Before AI Can Help, the Data Has to Work: Lessons from a 140,000-Person Manufacturer

Key points

• Jabil, a global manufacturer with over 140,000 employees and more than 100 sites across 30-plus countries, is standardising on SAP Integration Suite, a platform that connects separate software systems so they share data automatically.
• About 25 years of disconnected, site-specific tools and manual workarounds created gaps that slowed decisions and disruption responses.
• Jabil's IT director frames the core rule simply: any innovation without simplification adds more complexity.
• AI tools such as predictive supply-chain forecasting are next on the roadmap, once the shared data foundation is stable.

There is a common temptation in big organisations: buy a new technology to fix a problem, then buy another when the first one creates its own problems. After 60 years in business and expansion to more than 100 factories worldwide, Jabil, a contract manufacturer based in St. Petersburg, Florida, that builds products for more than 400 global brands, found itself deep in that trap.

Each site had its own tools, its own spreadsheets, its own ways of recording what was happening on the factory floor. Data lived in separate pockets, invisible to anyone outside that location. When a disruption hit one part of the supply chain, the rest of the network couldn't see it quickly enough to respond.

"We had limited ability to see issues early across plants, across regions," says Harish Manohar, SAP IT Director at Jabil, speaking on the Business Lab podcast, produced by MIT Technology Review in partnership with SAP. "This complexity created data silos, and in a way delayed decision-making."

Why didn't Jabil just add smarter tools?

Adding new technology on top of broken plumbing makes the plumbing worse. That was the lesson Jabil drew from its own history.

Manohar describes the company's response as a "simplify-first, then-innovate mindset." Standardise the underlying processes before bolting anything new onto them. Jabil chose SAP, the German enterprise software company, as its core digital platform, and is using SAP Integration Suite as the connective layer, routing as many processes as possible through one consistent system.

Not every site is there yet. Some carry legacy software dating back decades, while others operate in regulated industries with their own compliance rules. Standardising across all of them means changing how people work without stopping the factories that never close.

What does this have to do with AI?

Everything, in Manohar's view. AI tools, whether they forecast demand or plan production runs, are only as good as the data they read. Feed them inconsistent or outdated information and their answers are unreliable at best, costly at worst.

"The backbone of any modern organisation is data," Manohar says. "Before organisations can optimise or apply AI, data needs to flow across systems without gaps."

Our 2 September story on IBM and Confluent's partnership to run AI predictions on live data streams showed the same dependency from a different angle: the inference engine is only half the problem; the other half is what you feed it.

With a cleaner foundation taking shape, Jabil is now exploring predictive supply-chain insights and AI-driven planning tools. Those come second. The plumbing comes first.

The broader point reaches well beyond manufacturing. Any organisation, from a hospital network to a retail chain, faces the same sequence: clean, connected data is the prerequisite, not an optional extra. Skipping it and jumping straight to AI doesn't speed things up. It moves the problem somewhere harder to see.

Watch for: If your own workplace is being sold an AI tool as a fix for inconsistent reporting or missing data, ask what happens to the underlying data problem first. A good answer goes to the data. A bad answer goes straight to the demo.