When a pump starts overheating at an offshore wind platform, the old workflow goes like this: the anomaly shows up in some sensor log, someone files a ticket, a technician waits for transport, and three days later someone finally looks at it. With the integration SAP and ANYbotics announced this week, the sequence shortens considerably. The ANYmal robot detects the hot spot during its autonomous inspection run. The thermal data flows into SAP's asset management system. A work order is auto-generated and dispatched to the field team that can fix it. Nobody files anything. The maintenance ticket writes itself.
ANYbotics, a Swiss robotics company spun out of ETH Zurich in 2016 and led by Dr. Péter Fankhauser as co-founder and CEO, has been building the ANYmal — a four-legged inspection robot — for exactly these environments: the refineries, cement plants, offshore platforms, and processing facilities where stairs, heat, corrosion, and unreliable connectivity make wheeled robots impractical and human access expensive and dangerous. The robot climbs stairs autonomously, navigates cluttered industrial terrain, and carries a sensor stack that includes thermal imaging, ultrasonic leak detection, and acoustic anomaly monitoring.
The Northern Lights carbon capture and storage facility off the Norwegian coast — a joint venture between Equinor, TotalEnergies, and Shell — runs ANYmal missions without any personnel on site for routine inspection rounds, per the SAP press release. Operators dispatch the robot from a control room 30 minutes away, review the results, and send people only when the data says it's necessary.
More than 200 ANYmal units are now in productive operation worldwide, according to SAP's March 2026 press release, performing thousands of inspections weekly across oil and gas, mining, power, utilities, and metals industries. Its customers include Equinor, SLB, Siemens Energy, GE Vernova, Novelis, and Outokumpu — drawn from the customer roster in ANYbotics' Climate Investment announcement.
Nicole Zingg, director of technology partnerships at ANYbotics, put it directly in the SAP press release: "If customers are using SAP, SAP is where ANYbotics needs to be native."
One metals company running ANYmal inspections of furnace equipment scaled from initial deployment to full operational rollout in a few months, according to a customer testimonial published by Robot Report. The company is now running 18 automated missions every day with a single robot and is on track to save over $1 million by catching equipment failures before they cause unplanned downtime. The impact is not hypothetical — it shows up in the maintenance budget.
"Inspection robotics is really about data that is consistent and trustworthy," Zingg said. "ANYmal has to put data in the SAP system, just like human team members."
That comparison is the point. The integration, built on SAP's Business Technology Platform, is designed around the idea that ANYmal should slot into the same operational workflows as a human technician. The first integration point is SAP Field Service Management — the same dispatch system a plant manager uses to send a heating technician to a failing heat exchanger. The manager types the same work order, routes it to the same queue, and gets the same result back: inspection data and a recommended action. The robot executes its mission autonomously and reports directly into SAP. No portal, no manual upload, no waiting for someone to summarize what the robot saw.
The implication is not just efficiency. It is a different model for how robots enter industrial operations. Rather than deploying a robot and building a separate analytics layer to make sense of its output, the robot is simply another worker in an existing system of record. For plant operators already running SAP for asset management, the integration removes the friction of building custom pipelines between a robot vendor and their enterprise software. That is the sales argument for both companies.
The partnership fits within SAP's Project Embodied AI, an initiative to extend business AI beyond software workflows into physical operations — enabling robots to act on enterprise context rather than just sensor data. The division of control is explicit: ANYbotics controls what the robot physically does; SAP controls the business context — work orders, asset data, maintenance priorities. The robot does not decide what to fix. It reports what it found, and SAP routes the information to whoever needs it.
That boundary is where the interesting questions live. An inspection robot that generates data is a monitoring tool. A robot that generates work orders that trigger human actions is a different kind of actor in the maintenance workflow. One offshore wind customer SAP cites used ANYmal to eliminate months of personnel deployments at a remote platform — not because the robot fixed anything, but because the data it collected beforehand meant every site visit had a specific purpose instead of being a fishing expedition. The human expert still goes. The robot decided when and why.
ANYbotics is backed by more than $150 million in total funding, including a recent investment from Climate Investment, a specialist investor in industrial decarbonization. The round also included Aramco Ventures, Bessemer Venture Partners, NGP Capital, Qualcomm Ventures, and TDK Ventures. The company employs roughly 200 people across offices in Zurich and San Francisco.
The next product milestone is ANYmal X, which ANYbotics calls the world's first Ex-certified legged robot — designed for explosive atmospheres in oil and gas facilities where any electrical equipment must be certified not to ignite flammable gases. Customer deliveries are planned for 2026, per the Climate Investment announcement. If the certification holds up, it opens a category of industrial environments that have been effectively off-limits to autonomous inspection robots: the zones where a spark from conventional equipment could be catastrophic.
The partnership announcement is a soft launch — integration details, not a completed enterprise rollout — and the named case studies are selective. But the model is specific enough to evaluate. The question is not whether legged robots can navigate a plant. They can. The question is whether enterprise software vendors will treat robots as operational employees — with dispatch addresses, work queues, and performance metrics — rather than as sensor platforms that feed dashboards. SAP has answered yes. The results from metals plants and offshore platforms will determine whether the model holds.
