The Bank for International Settlements (BIS) published its Annual Economic Report on June 28, 2026, and buried inside its usual warnings about debt and inflation was something that caught the AI industry’s attention: a direct comparison between the current AI investment boom and previous boom-and-bust cycles, from canal building to railways to the dotcom era.

The BIS, essentially the central bank for central banks, identified four pressure points threatening global economic stability. One of them was AI. Specifically, the BIS flagged uncertainty over “the durability of AI-related investment” as one of the most alarming risks to global prosperity right now.

This is worth paying attention to. The BIS has called things like this before, and it tends to be measured in its language. When it reaches for comparisons to canal bubbles and dotcom crashes, that is not noise.

What the BIS Actually Said

The report outlined a specific concern: AI investment has grown so large, in both nominal terms and as a share of GDP, that leading technology firms can no longer fund it from operating cash flows alone. Financing is shifting toward debt, with private credit and hedge funds playing an increasing role. The report described this as “complex funding structures across the supply chain,” which is financial-speak for a system that has become fragile.

To put a number on it, hyperscalers including Amazon, Alphabet, Microsoft, Meta, and Oracle issued more than $100 billion in corporate bonds in 2025 to fund their AI infrastructure buildout. That is a lot of debt for an investment whose return timelines are still being debated.

The BIS noted that supply bottlenecks and intense competition could lead to the kind of overinvestment seen in previous technology cycles. The concern is not that AI is useless. It is that so much money is chasing the same infrastructure bets that the market eventually reprices sharply when expected returns do not materialise fast enough.

This Is Not About Your Business

Here is the part that matters for business owners: the BIS warning is almost entirely about speculative infrastructure investment at the macro level. It is about hyperscalers spending $100 billion-plus on data centres, compute, and chips. It is about hedge funds and private credit financing those bets. It is about what happens to financial markets if those bets disappoint.

It is not about a professional services firm in Brisbane using AI to automate its client reporting. It is not about a medical practice deploying a voice agent to handle after-hours calls. It is not about an accounting firm building a workflow that cuts month-end close time in half.

The businesses at risk in an AI bust are the ones making speculative financial bets on AI stocks, or the ones that invested millions building proprietary AI infrastructure they are still struggling to justify. The businesses that are fine are the ones adopting AI deliberately, with clear use cases, measurable outcomes, and realistic expectations.

The Real Risk for Business Leaders

That said, there is a genuine warning buried in the BIS report that applies to every organisation, and it is a strategic one, not a financial one.

The pattern the BIS described, where companies pour capital into a technology before the returns materialise and then face a painful correction, is not just a stock market story. It plays out inside organisations too. We have seen it with ERP implementations, with digital transformation programmes, and now we are starting to see early signs with AI.

The companies most at risk are the ones:

• Deploying AI without clear success metrics
• Assuming AI will pay for itself without changing the underlying workflows
• Chasing headlines about what competitors are doing rather than what their own customers actually need
• Treating AI as a cost centre without building the internal capability to get value from it

The BIS would not use these words, but the risk it is describing at the macro level is really just speculative deployment at scale. The same pattern shows up at the business level when organisations adopt AI out of fear of missing out rather than genuine strategic intent.

What Business Owners Should Actually Do

The BIS warning is not a reason to slow down AI adoption. It is a reason to adopt AI with more discipline.

A few things that matter right now:

Get clear on where AI is actually producing value. If you have deployed AI tools and you cannot point to specific, measurable outcomes (time saved, revenue protected, costs reduced), that is a problem worth fixing before you go further. The same financial scrutiny the BIS is applying to hyperscalers applies to your AI budget.

Avoid the infrastructure trap. Most businesses do not need to build their own AI infrastructure. The opportunity is in using existing AI tools and services, such as AI agents, automation platforms, and voice AI, on top of infrastructure that someone else has already built and is paying to maintain. You get the benefit without the exposure.

Invest in the capability, not just the tool. The organisations that will weather any AI market correction are the ones with genuine internal competency, meaning teams who understand how to use AI, how to evaluate it, and how to adapt when the tools evolve. That competency comes from training and practice, not from buying software licences.

Look for near-term payoffs. When capital markets get nervous about AI timelines, the projects that survive scrutiny are the ones with near-term, demonstrable ROI. If your AI investment takes five years to pay off, it is going to be harder to defend in a tighter environment. Start with the workflows where automation pays back within months.

The Bottom Line

The BIS is right to flag AI investment risk at the macro level. There is a genuine possibility that the infrastructure buildout runs ahead of demand and that financial markets reprice sharply at some point. Business owners do not need to predict when that happens or whether it will happen.

What they do need to do is make sure their own AI investments are grounded in real outcomes, not in speculation or fear of missing out, and not in the assumption that buying the right tools is the same thing as building the right capability.

That distinction is exactly what separates the organisations that will come out stronger from AI adoption from the ones that will have to explain to their boards why the programme did not deliver.

---

The practical next step is the free Working With Claude field guide. Thirty-two pages covering the ecosystem, Claude Code, and how to govern a rollout properly. Get your copy.