Applied Computing Raises $20M to Build an AI Model for Oil and Gas Plants

Applied Computing has raised $20 million in Series A funding to expand its AI model for oil, gas and petrochemical plants, betting that operators will pay for software that can understand an entire facility instead of isolated streams of industrial data. The London-based startup says its platform, Orbital, can help energy companies detect anomalies, test scenarios and predict plant behavior far faster than conventional workflows.

The round was led by engineering and construction heavyweight KBR, with Databricks Ventures also taking part. The funding lands as Applied Computing says it has already moved from stealth to double-digit millions in annual recurring revenue in less than 18 months, a sign that industrial AI is finding an early commercial foothold in one of the most complex sectors in the economy.

Why this funding round matters

Applied Computing is not trying to sell oil and gas companies another dashboard. It is building what it describes as a foundation model for whole facilities, designed to fuse live sensor readings with engineering documents and the physics and chemistry of industrial operations.

That approach targets a long-running problem in energy operations: companies collect enormous volumes of data, but much of it remains difficult to use together in real time. According to the startup, operators typically make decisions using less than 8% of the data available to them because information is fragmented across systems and teams.

If Applied Computing can prove that its system works reliably at scale, it could reshape how plants are monitored, optimized and maintained. That would matter not only for cost and efficiency, but also for safety, uptime and energy consumption across some of the world’s most capital-intensive assets.

What is Orbital?

Orbital is Applied Computing’s core product, and the company says it is built differently from a standard large language model. Rather than predicting the next word in a sequence, Orbital combines three layers of intelligence: a time-series model for live sensor data, a physics-based model for how a plant should behave, and a language model for reading and connecting technical information.

The goal is to build a digital understanding of what is happening inside a facility at a given moment. In practical terms, that means the system can watch changing signals from pumps, valves, compressors and processing units, interpret them against engineering constraints, and help operators reason about possible causes and consequences.

Applied Computing says Orbital can also run simulations. Technicians can ask how one change in a facility might ripple through the rest of the plant, giving them a faster way to test operational decisions before applying them in the real world.

How Orbital is supposed to work

The company says the model is designed to connect three sources of information that are usually kept separate in industrial settings.

• Sensor data: readings such as temperature, pressure, flow, velocity and viscosity.
• Engineering and operational documentation: plant designs, procedures and maintenance records.
• Physics and chemistry: the underlying rules that govern how equipment and materials behave.

By combining those inputs, Applied Computing argues that Orbital can help identify anomalies, trace likely causes, and test whether a fix in one part of a plant could create trouble elsewhere.

Callum Adamson, the company’s co-founder and chief executive, said the central challenge is not simply collecting data, but making different information sources work together in real time. He characterized that integration as the key technical hurdle the startup is trying to solve.

Adamson said the real problem is getting the plant’s data streams, documentation and scientific rules to communicate quickly enough for operators to act on them.

What kind of customers is Applied Computing serving?

Applied Computing is focused on upstream oil and gas, downstream refining and petrochemical operations. Those sectors are especially data-rich, but also highly fragmented, with many assets, software systems and engineering disciplines involved in every operating decision.

Adamson said the product is already in use at several large publicly listed companies across those segments, though he did not identify them. He also said the startup is working with a major U.S. upstream operator and expects to announce a partnership with a European oil major in the coming weeks.

The company’s existing partnerships suggest it is trying to move from pilot-style deployments into broader industrial integration. Its partners include Indian energy company Wipro and KBR, which has folded Orbital into its INSITE 3.0 digital platform for energy projects and is using it in ammonia production.

Where the company claims it saves time

Applied Computing says Orbital is designed to accelerate tasks that traditionally take teams days or even weeks.

The startup says the system can flag an abnormal condition, investigate the likely reason behind it, and evaluate the side effects of a proposed fix in a matter of minutes or less. Adamson argues that this compression of troubleshooting time can help operators cut energy use while preserving output.

That pitch is especially relevant in industries where an unplanned outage or inefficient process can quickly translate into large financial losses. Even small improvements in uptime or energy consumption can be valuable when applied across a plant that runs continuously and costs millions to operate.

How did Applied Computing grow so quickly?

The company says it went from operating in stealth mode to generating double-digit millions in annual recurring revenue in less than 18 months. For an industrial startup, that is a notable commercial ramp, especially in a market where sales cycles are often slow and implementation can be technically demanding.

Applied Computing has not disclosed the number of customers it serves, but the revenue claim suggests it has moved beyond early experimentation. In sectors like oil refining and petrochemicals, that usually means the software has found use cases where operators can justify the operational risk and integration effort.

That traction also helps explain why KBR took the unusual step of leading the Series A. For an engineering company with deep industrial relationships, an equity bet in a startup is often as much about distribution and strategic access as it is about financial return.

Why is KBR backing the startup?

KBR’s investment gives Applied Computing more than capital. The startup says the partnership provides access to operational data, technical expertise and introductions to potential customers, all of which are valuable in a field where trust and credibility matter as much as product performance.

KBR is already integrating Orbital into its INSITE 3.0 platform, which is used in energy project environments. The two companies are also applying the software in ammonia production, a process that depends on complex industrial control and tight operational margins.

For KBR, the relationship may help differentiate its digital offering. For Applied Computing, it may provide a route into larger deployments and a way to embed itself more deeply into the workflows of industrial customers.

Adamson said the strategic value of the KBR relationship lies in the company’s access to process knowledge, plant data and customer introductions.

What is the competitive landscape?

Applied Computing is entering a crowded industrial software market that already includes established vendors and newer AI-focused rivals. The company’s challenge is not simply to prove that its model works, but to prove that it is meaningfully better than existing tools and easier to adopt in real plants.

Among the better-known competitors are AspenTech and AVEVA, both of which sell simulation, optimization and process-modeling products aimed at industrial operations. On the data and workflow side, companies such as Cognite and Seeq help operators structure industrial data and build analytics-driven processes around it.

That means Applied Computing must compete on both technical performance and deployment speed. Industrial buyers are cautious, and switching costs can be high. To win, a new entrant typically needs to demonstrate not only better predictions, but also faster integration, better outcomes and clear return on investment.

How does Applied Computing say it is different?

The startup’s argument is that the hardest part of its business is artificial intelligence research, not access to data or knowledge of the energy industry. Adamson says the model itself is the moat, because assembling a team capable of building and training something like Orbital is the main barrier.

He pointed out that the operational data inside refineries and other facilities is generally private, while simulated data cannot fully recreate the complexity of a live plant. That gives companies with real deployments an edge, because every new installation can potentially improve the model further.

Adamson also framed the hiring challenge as a competitive advantage. In his view, the best AI researchers are more likely to be attracted to a company focused on frontier model work than to a traditional energy company.

Adamson argued that the startup’s competition is not a lack of industrial knowledge, but the difficulty of assembling top-tier AI researchers to build a model strong enough to match Orbital.

Why industrial data is so hard to use

Energy plants generate huge amounts of information, but the data is often messy, siloed and difficult to interpret. Sensors may feed into one system, maintenance logs into another and engineering diagrams into a third. The result is that plant operators can see pieces of what is happening, but not the full picture.

That fragmentation is especially problematic in facilities where a small change can have cascading effects. A temperature shift in one section may alter pressure elsewhere, which can affect output, efficiency or safety. Traditional analytics tools often struggle to link those signals quickly enough for real-time decision-making.

Applied Computing is betting that a model built around the physical structure of the plant can do better than software that treats the facility as a generic data problem. Its pitch is that industrial AI should understand the plant as a living system, not just as a collection of dashboards.

What makes the market attractive?

There is a large commercial opportunity in helping energy companies make better use of the data they already collect. Even modest gains in reliability, efficiency or throughput can produce meaningful savings when applied at refinery scale.

At the same time, the market is hard to penetrate, because customers demand strong evidence before allowing software to influence sensitive operations. That means startups can succeed if they are precise, technically credible and able to show results in real environments.

Applied Computing appears to be aiming squarely at that opportunity, with a product that combines predictive maintenance, process optimization and simulation in one system.

Where will the new funding go?

Applied Computing says the $20 million will be used to expand internationally, hire more researchers and engineers, and pursue further deployments with energy customers. The company is also opening a new office in Houston, adding to its existing base in London and operational center in Bengaluru.

The Houston move is strategically sensible. The city is one of the most important hubs for North American energy, and it places the company closer to two existing customers in the region. It also gives the startup a stronger U.S. presence as it looks to deepen relationships with large industrial buyers.

Adamson said the company is also planning an expansion into the Middle East, another region where large-scale oil, gas and petrochemical assets create demand for operational optimization tools.

Timeline: Applied Computing’s rapid rise

The startup’s progress has been unusually fast for industrial software. The following timeline captures the main milestones disclosed by the company.

What does this mean for AI in energy?

Applied Computing’s funding round is another sign that enterprise AI is moving deeper into specialized industries where the value proposition is concrete and measurable. In energy, the most promising use cases tend to be the ones that save time, reduce waste, improve uptime or help operators avoid costly mistakes.

The challenge is that these environments are much less forgiving than consumer apps or generic business software. A plant model must be accurate, explainable enough for engineers to trust and robust enough to function under changing conditions.

If Orbital proves successful, it could become a template for a broader class of industrial foundation models built around physics-heavy sectors such as chemicals, manufacturing and power. If it falls short, it will still have helped define the standard against which future systems are judged.

What happens next?

Applied Computing’s next phase will likely hinge on two things: whether it can turn its current deployments into larger contracts, and whether it can maintain product performance as it expands into new regions and plant types.

The company’s immediate priorities include hiring, expanding internationally and completing more integrations with large energy operators. The planned European oil major partnership, if announced, could be an important validation point. So could the company’s continued progress in North America and the Middle East.

For now, the startup has done enough to attract major strategic capital and industry attention. In a sector where software adoption often moves cautiously, that is itself a significant milestone.

At a glance

• Company: Applied Computing
• Base: London, with operations in Bengaluru and a new office in Houston
• Product: Orbital
• Funding: $20 million Series A
• Lead investor: KBR
• Also involved: Databricks Ventures
• Target sectors: Oil, gas, refining and petrochemicals