---

• As tech leaders are in a rush to keep up with the AI boom, worries about an AI bubble bursting are coming to the fore.
• Hyperscalers such as Microsoft may be slowing down investment in AI.
• Some midmarket firms are running their finance functions on outdated legacy systems which are not AI-ready.
• Pragmatism through more measured AI adoption is a better way to go.

---

Artificial intelligence (AI) dominates the business agenda today. One moment it’s an immediate threat to jobs and trust, the next it’s heralded as a miracle technology that will save money, unleash productivity and transform industries overnight.

And then comes the classic sales pitch: “If you don’t have it, what are you waiting for?”

This hype is stirring up a dangerous cocktail of anxiety and AI FOMO (the fear of missing out). Tech vendors are racing to market with flashy AI add-ons. Executives – worried about being left behind – often feel pressured to act quickly, rather than strategically, before asking whether their organisations are truly ready.

But while AI will change the way we live and work, it won’t happen overnight. As someone who ran a scale-up during the dot-com bubble, I can tell you that this feels eerily familiar.

The ghosts of 2000

In the late 1990s, there was definitely a period when everyone lost the plot thinking that the internet would instantly revolutionise the world. Did it change the way we work and do business globally? Yes. Did it happen overnight? Absolutely not.

We’re now watching this same cycle play out, in real-time, with AI. Greed and fear of missing out are fuelling a frenzy. Billions of dollars have been poured into start-ups promising the world, yet meaningful, sustainable results are proving much slower to materialise.

However, too many over-45s in business today seem to have forgotten the dot-com crash, while the younger generation of business leaders haven’t lived the experience of it at all. Together, that makes us vulnerable to repeating the same mistakes.

Warning signs from investors

The investor community is on edge, too. Artificial intelligence has attracted extraordinary sums of capital – 50 per cent of venture dollars in the first half of 2025 went to AI start-ups, according to CB Insights.

Yet the likes of Goldman Sachs are now warning of a looming slowdown in AI investment, particularly from major hyperscaler firms such as Microsoft, Amazon, Meta and Alphabet. Sundar Pichai – head of Google’s parent firm, Alphabet – is the latest to weigh in, warning that no company would be safe if an AI bubble were to burst.

The slowdown is no doubt linked to mounting fears of overly optimistic growth/ROI ambitions. The same research note from Golman Sachs claimed that many “Phase 3” companies – those expected to be benefiting from AI’s ripple effects by now – have yet to show any meaningful earnings impact. Not surprising, when a recent MIT report found that 95 per cent of enterprise AI pilots are failing to deliver rapid revenue acceleration. That alone should temper expectations.

The parallel to 2000 is uncomfortably clear: exuberance is outpacing substance.

Yet, vendors are upping the ante

As a UK tech startup leader, I see the hype cycle play out daily. AI is an incredible opportunity for innovation, but it’s also become the latest marketing badge of honour.

We’re in the accounting software space and are now seeing competitors making some very bold claims for artificial intelligence. Some are freely applying the label ‘agentic AI’ to things we just call automation. Others claim AI can breathe new life into legacy systems. But layering AI onto outdated infrastructure is like slapping a plaster onto a broken leg.

Even high-profile companies have fallen prey to the AI propaganda, getting miles ahead of themselves. Klarna, for example, rushed into AI with great fanfare, laying off 700 roles only to rehire many of them after the fact. It’s a stark reminder of what happens when organisations believe tech vendor hype, or get swept along by FOMO, rather than focusing on readiness.

The truth is that adoption has been bumpy. Some early projects have failed to show a clear return. That doesn’t mean AI isn’t delivering value – it simply shows that most organisations are still building the structures needed to support it effectively.

The overlooked midmarket

Adding to that, much of the AI conversation tends to be centred on global enterprises with vast budgets, chief AI officers, and dedicated AI teams. But what about the midmarket, so often overlooked in these conversations, despite making up a large chunk of the UK’s economy – 60 per cent of employment and nearly half of turnover – for that matter?

Our own intel tells us that there are tens of thousands of midmarket organisations still running their finance functions on outdated legacy systems from the noughties, stitched together with manual spreadsheets. Just recently, we saw the UK’s National Crime Agency exposed for this very issue. These systems are cumbersome, siloed, and nowhere near AI-ready.

For them, the priority shouldn’t be rushing into generative AI pilots or looking into agentic AI stacks. They need a powerful cloud core and clean data structures first and foremost – the foundation on which any effective AI tool can be layered. Without that groundwork, attempts to deploy AI are premature at best, and costly failures at worst.

What’s really needed: pragmatism

None of this is to say AI is failing. Quite the opposite. It’s already proving its worth in many areas, from streamlining workflows to improving customer support and product innovation – including within our own organisation.

As a tech leader in 2025, I’ve worked closely with my c-team to roll out an extensive AI programme within our own business, utilising AI for operational efficiency, support, and product enhancement.

That said, I don’t advocate pushing AI onto customers that aren’t ready for it. Instead, as a business, we focus our efforts on ensuring our customers have the correct foundational elements and provide a programme of migration that will stand the test of time.

I believe AI will ultimately prove significantly more transformative than the internet. But just as the internet’s potential unfolded over decades, AI’s revolution will take years, not months.

Slower adoption isn’t a sign of failure. It’s a sign that organisations are waking up to the work that must be done first. The real opportunity now is to combine excitement with realism and to balance innovation with the infrastructure that makes it sustainable.

Companies are realising that meaningful AI adoption depends not just on bold ambition, but on getting the fundamentals right: modern systems, integrated data, and correct processes – including security.

If businesses, particularly SMEs, take a pragmatic approach – laying solid foundations before layering in advanced capabilities – then they will be well placed to harness AI when the hype inevitably fades and the real value begins to emerge.

Will AI go the way of the dot-com bubble?

Not immediately. But in time, yes, a correction will come. The winners will be those who avoided the frenzy, ignored the FOMO, and took a measured approach.

That was true during the dot-com bubble. It will be just as true in the AI era. AI’s potential is vast – but it rewards preparation, not panic. AI will change the world – but only for those who build the foundations strong enough to carry it.

For more on how to adopt AI in a sustainable way, head over to our sister site Information-Age to read Eight steps to a successful AI implementation.

Lyndon Stickley is the CEO of iplicit.

Read more on technology

6 top decision intelligence solutions for 2026 – The right decision intelligence software for your organisation is the one that makes it easiest for your team to unlock the value that lies in data, helping to improve decision-making across the enterprise

6 top phishing simulators for training employees in 2026 – 6 top phishing simulators for training employees in 2026

The best business VoIP providers for UK businesses – We’re diving into seven of the best Voice over Internet Protocol (VoIP) systems and which businesses they’re most suitable for