Most firms raise AI spending but see little return: Lark

Firms must ask where the value is leaking out.

Nine in 10 companies in the region are increasing their artificial intelligence (AI) spending this year, yet 60% see minimal or no return on their investment.

“In some of them, up to 40% of the technology budget is going to AI, and some firms have told me they have an unlimited budget for it,” Olivier Adam, general manager for Asia Pacific at Lark, said at the Retail Asia Summit 2026 in Singapore.

Fewer than 5% of companies report tangible results, he added. “The money is leaving the building,” Adam said, describing the spending as an experiment that has yet to pay back for most firms.

He said this is not a reason to stop spending, but a reason to ask where the value is leaking out.

“Only 3% of respondents named lack of executive sponsorship as a blocker,” Adam said. He noted that the main obstacles include talent that cannot be hired fast enough, systems that do not connect, and an inability to prove return on investment.

In a typical retailer, he said, sales data sits in the enterprise resource planning system, monthly figures in a spreadsheet on one person's computer, decisions in WhatsApp groups, and standard operating procedure checklists in a self-built app that connects to nothing.

“AI tools then see only fragments, and staff revert to running the business on gut feel.”

Meanwhile, he noted that the companies that do capture value have redesigned how work gets done.

“Firms that redesign around AI are twice as likely to capture value from the investment, with the same tools and similar budgets,” Adam said.

He reiterated that businesses must bring context together, redesign workflows so AI sits inside them, before bringing in AI agents where the business happens.

“Companies should avoid betting on a single AI model since a new one appears every few months and leading platforms will probably not be state of the art within three to six months,” he warned.  

Adam said the path has three stages—digitise, transform and augment with AI. “Each compounds the last, and value can start at stage one, so companies do not have to wait three years for results”.