Saudi Arabia is about to operate a very large stock of new buildings and plant, and the maintenance industry is moving from fixed-interval servicing toward condition-based and predictive work. The technology is proven in industry, but new assets lack the failure history that predictive models are trained on.
Maintenance in most of the Gulf still runs on the calendar. A service contract specifies quarterly, half-yearly and annual visits, technicians attend whether or not anything needs attention, and the failures that matter happen between visits anyway. It is a system designed around labour availability rather than around equipment behaviour, and it survives because it is easy to price and easy to audit.
It is also expensive in a way that does not show up in the maintenance budget. Fixed-interval servicing over-maintains equipment that is running well, under-maintains equipment that is degrading, and produces almost no information about either. A chiller that fails in August in Riyadh does not fail because it missed a service. It fails because a bearing had been degrading for four months and nobody was measuring it.
The alternative has been proven in industry rather than in buildings. Continuous vibration, temperature and current monitoring on rotating equipment, feeding models that flag developing faults weeks before failure, is now standard practice in manufacturing and process plants. The maturing part of that market is the link between the alert and the work: Augury and MaintainX integrated machine-health monitoring with maintenance execution so that a detected fault raises a work order with the diagnosis attached, rather than an email into a reliability team's inbox. Robotic inspection is following the same path, with ADNOC deploying a heavy-duty inspection robot at a gas plant to take routine readings that previously required a person in a hazardous area.
Saudi Arabia has an unusually strong case for making that transition in buildings as well as plants. The facilities management market in the Kingdom is put at roughly $52bn to $55bn for 2026 depending on the source, and is forecast to grow faster than construction output for the rest of the decade. The asset base is new, which means it was installed with networked controls, sensing and metering already in place. The cost of retrofitting condition monitoring onto an old building is what kills most predictive maintenance business cases; on a building completed in 2025 the sensors are largely already there.
The Saudi building automation market has been valued at about $1.5bn in 2022 with forecasts near $3.3bn by 2030, and the broader building energy efficiency systems category is put at around $3bn in 2025. Those estimates come from different houses on different definitions and should be read as direction. The direction is that the instrumentation layer is being installed at scale regardless of whether anyone has decided to do predictive maintenance with it.
The obstacle is subtler than sensors, and it is specific to a new asset base. Predictive models are built on the relationship between a measured signal and a subsequent failure. That relationship is learned from failures. A portfolio of buildings commissioned in the past three years has no failure history, no baseline for what normal looks like on that specific chiller in that specific plant room, and no record of the small anomalies that later turned into something. Generic models trained on other people's equipment help, but they are noticeably worse, and a system that produces false alerts in its first year loses the technicians' confidence permanently.
The second obstacle is organisational. A predictive programme changes what a maintenance contract is buying. Fixed-interval servicing is a schedule of attendances, which is straightforward to specify, price and verify. Condition-based maintenance is an outcome, which requires the client to accept that a technician may not attend for six months and to have a way of knowing that was the right decision. Very few Saudi facilities management contracts are currently written that way, and the ones that are tend to sit inside single-owner portfolios where the operator and the asset owner are the same party.
The third is the handover problem that runs through the whole operating phase. Predictive maintenance needs an accurate asset register, equipment nameplate data, commissioning baselines and control system access. On a great many projects that package arrives incomplete or as unstructured documents, and the operator's first year goes on reconstructing it. A predictive programme cannot start until it is done.
Where this is heading is reasonably clear. Critical rotating plant, cooling systems, lifts and electrical distribution will move to condition-based regimes first, because their failures are expensive, their signals are well understood and they are already instrumented. Fabric, finishes and the long tail of small equipment will stay on the calendar, because monitoring them costs more than fixing them. The industry conversation is converging on the interface between the two, which is why coordination between design, installation and facilities management now appears on the agenda at events like Big 5 Construct Saudi rather than only in reliability journals.
The Kingdom will spend the next decade learning how its new asset base actually fails. The organisations that instrument it properly now will own that knowledge, and it is not transferable.