London-based startup Applied Computing has raised a $20 million Series A round led by KBR, with Databricks Ventures participating, to build out Orbital, a foundation AI model designed for the oil, gas, and petrochemical industry. TechCrunch

Founded in 2023 by co-founder and CEO Callum Adamson, Applied Computing exited stealth less than 18 months ago and reports it has already reached double-digit millions in annual recurring revenue. The company says Orbital is currently deployed at large, publicly listed upstream oil and gas, downstream refining, and petrochemical companies, though it has not named those customers.

Orbital combines three model types into a single system: a time series model (which finds patterns in data collected over time, like sensor readings), a physics-based model (which enforces the physical and chemical rules that govern how a plant actually behaves), and a language model (which lets people interact with the system in plain language). Together, they let plant technicians run simulations of how a change in one part of a facility could ripple through the rest of its operations. Applied Computing claims Orbital can flag anomalies, investigate root causes, and model whether a proposed fix could create problems elsewhere in a facility, all within minutes rather than the hours or days that traditional process simulation typically demands.

Adamson frames the problem in terms of data utilization. He states that facilities currently make operating decisions using less than 8% of the data available to them. The remainder, captured by sensors and historians (industrial data-logging systems) across a plant, goes unused in day-to-day decision-making. Orbital is designed to ingest that broader data corpus and surface actionable signals.

KBR's lead investment comes with a board seat, secured as part of its strategic stake. Investing.com KBR has already integrated Orbital into its INSITE 3.0 digital platform for energy projects and is using the model for ammonia production. Wipro is also listed as a partner, though the specifics of that relationship have not been detailed.

On the customer pipeline, Applied Computing says it is working with a major U.S. upstream operator and plans to announce a partnership with a European oil major in the coming weeks.

The multi-model architecture is where the technical design gets interesting. Industrial process plants generate enormous volumes of structured time series data from distributed control systems, and their physical behavior is governed by thermodynamic and chemical constraints that a language model alone cannot reliably encode. By pairing a time series model for sensor-driven pattern recognition with a physics-based model that enforces conservation laws and reaction kinetics, then wrapping the system in a language model interface for natural-language interaction, Orbital is attempting to bridge the gap between first-principles engineering simulation and the pattern-matching capabilities that transformer-based models excel at.

The broader context here is that foundation models have proven their value in text, code, and image generation. Applying them to industrial process data, where the cost of a wrong recommendation is measured in equipment damage or safety incidents rather than a bad output, is a substantially higher bar. The architecture Applied Computing has chosen, anchoring the language model to physics-based constraints rather than letting it operate as a free-form recommender, is a pragmatic response to that risk.

There are open questions worth noting. Whether the minutes-timescale claim holds in production environments at scale is unverified. The company's ARR figure and customer roster, while reportedly real, are self-reported. The involvement of KBR and Databricks Ventures, however, signals that both an established engineering procurement firm and a data infrastructure company see enough technical and commercial viability to commit capital and integrate the product into existing platforms.

For plant operators, the potential payoff is concrete: faster anomaly detection, faster root-cause investigation, and the ability to test corrective actions in simulation before committing them to a live facility. Whether Orbital delivers on that potential across diverse plant configurations and feedstock types will determine whether Applied Computing's rapid revenue growth reflects durable demand or early-stage pilot enthusiasm. The European oil major announcement, when it lands, will be a data point worth watching.