Is AI a bubble that’s about to burst?

Earlier this year, at a time when multinational organisations were congratulating themselves for having survived a year of tariff-induced chaos, it was becoming clear that the apparent resilience of the global economy is being shored up by something very different: AI. Research by McKinsey, published in March this year, estimated that AI-related trade accounted for over 30% of global trade growth in 2025. Fear is increasing that investors are putting all their eggs into a single, AI basket.

Is this changing how clients think about AI? After years of asking clients questions about how they plan to deploy the technology, we used our Quarterly Client Survey in Q2 to ask a much more basic question: Do they think AI is a bubble that will burst? In other words, is the value of AI, like that of the first attempts to monetise the internet 25 years ago, being overstated? Clients wouldn’t know the answer, any more than technologists or economists would. What we wanted to understand were the factors influencing their views.

The simple answer

The simple answer to our simple question is that 54% of clients think AI is a bubble, 43% don’t think it is, and a miserly 3% are unsure.

 

The complicated answer

Why do they think this? Not seeing any clear patterns from a sector, geographic, or functional perspective, we’ve analysed the “yes” camp (who think AI is a bubble about to burst) and the “no” camp (who think the opposite) in terms of their responses to a wide variety of other questions asked in the same survey. The greatest divergence between these two groups falls into three areas:

• Their attitude to the world they work in.
• Their priorities as an organisation
• The way in which they work, or have worked, with consulting firms

Client attitudes to the world they work in: Executives in the “yes” camp are more likely to say their confidence has been severely damaged by geopolitical and macroeconomic turmoil than those in the “no” camp. Consistent with this, their confidence is more likely to have been damaged by recent events, such as the conflict in the Middle East, than that of the “no” camp. Unsurprisingly, those in the “yes” camp are also more pessimistic about the length of the “poly-crisis”.

We can surmise that the “yes” camp, having been more significantly rocked by geopolitical turbulence, feel more pessimistic in general. They anticipate a continuation of this negative environment and fail to see AI as the silver bullet it is lauded to be.

Organisational priorities: The “yes” and “no” camps agree on some things: Both, for example, say that better technology infrastructure and improved productivity are near the top of their corporate agenda. However, 31% of the “no” camp say that AI is one of their three most strategic priorities, compared to only 23% of the “yes” camp. It makes sense that those that back AI to deliver on its expected value are more likely to be prioritising it in their own business plans.

Almost every organisation we survey at the moment says that resilience is critical, but they have different views about the best way to achieve this. Thirty-six percent of people in the “no” camp say that AI is one of the main ways to be more resilient, compared to only 22% of people who think AI is a bubble that’s about to burst. Instead, the “yes” camp is more likely to say that better data and analysis, together with recruiting people with deep expertise in their industry will help. In a similar vein, only 18% of the “yes” camp say that they’re using AI to help them manage risk more effectively, compared to 36% of their equivalents in the “no” camp.

A couple of things stand out here. Firstly, it feels counterintuitive that a proportion of people who think that AI is a bubble that’s about to burst, roughly 20%, also think that it will help them increase resilience and manage risk more effectively. What we hear in client interviews is that these executives recognise the long-term potential of AI—they just don’t see it delivering this right now.

Secondly, there’s tentative evidence in our research suggesting that people in the “yes” camp, that see AI as a bubble set to burst, may have had a more negative experience with AI. They feel, for instance, that the technology has not had any positive impact on their work yet: They’re more likely to say that both day-to-day and strategic change are harder than they were a year ago—and while this isn’t directly in reference to the impact of AI, we can glean that they are not yet seeing the productivity gains the technology has been predicted to deliver.

Working with consultants: Clients turn to consulting firms to help them feel more in control in an increasingly unreliable world. Forty-one percent of the “no” camp says that one of the best ways to do this is through using consultants to speed up their adoption of AI, but only 20% of the “yes” camp agree, which speaks to a degree of reticence in their willingness to embrace the technology. In contrast, the more cynical “yes” camp is more likely to want consultants to help them feel in control by providing ad hoc expertise to their staff, highlighting new risks, and helping them understand the implications of political and economic instability—they don’t automatically assume that AI is the solution.

Interestingly, people in the “yes” camp are more likely to have used AI support to facilitate fundamental change to their business model, while the “no” camp are much more likely to have focused their use of AI support on everyday activities. This suggests that those that have attempted to leverage AI to revolutionise their businesses are less convinced of its ability to do so—a finding that should be alarming to firms, most of whom are making it a core part of their offerings. Furthermore, this “yes” camp seem unconvinced that firms can add value in this space: They have had greater exposure to firms’ own consulting tools (77% have seen consulting firms deploy AI-enabled tools to a great extent, compared to 55% among the “no” camp) but 68% believe they know more about AI themselves than consultants do. What’s even more concerning for firms is that these unconvinced clients in the “yes” camp are the more important market for them: They are more likely to be heavy users of consulting services already and to expect to spend more in the future compared to the “no” camp.

What does this tell us?

Comparing the attitudes of the “yes” and “no” camps has led us to hypothesise that:

The “yes” and “no” camps have very different business outlooks and priorities. The “yes” camp is more sceptical about the value of AI and are more likely to turn to solutions that are not entirely technology dependent.

The “yes” camp has had more exposure to AI-related consulting support: They’re more likely to have used AI consultants to make fundamental changes to their business and seen them make use of AI-enabled tools—but think that consultants know less about AI than they do. Although they’re still expected to spend more on consulting services in the future, they may be less likely to use them for AI-related work.

Less may be more: Consulting firms should re-consider their approach to deploying AI with clients. Starting small (as we’ve written about before in this blog) and focusing on everyday improvements will make clients see AI as more than hype. Taking a business-led approach to using AI will resonate more with those that see AI as a bubble set to burst (hint: technologists aren’t always the best people to talk about technology). Avoid making grandiose claims that your in-house tools will revolutionise their businesses—set expectations that you know you can live up to, or else risk disappointing your clients.

What can firms do next?

Build a clearer picture of your clients’ needs and pain points around AI through our in-depth surveys and conversations with senior decision makers. Our research blends this powerful information with market sizing data, to give you a complete view of the opportunities available. To find out more about how our insights can shape your strategic decisions, contact us.