The business press is increasingly concerned about the AI stock market bubble. Concerns are being highlighted about over-concentration in AI-related giants, noting that the hype around AI may be inflating valuations while delivering lacklustre returns (Tech Stocks are sending a warning, Financial Times, 23 August). A recently published Massachusetts Institute of Technology study shows that 95% of AI investments yield no measurable returns, underscoring this caution. Many AI startups remain unprofitable, with valuations remaining sky-high. Some experts fear that if the bubble bursts, it could produce losses dwarfing those of the dot-com crash of the early 2000s (Tech guru says investors will ‘suffer’ far more from the AI boom than the dot-com crash, Business Insider, 15 August). Recent slides in AI-linked tech stocks now appear to flow from valuation anxieties and macroeconomic jitters, such as relatively elevated interest rates, which remain high against fears around stubborn inflation.
There are also shades of 2008 via the existence of speculative ‘special purpose vehicles’ acting in the markets. These ‘SPVs’ essentially pool money from investors to buy shares in a startup, and some apparently have snake oil-like structures. They are little more than clever marketing vehicles to extract fees from excited investors who have a ‘fear of missing out’ on the tech bubble. The fees charged by SPVs are often high, their structures opaque, and many retail investors may not fully understand the risks (AI Fomo could be fueling a risky bubble in AI’s hottest companies, Business Insider, 22 August).
History repeats itself as farce.
This bubble-like behaviour is an inevitable feature of market economies and it is not just irrational exuberance as a Keynesian might claim. Individual investors, funds, and corporations feel compelled to join the rush because if they do not, their rivals will. It is the coercive law of capitalism at work. It also has another structural aspect insofar as it represents capitalism’s way of recycling surplus money into risky bets when safe, profitable outlets are scarce.
There is also a considerable dose of technological fetishism going on here with investors treating AI as if it has almost magical powers to generate value and super-normal profit, detached from real-world productivity. And like all bubbles that preceded it, there will be winners and losers. You can bet the current beneficiaries are the venture capitalists, tech executives, asset managers, and those who buy and sell at the right time. And while some of the latter will be among the losers when the bubble bursts, so too will inevitably be the ordinary retail investors and workers who came to the party a little too late to get out safely, whether that be through direct investments or indirectly via pensions and index funds.
That said, an AI bubble burst might not be a 2008-style collapse. If AI equities drop hard, high-income households (who own most stocks) pull back a bit on spending; that will have some effect but is not likely to induce the collapse of capitalism. If the bubble burst feeds into a cut or delay in data centre builds and microchip orders, the impact becomes more significant. The current spend is substantial (JPMorgan and MUFG near $22bn data centre financing deal in Texas, Financial Times, 20 August), so a pause would ripple through construction, industrials, and the banks that financed the projects—a case of watch this space.
