AI in MRO: What Singapore Airlines Gets Right — and What Everyone Else Is Missing

In February 2026, Singapore Airlines, SIA Engineering Company, and A*STAR launched two new joint laboratories. One focuses on advanced manufacturing for cabin components. The other — the one worth paying attention to — is building GenAI-enabled solutions for airline operations, maintenance planning, and operational resilience. It’s been running since October 2025 and will continue until 2028.

This isn’t a press release exercise. SIA and SIAEC have been doing this since 2019. Phase one produced real results — over 100,000 cabin components manufactured locally, a 30–50% reduction in parts lead time, AI-driven defect detection deployed across maintenance operations, and 28 local SMEs integrated into the aerospace supply chain.

They’re not talking about AI in MRO. They’re doing it. And the way they’re doing it matters.

Where SIA Is Applying AI — and Why It’s Working

The second joint lab is focused on applying advanced analytics and GenAI across three areas: detecting potential disruptions earlier, optimising operational workflows, and strengthening overall operational resilience and service quality.

In plain English: they’re using AI to spot problems before they escalate, reduce the manual coordination that slows everything down, and make the entire operation more resistant to the kind of cascading failures that ground aircraft and cost millions.

This builds on earlier work where A*STAR’s Institute for Infocomm Research and Institute of High Performance Computing developed predictive analytics that detect recurring defects and predict component failures. The technology reduces the risk of flight delays by catching issues before they reach the flight line.

But here’s what makes SIA’s approach different from the digital twin headlines and conference presentations you see everywhere else: they started with the operational layer, not the theoretical one.

The Problem Most MRO Operations Actually Have

Across the industry, the AI conversation in MRO is dominated by two things: predictive maintenance and digital twins. Both are genuinely valuable technologies. Both require enormous data infrastructure investments. And both assume that the operational communication underneath them is already working.

For most MRO shops, that assumption is wrong.

The daily reality for the majority of maintenance operations — especially independent shops, regional carriers, and parts trading companies — isn’t a shortage of sensor data or predictive models. It’s a shortage of clear, timely communication between the people who need to coordinate work.

The quote that took 45 minutes to draft because the original RFQ was missing three critical details. The AOG response that generated five follow-up emails before anyone had the full picture. The shift handover that lost context because it lived in someone’s inbox instead of a shared system.

These aren’t edge cases. They’re the daily grind that eats productive hours across every MRO operation that runs on email.

What SIA Gets Right — and What the Rest of the Industry Is Missing

SIA’s joint lab approach works because it’s solving the problems in sequence. Phase one laid the manufacturing and data foundation. Phase two is applying AI to the operational coordination layer — the workflows, the planning, the disruption management. They’re not leapfrogging to digital twins. They’re building the intelligence into the layer where decisions actually get made and communicated.

Most of the industry is trying to do it backwards. They’re investing in the sophisticated end of the stack — the sensors, the simulations, the virtual replicas — while the basic coordination that connects maintenance decisions to maintenance actions is still running on unstructured email and phone calls.

A*STAR’s optimisation technology helps SIA and SIAEC improve operational efficiency through better maintenance interval planning, workflow sequencing, and manpower allocation. That’s not glamorous. It doesn’t generate breathless LinkedIn posts about the future of aviation. But it’s where the actual efficiency gains live.

The Communication Layer Nobody’s Investing In

SIA has the resources, the government backing, and the institutional commitment to build these capabilities in-house. Most MRO operations don’t. And that’s fine — not every shop needs a joint research lab with a national science agency.

But what every MRO operation does need is a communication layer that doesn’t actively work against them.

Right now, the typical MRO email workflow looks like this: someone writes a message under time pressure, misses a critical detail, sends it. The recipient reads it, realises they need clarification, writes back. The original sender is now on a different task. The follow-up sits for an hour. The part that was needed four hours ago still hasn’t been confirmed.

SIA is applying AI to detect disruptions earlier. Most shops would settle for an email that contains the right information the first time.

Start Where You Are, Not Where SIA Is

The lesson from Singapore isn’t that every MRO operation needs a multi-year AI research programme. It’s that the sequence matters. You solve communication before you solve prediction. You fix the information flow before you model the digital twin. You make the daily coordination work before you invest in the future-state architecture.

SIA understood that. Their Phase one wasn’t AI for the sake of AI — it was practical improvements to how parts were made, how defects were detected, and how maintenance was planned. Phase two builds on that foundation with GenAI applied to the operational layer.

Most MRO teams don’t need a research lab. They need their next AOG email to land clearly the first time.

That’s not a small thing. In an industry where every hour of aircraft downtime costs thousands, the difference between a clear first email and a four-message thread to get to the same answer is real money. Real time. Real capacity freed up for the work that actually matters.