AI, the Bubble That Could Break the US Stock Market’s Back?
AI has completely grabbed our attention, having dominated financial news and market analyses for some time already. Ever since OpenAI launched its first version of ChatGPT, the new technology has taken the world by storm, spawning numerous competitors and countless different applications. From chip producers to data centers and the energy sector, all have become core parts of the outlook and the debate over the future and impact of the nascent sector.
But it hasn’t stopped there. It didn’t take very long for investors to flood the space and for governments to officially enter the AI race. Unprecedented amounts of capital have been injected into AI and AI-adjacent companies and the stock market soon came to resemble a Tech ETF rather than a reflection of the broad economy and the corporate sector at large. Now, however, there is growing concern that the same forces that have been powering the US market to ever higher highs could be the very ones that currently pose the greatest systemic risk.
This article was taken from BFI Bullion’s recent newsletter, the Digger, published on December 18, 2025. To read the latest Digger in its entirety, click here.
A bull market on thin ice
As we recently pointed out in our article “the Everything Rally”, US equities aren’t powering along because “everything” is showing solid, sustainable growth across the board. Instead, the record-breaking gains of the last couple of years are being carried almost exclusively by a narrow cluster of mega-cap tech companies riding the AI wave. Their dominance has now reached levels without historical precedent, with the top 10 US stocks accounting for over 42% of the entire S&P 500. This absurd level of market concentration is masking profoundly worrying weakness elsewhere. Indeed, small-cap stocks are underperforming the Nasdaq by the widest margin ever recorded.
Furthermore, vast sections of the economy are stagnating or even deteriorating, with job openings in decline and multiple waves of layoffs over the last year. Of course, this might come as a surprise to those only tracking indices and following mainstream financial news, but to the average working man, who can’t watch ticker symbols on a screen all day but is instead watching his paycheck shrink, it’s last year’s news.
The cracks in the real economy are becoming increasingly obvious and the seemingly never ending, yet illusory, bull market is a very dangerous distraction for what is really going on under the hood. Despite official data, consumer prices are still exerting enormous pressure on countless working households. This is extremely important to note, because, in the case of a significant market correction, central bank interventions could become even more dangerous. In other words, if the global investor enthusiasm over AI subsides - if there are any doubts over the grand promises that the sector has made - and that change in sentiment takes the entire market with it, then the usual “remedies” of money printing and rate cutting will be much worse than the “disease”. It would usher in another inflationary wave before the last one was even dealt with.
The productivity promise
Most AI founders and investors have made extraordinary claims about the potential of the new technology. Elon Musk recently prophesied that in 10 to 20 years, work will be optional and money irrelevant thanks to AI and robots. Sam Altman of OpenAI claimed that in “some small single digit number of years, not very far,” entire companies will be run by AI CEOs. Regardless of how realistic these forecasts of a “Brave New World” might be, most AI enthusiasts are not investing in these companies merely due to these promises. Rather, they are buying into a much more practical, and seemingly within reach, pledge: that by adopting and embracing AI, companies can expect an explosion in productivity gains, efficiency and cost-cutting through replacing imperfect and fallible humans with machines. And the promised timeline for that was not in “10 to 20 years”, it was now.
What we’re seeing instead, however, is underwhelming at best, with many companies describing AI’s “contributions” as “workslop”. The suddenly popular term first appeared in a 2025 Harvard Business Review report by BetterUp Labs in partnership with Stanford Social Media Lab and is defined as AI-generated work content that masquerades as good work but lacks the substance to meaningfully advance a given task. Workslop can affect anything from emails, reports, presentations, or code and it goes beyond producing a merely subpar or imperfect work product. It consists of vast amounts of low-quality, obviously AI generated output that can increase friction and decrease efficiency. Opposite to its purpose, it can require more human oversight than a flesh-and-blood employee, not less.
The aforementioned Harvard Business Review survey found that workslop could be one explanation for the 95% of organizations that have tried AI but reported seeing zero return on that investment. The ongoing survey included 1,150 full-time, US-based employees, with 40% of them reporting they had received workslop in the past month. As the researchers highlighted: “The insidious effect of workslop is that it shifts the burden of the work downstream, requiring the receiver to interpret, correct, or redo the work.”
What is very important to note, however, is that we are still in the very early days of the technology and none of this means that AI is totally useless or that future breakthroughs will not eventually deliver on these promises. It is also possible that the workslop phenomenon is not entirely AI’s fault. Many companies publicly embraced the technology just to show investors and the public that they will not be left behind and that they are keeping pace with the latest tech developments. AI tools and applications were adopted without really thinking through what the aim was or how it would effectively factor into their businesses and their operational models. Reminiscent of the first crypto craze era, where companies quickly added the word “blockchain” to their marketing materials even when it really didn’t make any sense, the AI craze of today has also seen similarly quick-to-react, misguided cases.
Cracks inside the sector
The most worrying alarm bells are coming directly from the companies that are leading the AI charge themselves. In early November, a letter from OpenAI’s chief global affairs officer, Chris Lehane, and addressed to the White House’s director of science and technology policy, Michael Kratsios, made the rounds. In it, Lehane argued that the government should consider expanding the Advanced Manufacturing Investment Credit (AMIC), i.e. the 35% tax credit from the Biden administration’s Chips Act, beyond semiconductor fabrication to cover electrical grid components, AI servers, and AI data centers. OpenAI’s CFO also said the company was hoping the federal government would help guarantee the financing of chips behind its data center investments, though she walked back her comments after they sparked widespread and fierce criticism that OpenAI was seeking a government bailout.
Also in November, Jensen Huang, Nvidia’s CEO, issued a thinly veiled ultimatum to US policymakers, warning them that unless Washington provides preferential treatment to AI firms, America could lose the “AI race” to China. He argued that the conditions under which China is investing in the sector, especially through energy subsidies, state subsidies, scale, and policy favorability, are putting China “nanoseconds” behind the US in AI development. Put simply, in his view, if the US wants to stay in the race, the government will have to make exceptions for, and concessions to, the industry.
When a nascent sector that has already attracted unprecedented private investment flows is looking for state support and special treatment, it certainly doesn’t inspire much confidence over its outlook. Nevertheless, the most concerning trend in the industry lies in the “fine print” of some of their earnings reports. Prominent investor Michael Burry recently came out as extremely bearish on the AI sector accusing major AI “hyperscalers” (including Nvidia’s customers and partners) of “suspicious revenue recognition.” In particular, he argued that firms are mis-estimating depreciation of AI-related hardware (GPUs, servers), thereby extending useful life schedules in ways that artificially boost reported earnings.
Finally, there are serious concerns over the “circular financing” prevalent in the sector. This refers to arrangements in which leading companies in the AI supply chain simultaneously play overlapping roles in a closed loop. A simplified version of what has been going for months in the AI space goes as follows: a chipmaker invests capital in an AI startup, which the startup uses to build out infrastructure and to purchase chips from the same chipmaker. That purchase becomes revenue for the chipmaker, as it is clearly funding its own future sales via its investment. To muddy the waters further, the startup may itself be part-owner or customer of other firms in the same network, which again invest funds, buy hardware/services, and keep the loop of recycled capital going. This obviously inflates revenues and valuations without necessarily reflecting any real external demand, leading the sector to look more like a pyramid scheme and less like a genuinely booming industry.
Was kommt als Nächstes?
There is no doubt that over the long term, AI will undoubtedly prove to be transformative in ways we haven’t even imagined yet. It will bring about a major paradigm shift in the way we work, the way we research, develop and produce, and perhaps in the way we think about the economy at large too. However, the current market dynamic is not about technological progress or about real-world advantages and contributions. Right now, it is about concentration, promises and projections, and very likely, about investor exuberance and expectations that have outpaced economic reality.
This is a clear and present danger for all investors and even for ordinary people that have no direct exposure to the market. If the AI trade loses momentum, the S&P 500 will not simply cool off; it will enter a severe correction, given the concentration levels in just a handful of AI-dependent companies. And, as we mentioned before, central bankers will be forced to jump in and provide artificial support once again. Only this time, the impact on the broader economy will be grave, as it still continues to absorb the easing of the covid crisis.
The resulting inflationary wave would likely be catastrophic, especially for the working households that are already struggling to make their paychecks stretch until the end of the month.
Read the entire BFI Bullion Digger here.
