• McKinsey & Company found that AI is delivering monetary gains in an analysis of 20 companies.
• McKinsey says AI adoption at those companies boosted profits by 20% after 2 to 4 years.
• The companies succeeded by focusing on key domains rather than widespread implementation.

AI might finally be paying off.

For all the hype, there's been lingering skepticism about whether AI is actually delivering financial returns. Companies have adopted all sorts of creative metrics to measure — or at least approximate — the potential gains from AI adoption. Some are tracking the number of users a particular AI tool has, or the amount of time workers are saving using AI.

To most companies, though, it's the bottom line that matters most — and McKinsey & Company says the numbers are starting to look promising.

Among a group of top-performing companies, the firm found that AI delivered returns of about $3 for every dollar invested in the tech.

"In the grand scheme of returns that you can get, that's not too shabby," Kate Smaje, a senior partner at McKinsey, said on a recent episode of the firm's podcast.

The consultancy analyzed 20 companies that had fully implemented its "Rewired" framework to understand what drives successful AI adoption and returns.

Smaje, along with her McKinsey colleagues Eric Lamarre and Rodney Zemmel, distilled the framework in a 2023 book titled "Rewired," drawing on the firm's analysis of 200 large-scale digital and AI transformations and years of client work. It lays out how companies can consistently turn digital investments into lasting business value by overhauling their talent, operations, technology, and data.

After about one to two years, the majority of companies were generating cash from AI adoption, Smaje said. A few years later — typically another two to four — the real gains started to come through, with core profit up about 20% on average.

If there's a lesson to take away, she said, it's that AI adoption is about quality, not quantity. Smaje said that about two-thirds of the cohort focused on incorporating AI into three or fewer domains.

"They're not papering AI everywhere across their organization," she said. "They're being incredibly focused in where they point the resource."