Most companies are not doing AI. They are buying AI.
Very different hobby.
Deloitte’s new State of AI report is basically a 41 page obituary for corporate theatre. Access is up 50% in a year, from under 40% of workers to around 60%, yet fewer than 60% of those with access use it in their daily workflow. So we have managed to turn the biggest technology shift in years into another unused enterprise licence.
Then it gets better. Only 25% of companies have moved 40% or more of their AI experiments into production, but 54% expect to get there in the next three to six months. Of course they do. Every bad strategy meeting ends with “next quarter will be transformational.”
The most revealing stat in the report is this: 74% hope AI will increase revenue. Only 20% say it is doing that today. That is the whole story. Most firms are using AI to make existing mediocrity slightly faster, not to create a better business.
And while everyone is busy posting thought pieces about the future of work, 84% of companies have not redesigned jobs around AI at all. Not redesigned the roles. Not redesigned the workflows. Not redesigned the operating model. Just added a tool and called themselves innovative, which is a bit like buying a Peloton and announcing you’re an athlete.
The really reckless bit is agentic AI. 74% plan to deploy it within two years. Only 21% have mature governance for autonomous agents. So the plan, apparently, is to let software make decisions before anyone has worked out who is accountable when it goes sideways. Splendid.
One more thing private equity should pay attention to: 77% of companies now factor the country of origin of AI solutions into vendor selection. Sovereign AI is no longer a government issue. It is a commercial one. If your portfolio strategy assumes software is borderless, that assumption is starting to look a bit 2023.
The winners here will not be the firms with the most pilots, the most licences, or the loudest LinkedIn posts about copilots.
They will be the ones that redesign work, build governance before scale, and actually use AI to change the economics of the business.
Everyone else is just attending the conference and calling it value creation.
