Rolls-Royce’s Predictive Maintenance AI Isn’t One System. It’s Three, and a 1962 Business Decision

Key Takeaways

• Rolls-Royce’s predictive maintenance AI isn’t one system, it’s at least three distinct ones, covering different engines and different parts of the business, and most retellings of “Rolls-Royce uses AI to predict engine failures” collapse them into a single story that doesn’t quite exist. A UK C-suite reading this case study should notice the precision, or lack of it, in how the story usually gets told.
• The broadest is Engine Health Monitoring (EHM), Rolls-Royce’s own current figures put it at monitoring “over 8,000 aircraft across Civil and Business aviation” fitted with RB211 and Trent engines, evolved over more than two decades from an original service that has since had AI and advanced analytics layered onto it.
• A newer, narrower system, the Engine Vibration Health Monitoring Unit, deployed specifically on Rolls-Royce’s Pearl business jet engine family and detailed in a Microsoft case study published in April 2025, tracks more than 10,000 engine parameters and detects and prevents around 400 unplanned maintenance events a year. That figure belongs to the Pearl family specifically, not Rolls-Royce’s entire in-service fleet, a distinction the case study itself is careful about that a lot of secondary coverage isn’t.
• A third, entirely separate AI system, nicknamed Signature Analyzer, has nothing to do with engines already in service. It uses machine vibration analysis and generative AI to inspect the roughly 2 million cooling holes Rolls-Royce used to check manually every month on turbine blades during manufacturing, lifting machine utilisation by 30% and cutting erroneous scrap classifications. It is a factory quality-control tool, not a predictive maintenance one.
• The reason Rolls-Royce has invested this heavily and this specifically in monitoring engines it doesn’t operate traces back to a 1962 business model decision, not a recent AI strategy: Power-by-the-Hour, and its modern successor TotalCare, charge airlines a fixed rate per flying hour rather than for parts and labour, meaning Rolls-Royce itself, not the airline, absorbs the cost of an unplanned engine removal. For a UK C-suite, that’s the real lesson: Rolls-Royce’s predictive AI investment makes sense because its revenue model was already built to reward exactly the outcome the AI produces, decades before the AI existed to produce it.

Rolls-Royce is one of the most cited names in enterprise AI case studies, usually reduced to a single sentence: it uses sensors and AI to predict when a jet engine needs maintenance before it fails. That sentence is true, and also flattens at least three separate systems, tracking different engines, run for different reasons, into one story. Read precisely, what Rolls-Royce has actually built says less about the cleverness of any individual algorithm and more about a company that restructured its entire business model around outcomes decades before it had the AI to optimise for them.

Engine Health Monitoring: the broad, long-running foundation

The oldest and broadest of Rolls-Royce’s monitoring systems is Engine Health Monitoring, EHM, available for the company’s RB211 and Trent engine families that power the majority of the world’s wide-body commercial aircraft. Rolls-Royce’s own current EHM service page describes a system that “analyses data from every engine and every flight,” now enhanced by AI, and states that it “monitors over 8,000 aircraft across Civil and Business aviation.” Where a system fault is detected, EHM can generate a specific troubleshooting recommendation, referencing an exact Fault Isolation Manual or Maintenance Manual task, or a customised notification sent directly to the operator, rather than a generic alert requiring an engineer to diagnose the problem from scratch. Rolls-Royce sells EHM as a subscription service, with a one-year initial term followed by rolling monthly renewal, a structure that itself signals how central the service has become to the ongoing customer relationship rather than being a one-off add-on.

IntelligentEngine: the 2018 vision, and what it was actually meant to fix

Rolls-Royce formalised its broader ambition for this data at the Singapore Airshow on 6 February 2018, launching what it called the IntelligentEngine, described in the company’s own press release as the third evolution of its service model, following a shift from repair-based servicing to Engine Health Monitoring under the TotalCare brand, and now to a fully data-driven approach. The company said it expected to generate 70 trillion data points annually from its in-service fleet by the end of 2018 alone, built on foundations including R2 Data Labs, a dedicated data science unit Rolls-Royce had launched two months earlier, in December 2017, plus partnerships with Tata Consultancy Services and Microsoft, and data innovation hubs spread across the UK, US, Germany, Singapore, India and New Zealand. “Digital information is the lifeblood of the IntelligentEngine,” said Dominic Horwood, Rolls-Royce’s then-Director of Customer and Services for Civil Aerospace, in the announcement. That 70-trillion figure describes the scale of raw data Rolls-Royce expected to be collecting by 2018, not a processed or actionable-insight figure, and it’s worth reading as an infrastructure ambition statement rather than a performance claim.

The Engine Vibration Health Monitoring Unit: a newer, narrower, better-quantified system

The most concretely quantified of Rolls-Royce’s monitoring systems is also its newest and narrowest. Detailed in a Microsoft customer case study published on 1 April 2025, the Engine Vibration Health Monitoring Unit, EVHMU, is deployed specifically on Rolls-Royce’s Pearl engine family, which powers Gulfstream business jets rather than the wide-body commercial aircraft EHM more broadly covers. Built on Microsoft Cloud for Manufacturing, the system tracks more than 10,000 individual engine parameters and, across the fleets it covers, detects and prevents approximately 400 unplanned maintenance events a year, Rolls-Royce told Microsoft, avoiding what the case study describes as millions in cost from unplanned repairs and disruption. Fault resolution, previously taking days, now happens in near real-time. “By digitizing operations, we’re driving engineering innovation while reducing cost and time to market, to deliver an enhanced customer experience,” said Kaveh Pourteymour, Rolls-Royce’s Group Chief Digital and Information Officer, in the case study. Nikki Grady-Smith, the company’s Chief Transformation Officer, was more measured about how far the work still had to go: “Digital is a key enabler of our transformation. And whilst we are already seeing clear benefits, whether in engine design, operations, or customer solutions, there’s still much more to unlock.”

Signature Analyzer: the AI system that has nothing to do with engines in service

The same 2025 Microsoft case study describes a third system that is genuinely easy to mistake for part of the same predictive-maintenance story, and isn’t. Signature Analyzer uses machine vibration analysis combined with generative AI to inspect turbine blade cooling holes during manufacturing, not in-service monitoring at all. Rolls-Royce used to inspect roughly 2 million of these cooling holes manually every month, checking for defects that could compromise a blade’s cooling performance once the engine was in use. “Historically, we inspected about 2 million cooling holes per month manually. With Microsoft AI, we’re able to optimize manual inspection, dramatically reducing the processing time of each component,” said Tim Pickup, Rolls-Royce’s Digital Manufacturing Lead, in the case study. The result was a 30% increase in machine utilisation and a sharp reduction in components incorrectly classified as scrap. It’s a legitimate, well-quantified AI deployment, but it belongs to Rolls-Royce’s factory floor, catching manufacturing defects before an engine is ever built, not to the fleet of engines already flying that EHM and EVHMU are built to watch.

Why any of this makes commercial sense: a 1962 decision, not a recent one

The detail that actually explains why Rolls-Royce has invested this specifically in predicting failures in engines it doesn’t itself fly predates AI by six decades. Power-by-the-Hour, a service model Rolls-Royce pioneered in 1962 to support the Viper engine on an early business jet, charged operators a fixed rate per flying hour rather than billing separately for parts and labour when something broke. The modern version of that model, TotalCare, works the same way: airlines pay Rolls-Royce a fixed rate per flying hour, under contracts that commonly run 10 to 20 years, occasionally longer, and Rolls-Royce is only fully rewarded when engines keep flying without unplanned removals. An unplanned maintenance event under that structure isn’t primarily the airline’s cost to absorb; a meaningful share of it lands on Rolls-Royce. That single fact is why EHM, IntelligentEngine and EVHMU exist in a coherent line at all: each one is a more capable way of protecting a revenue model that already rewarded exactly the outcome, engines that don’t need unplanned maintenance, the AI was eventually built to produce.

What a UK C-suite should actually take from this

The instructive part of Rolls-Royce’s story isn’t any single accuracy figure or data volume, most of which, read carefully, apply to one specific engine family or one specific factory process rather than the business as a whole. It’s the sequencing: the commercial incentive to predict and prevent failure was built into Rolls-Royce’s contracts in 1962, more than sixty years before the current generation of AI existed to serve it well. EHM, IntelligentEngine and EVHMU are three successive, increasingly capable attempts to get better at a job the business model had already assigned itself. For a board considering its own predictive-maintenance or reliability AI investment, the useful question Rolls-Royce’s history actually raises isn’t which vendor or platform to copy. It’s whether the organisation’s own revenue model rewards the outcome the AI is meant to produce, in the specific way TotalCare rewards Rolls-Royce for engines that don’t fail unexpectedly, or whether the AI is being asked to manufacture an incentive the business itself doesn’t yet have.