Rob Hornby
Co-Chief Executive Officer - London
The Supremes famously sang that, “You can’t hurry love”.
You won’t find many songs written about waiting for technical maturity in hype cycles, but we can usefully apply the Supremes’ relationship advice.
To make the point, I want to begin with the “Digital” innovation curve that entered a critical new phase in the early 2010s. Digital computing dates to the early 1940s, so the suggestion that we were moving from an analogue world was always a false premise. However, in common with other ill-defined movements, something important was happening, even if it was hard to pin down.
The confluence of smartphones, cloud, big data, social media, agile, broadband, and a focus on user experience slowly changed our relationship with technology and then each other.
New types of business emerged, from platform-based fintechs to SaaS providers like Salesforce, Zoom, Snowflake, and Workday.
For technologists, Digital enabled a move from the back of the house into sneaker-wearing relevance all over the business. Once through that transition, the digital tribes looked with disdain at any remaining “IT” groups.
However, “Digital” imperatives were relentlessly over-hyped, and companies spent unwisely. Valuations lost touch with reality. The human experience of hyper-connection turned out to be very mixed, especially for young people. Big data was an ocean boil, usually abandoned or heavily pruned after a year. Cloud became very expensive. Most digital projects never made it past the front end (and “IT” was mysteriously unhelpful).
But why bring all this up now?
Well, it would be easy to think that Digital is old news – replaced by AI. Not so. I recently discovered that interest in digital topics rose dramatically in 2025 to levels not seen since 2010.[1]
On closer inspection, it turns out that when the hype-ecosystem looked elsewhere, Digital kept going on the ground. Technologists and business leaders took the maturing offerings and embedded them deeply and effectively in their enterprises.
This dedication has paid off in a period of sustained disruption. Supply chains have been reconfigured using digital twins. Pricing has been optimised by proven analytics platforms. Business model changes have been enabled by more flexible front ends and configurable rules engines.
Digital’s peak value was delivered after we stopped looking.
In contrast, generative AI is still early in the hype cycle, and now we are seeing growing evidence of problems and limitations. Recent MIT research has gone viral, showing that 95% of generative AI pilots in their sample did not produce an economic payback. It has also become clear, theoretically and practically, that, even though hallucinations can be reduced, they cannot be engineered or scaled out of existence.
As a result, some commentators are ominously predicting an AI bubble burst, sending stock markets into jitters.
Early generative AI promises and claims were ludicrously overstated (AGI, the end of scarcity, human labour replacement, even immortality), so the emerging messy and diminished reality should not have come as a surprise. But we are at the end of the beginning, not the beginning of the end.
Just as in the Digital era, I predict AI will mature on the ground, leaving society, business, and individual lives transformed. Many of the changes will be for the better, but unforeseen problems will also arise without simple solutions. I expect something like quantum computing, robotics, or biotech will already have triggered the next hype-cycle; so, as with Digital, our attention is likely to be elsewhere when AI’s value finally peaks.
The big challenge is navigating the liminal space between now and then. To that end, I have three simple suggestions arising from this long view:
Firstly, leaders should resist false urgency. AI is not an immediate existential issue for most businesses, and many early adopters will pay a heavy price for moving too soon. Second-mover advantage is the optimum strategy for many organisations that can afford to build internal capability incrementally and wait for someone else to crack the code.
Secondly, be clear about what is mature and what is still flaky. Machine learning is a proven technology, and measurable benefits can be obtained quickly from it. Generative AI already provides ‘bottom-up’ productivity enhancements in content creation, coding, and conversational tasks (the three “C”s), but cannot yet orchestrate activities that mimic a real job. This means top-down AI-driven organisational transformational strategies are two years early. Although developing rapidly, agents are still nascent and pose complex governance issues around autonomous AI actions.
Thirdly, find somewhere to talk openly about these matters. Whether it is a trusted advisor, a peer relationship, or an off-the-record group, leaders need a forum for genuine dialogue. The relentless pressure to show AI progress makes this hard in any public setting, and median progress on adoption is way behind the rhetoric.
The bottom line is, as the Supremes already told us, some important things can’t be rushed, and I believe AI is very much one of them.
Patience may be the counterintuitive unlock for the long AI journey ahead.
If you'd like to listen to an audio version of this article, please click here.
[1] Based on Google Trends search analysis.