level: business
applied computing, a london-based startup, raised a $20 million series a led by kbr with databricks ventures participating. its foundation model, orbital, targets oil, gas, refining, and petrochemical facilities. these plants have thousands of sensors but use less than 8% of available data for decisions. orbital combines a time series model, a physics-based model, and a language model to predict a facility's state by analyzing sensor readings, physics, chemistry, equipment constraints, and operator activity. it can flag anomalies, investigate causes, and model fixes in minutes, compressing work that once took days or weeks.
the startup says it reached double-digit millions in annual recurring revenue within 18 months of leaving stealth. orbital is used by large publicly listed upstream, downstream, and petrochemical companies, though customer numbers are undisclosed. partners include wipro and kbr, which integrated orbital into its insite 3.0 platform for ammonia production. applied computing is also working with a major u.s. upstream operator and plans a european oil major partnership. the company faces competition from aspen tech, aveva, cognite, and seeq, but its ceo argues the key challenge is assembling ai researchers, not accessing data or energy expertise.
the kbr partnership provides operational data, industry expertise, and customer introductions. applied computing will use the funding to expand internationally, hire research and engineering staff, and explore more energy client deployments. it opened a houston office to be near north american customers and plans middle east expansion. the model's advantage comes from real operational data from deployments, which is not publicly available, unlike simulated data. this data helps improve the model's accuracy for real-world plant conditions.
why it matters: this model could help energy operators use more of their sensor data to reduce energy use and maintain output, making industrial ai more practical for complex physical systems.