A single refinery can carry thousands of sensors measuring temperature, pressure, velocity, and viscosity. According to Applied Computing, operators make decisions using less than 8% of what those sensors tell them.
The London startup has raised a $20m Series A to close that gap, led by engineering giant KBR with Databricks Ventures participating. Founded in 2023, it is building a foundation model for oil, gas, refining, and petrochemicals.
The problem is not collection, according to co-founder and chief executive Callum Adamson. Operators already gather the information.
They cannot combine sensor readings, engineering documentation, and the underlying physics and chemistry fast enough to predict anything useful from them.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!
The shape will be familiar: a foundation model trained on proprietary industrial data, with a large incumbent as both investor and route to market.
Mistral launched its industrial engineering tier with Airbus, BMW, and EDF as named customers on the same logic.
“It's getting those three data sources to talk to each other in real time,” Adamson told TechCrunch. “That's the real key.”
Its model, Orbital, is not a language model with an industrial skin on it. The company says it fuses a time series model, a physics-based model, and a language model to predict the state of a facility, reading sensor data while accounting for chemistry, equipment constraints, and what the operators are actually doing.
It also lets technicians simulate how a change in one part of a plant would ripple through the rest. That is the part the industry has historically paid consultants and weeks of downtime for.
It is also where the stakes sit. A refinery is not a customer-service queue, and Amazon has already warned that human oversight of AI degrades precisely because people stop scrutinising a system that is usually right.
The pitch, in the end, is speed. Applied Computing claims Orbital can flag an anomaly, work out what caused it, and model whether a proposed fix creates a problem somewhere else, all within minutes. Adamson says investigations that took days or weeks compress into seconds.
Some of this is landing. The company says it went from stealth to double-digit millions in annual recurring revenue in under 18 months, with Orbital deployed at unnamed “large, publicly listed” upstream, refining, and petrochemical companies.
Adamson declined to say how many customers it has, which is the sort of omission worth noticing next to a revenue claim.
KBR has integrated Orbital into its INSITE 3.0 platform and is using it for ammonia production. Adamson said the company is working with a major US upstream operator and expects to announce a European oil major in coming weeks.
The competitive picture is crowded and old. AspenTech sells simulation and AI-powered modelling across upstream, refining, and chemicals, while AVEVA does physics-based process simulation and what-if modelling.
Cognite and Seeq work the data layer. None of these are startups that can be outrun.
Adamson's answer is that none of them are competing for the right talent. “It's an AI problem. It's not a data problem, and it's not an energy problem,” he said. “If you're a tier-one AI researcher, where are you going to work? I don't think Shell's on that list.”
It is a good line, and it is also the entire bet. The claim is that the moat is neither industrial data nor process knowledge, both of which the incumbents have in depth, but the ability to assemble researchers who can build a model that beats Orbital.
Whether a $20m Series A buys that against AspenTech's installed base is the open question.
There is a second-order argument underneath. Adamson notes that operational data from working refineries is not public, and that simulated data cannot reproduce what happens inside a live plant, which makes deployments themselves the asset.
The KBR partnership matters for the same reason: it brings operational data, industry expertise, and introductions.
That reasoning is why heavy industry keeps ending up here. UPS is running a real-time digital twin of its entire logistics network on much the same basis.
The money goes on international expansion and research and engineering hires. The company opened a Houston office on Thursday, adding to its London headquarters and Bengaluru operational hub. The Middle East is next.