September 11, 2026

Who Is Really Paying for the AI Boom?

by Ewen Emmerson

in Insights Research

We have written before about how concentrated an ordinary, diversified portfolio has quietly become in a handful of very large technology companies, and how that concentration is often bigger than investors realise. This month, we want to look at a related but different question: not how much money is being spent on artificial intelligence, but how that spending is actually being funded, because the answer has changed, and it matters.

For the first couple of years of this investment cycle, the story was simple. A small number of very large, very profitable companies were spending their own cash on new computing infrastructure, funded largely from their own enormous profits. That is a comfortable story for investors to hear. A company spending money it has already earned, on a business it understands, is doing something well within its own means.

That story has moved on. The sums involved have grown so large that some of the biggest spenders are no longer funding it purely from their own pockets. They are increasingly turning to debt, both their own and, in some cases, debt taken on by joint ventures set up specifically to build and lease out the infrastructure. And a newer, more unusual pattern has emerged alongside this: some of the companies that make the computing equipment are themselves investing in, lending to, or guaranteeing payments for the very customers who buy that equipment.

What this means in practice, is that in parts of this industry, the supplier, the customer, the lender and the investor are increasingly the same small circle of companies, dealing with one another. A chip maker invests in an AI company. That AI company uses the money to buy chips, and to lease computing capacity from a data centre operator. The chip maker, in turn, guarantees some of the payments that make that lease possible. Money moves around the circle, and at each step, it looks like more revenue and more investment. It can be entirely genuine activity. But it also means that if demand for any part of this chain were ever to disappoint, the effect would not stay neatly contained to one company. It would likely be felt by several of them at once, for the same underlying reason.

This is not a hypothetical concern. In the past fortnight, one of the world’s largest providers of computing infrastructure reported results that, on the surface, looked strong, with revenue growing very quickly. But the market’s attention went straight to two other numbers: the company’s cash flow, which turned negative as spending on new infrastructure accelerated, and its own guidance for future capital spending, which was raised again, sharply. The shares fell by more than five per cent on the day, and the cost of insuring that company’s debt against default rose too. This was one company, in one quarter, and it does not tell us the wider AI investment story is faltering; demand for computing power, on the evidence we see, remains genuinely strong. But it is a clear, public illustration of exactly the dynamic we are describing: rapid growth and rising financial strain can arrive in the same set of results, and when a company’s future is tied up with the fortunes of its own customers and backers, that strain does not necessarily stay isolated.

We want to be careful not to overstate this. The companies at the centre of this activity are, by most conventional measures, financially strong, with far less debt relative to their size than the average large company. The concern is not that any single business is in immediate difficulty. It is a structural one: as more of the industry’s growth becomes financed through debt and through arrangements between a small number of interconnected companies, the industry as a whole becomes more sensitive to any disappointment in demand, and less like the simple, self-funded growth story it was two or three years ago.

For our portfolios, this reinforces rather than changes the approach we have already described. We continue to hold this theme through diversified, actively managed exposure rather than through concentrated positions in a small number of names, precisely because it allows a manager to weigh these financing questions on a company-by-company basis rather than assuming the whole sector carries the same risk. We are not making any change to portfolios as a result of this month’s news, and we would caution against reading too much into a single quarter from a single company. But we think it is right to explain plainly what is happening, so that the strength of the headline growth numbers you read about is not mistaken for the whole picture.

As ever, our approach is to explain what we see and why it matters, not to predict how it resolves. Perspective, not prediction, remains the principle we hold ourselves to.

If you would like to discuss how this theme is reflected in your own portfolio, please contact us and we will be happy to arrange a conversation with one of our advisers.