and heat generation. AI models are getting larger and more power-hungry, and the infrastructure needed to run them is bumping up against physical limits. Orbital is using AI to solve one of AI’s own infrastructure bottlenecks — a recursive loop where AI improves the conditions for its own advancement.

The Bottom Line

Orbital Industries is the kind of company that makes you optimistic about AI’s potential to solve real problems. While the AI industry has been busy building chatbots and image generators, Orbital has been quietly using AI to design new materials at the atomic level — and their first product could help solve the data center cooling crisis that threatens to choke the AI boom itself. With Nvidia backing, a 10x speed advantage over competing models, and perfect timing with PFAS regulations, Orbital Industries might be the most important AI company you have never heard of. At $50 million in Series B funding and a 50-person team, they are punching far above their weight — and the data center industry is about to find out just how hard they can hit.