Applied Computing wants to give oil and gas operators an AI model for the entire plant

Refineries are drowning in data they barely use. Applied Computing, a London startup founded in 2023, says the average oil, gas or petrochemical facility runs on less than 8% of the information its own sensors generate. Not because the sensors aren't collecting it, but because nobody has found a fast way to fuse three very different data streams: live readings on temperature and pressure, decades of engineering documentation, and the underlying physics and chemistry that govern how a plant actually behaves.

The company's answer is a model called Orbital, and it's built differently than the chatbots most people associate with AI. Rather than predicting the next word in a sentence, Orbital stitches together a time-series model, a physics-based model and a language model to estimate the real-time state of a facility. CEO Callum Adamson says that combination lets Orbital flag an anomaly, trace its cause, and simulate whether a proposed fix might cause trouble somewhere else in the plant, all inside a few minutes. He claims work that used to take investigators days or weeks now takes seconds.

That pitch has apparently landed. Applied Computing says it went from stealth mode to double-digit millions in annual recurring revenue in under 18 months, with customers among large, publicly traded upstream and downstream energy companies, though Adamson wouldn't say exactly how many. Wipro and KBR are named partners; KBR has folded Orbital into its INSITE 3.0 platform and is using it on ammonia production specifically. A major U.S. upstream operator is already in the mix too, and Adamson expects to announce a European oil major partnership within weeks.

The $20 million Series A was led by KBR itself, with Databricks Ventures joining in, which is a notable arrangement since KBR is both investor and customer. Adamson argues that relationship gives Applied Computing something rivals can't easily buy: real operational data from live plants, industry credibility, and warm introductions to future clients. He's dismissive of the idea that data access is the real competitive barrier, insisting the harder problem is recruiting AI researchers who'd rather build models than work for an oil major directly. Whether that argument holds against entrenched players like AspenTech and AVEVA, or data-layer specialists like Cognite and Seeq, is the open question the market will answer.

For now, the money goes toward international growth: a new Houston office joins headquarters in London and an operational hub in Bengaluru, putting the company closer to existing North American customers, with a Middle East push reportedly next.