There is no doubt that recent years have seen some exceptional growth figures on stockmarkets worldwide, and more particularly on the S& P 500. Over the 5 years to September 2026, the S&P 500 has delivered returns of 73.8%.
A definite contributory factor to these figures has been what we could be referred to as the AI bounce, whereby many companies by virtue of having increasingly adopted Artificial Intelligence within their business practices have been able to reduce their costs, thereby increasing profits margins and ultimately profits.
Whilst it is questionable whether this is ultimately to the benefit of society in general, it has without question been highly beneficial to stock markets worldwide, and will in all likelihood continue to be so for many years to come, as AI uptake and usage continues to increase.
However, one very interesting question has arisen recently around the whole AI debate, namely is it possible for the frontier AI firms to develop positive returns on the colossal amount of funds that are being invested into them, and ultimately to become profitable. The answer to this is far from clear cut.
Let’s take Open AI (owners of Chat GPT) as an example. In 2025 Open AI lost an eye watering $38.5 billion on revenues of $13 billion, and is not projecting to become profitable until 2030, by which time it will have spent a further $278 billion. The question then becomes whether it will be able to generate enough revenues to cover not only the enormous expenditure on infrastructure build out and computing capability, but also the extremely high ongoing expenditure required on the Large Language Models (LLMs) that are required to run AI models, and also the enormous electricity and cooling costs required to run the AI data centres.
Whether they will be able to do so is highly debatable for two primary reasons: firstly there is a high level of competition within the American AI market, with the likes of Open AI, Anthropic and Xai competing against each other for market share, and early signs indicating that there is no great amount of customer loyalty from customers towards particular AI providers.
Secondly, and arguably more seriously, there is competition from Chinese AI firms. Until recently American AI firms appeared very confident in stating that their Chinese counterparts were a long way behind them in terms of their AI development. However, recent events have shown that to be far from the case, with Kimi K3, a new AI model owned by the Chinese operator Moonshot AI proving to be only marginally inferior to the American models.
And crucially, unlike American AI models, Chinese ones are open source, meaning their coding and architecture is publicly available information, and therefore it is not necessary to pay to access them.
As a result, Chinese AI models are a lot cheaper than American ones. The question then becomes; will AI users worldwide be happy to pay significantly higher prices for American models when they can pay a lot less for Chinese ones that are nearly as good? And also, if there is widespread competition amongst AI providers, will AI itself become largely commoditised and therefore subject to significant price competition?
If the answer to this is yes, then it becomes difficult to see how the enormous current and future spend on building frontier AI models can ever be recouped, and if some of the American providers can ever become profitable (interestingly if you ask Chat GPT if Open AI will ever be profitable then the answer you get is that it is far from certain.
The danger in all of this is that it appears that the US have to a large extent bet the house on AI, confident initially that it was a sure fire winner. But that is looking increasingly less likely. Open AI have now delayed their IPO, originally scheduled for this year, until 2027 at the earliest. Cynics would suggest that this is because their figures are currently nowhere near strong enough to stand up to the public scrutiny that accompanies an IPO.
And if it turns out to be the case that US AI firms start to fail, will that bring down the entire US stock market with it, given the enormous amounts invested? At the moment it is very difficult to say with any certainty, but it is starting to look a distinct possibility that at least some of the AI story could be one of a bubble that is about to burst.
