As the World Cup in North America draws to a close, let’s talk about the United States. The competition was an undoubted success: huge stadiums, enormous crowds, lots of tourists, buckets of revenue and great football. Despite this success, the World Cup underscored yet again how much of an outlier the US remains. Once their own side was beaten, the American public largely moved on. Sure, as it is hosted by their cities, there’s a residual interest, but nothing like the devotion elsewhere. Soccer is beloved by the rest of the globe and, although popular among America’s hipsters, it is a sporting sideshow in middle America. Locals are more invested in their “own” sports: baseball, American football and basketball. American exceptionalism is alive and well in sport.
Another place where American exceptionalism is alive and well is in finance. The Yanks obsess about Wall Street the way Europeans fixate on football. For most Europeans, the stock market is about as relevant as baseball. For Americans, it seems to be an essential part of their daily concerns. Reams of ticker-tape news fill all sorts of business programmes, 24-hour finance updates are the norm. CEOs of companies are interviewed like Premier League managers and an entire media industry relies on the daily gyrations of the stock market. Why? The reason is simple: about 50 per cent of working Americans actively invest in the stock markets through their “401k” self-managed pensions. That is more than 80 million people.
In the past few years, investment by the average American in Wall Street has increased significantly. This week, a senior Barclays executive revealed for the first time that more of American household wealth is now held in stocks rather than the traditional asset, real estate. Americans’ equity portfolios are now worth more than their homes. That means 34 per cent of American wealth is now held in equities and a lower 26 per cent in property. The major reason for this is that the equity markets have surged in recent years, largely on the back of the AI and tech story focused on a narrow half a dozen or so companies. Stock portfolios have increased in value in tandem. Is such concentration of wealth in only a few companies, and only in one sector, safe or is the AI story vulnerable to a shock and, if so, what would happen to the wealth of middle America? Is the American AI bonanza a fundamentally secure investment in the future or an old-fashioned bubble, driven by human psychology and Fomo not just on the part of small investors, but of the large so-called superscalers, those companies spending billions?
Let’s look at the numbers.
Total investment in US AI was $256 billion in 2025, up from $95 billion in 2024. It is expected that AI spending will be between $650 and $800 billion this year, while this could go above $1 trillion in 2027 and $3 trillion by 2035. We are talking about 2 per cent of total US GDP already. A few American companies are now spending on AI just shy of what America spends in total on higher education, and soon US private companies will be spending more on AI than the US government does on defence. This boom is already far greater than the ill-fated dot.com bubble, and AI spending far outstrips the 0.4 per cent of GDP spent on both the 1960s and early 1970s on space exploration or the interstate highway system of the 1950s.
When private companies are borrowing such huge sums from investors, they’d better start making money to pay back the loans. For this to happen, companies using and buying AI products from the AI providers must show that it is making a difference to the bottom line, otherwise why use it? Thus far, the results are underwhelming. Last year MIT (Massachusetts Institute of Technology) found that 95 per cent of companies in the US have experienced no return in investment from using AI, while 99 per cent of companies reported financial losses on AI spending. The key metric in corporate speak (that yucky language deployed by management consultants) is the “activation gap”. This “gap” is the difference between job disruptions/losses due to AI and productivity gains from the technology. If this gap isn’t closed, companies could find they are laying off productive workers for less productive AI tools, which could destroy profitability, sparking panic in financial markets.
Assuming that AI will be revolutionary at some stage, the critical short-term issue is the debt mountain that is building and the interrelated nature of who is lending to whom. Who will hold the can if this thing comes crashing down? What happens if the superscalers – the companies leading the AI charge such as Nvidia, Oracle, Microsoft and Open AI – can’t pay the money back? What happens to the shares of such companies if the mood of the market swings from techno-optimism to commercial-pessimism? What could trigger such a reversal?
Let’s focus on one source of obvious concern: China.
[Why the American dollar was more important than US constitutionOpens in new window ]
The US has privatised the AI surge: private companies, privately financed, have employed the best brains in the tech world on enormous salaries to come up with the best AI products. They hope to charge top dollar for these AI tools. With high prices for the products, it is hoped that ensuing revenues from AI will pay back all the investors who have bet billions on the AI future. That makes sense in what could be described as a typically “closed end” market. The people who gamble most win.
Imagine this model is disrupted by an “open source” model where the development of AI is not limited to very well-paid engineers trying to come up with the most commercially profitable use of AI? What might happen if a big country such as China pushed “open source” AI development where millions of engineers driven by a greater goal than profitability were to be deployed to develop AI? Imagine a sort of “Wikipedia for AI” development. Well you don’t have to imagine it, it is already happening. China is “open-sourcing” its domestic AI.
Despite all the investment in US tech companies, China still has six of the top 10 most popular AI models worldwide. DeepSeek, China’s major AI company, released a new flagship model in late April called V4, which is as powerful as US models released about half a year ago, and a fraction of the cost. The US is focusing on the technologically best – the model with the most “wow” factor – while China is focusing on the best use at the best price. Could China – as it has done with cars, drones and mobile phones – simply undercut the Americans, providing better AI models more cheaply to the world?
If the American companies lose their global pricing power, who pays back all the borrowed money? And what happens to the wealth of those 80 million Americans, tied up in the stock market, itself increasingly a giant bet on expensive AI? We are entering dangerous territory, where tech and profits meet geopolitics and strategy. Who wins is the issue of our times.
Now back to the World Cup for one last hurrah.
[A conflict between the US and China now looks inevitableOpens in new window ]