Whether AI-related stocks are in a bubble has been a subject of intense debate in financial markets over the past six months. Personally, I firmly believe that AI stocks are exhibiting significant signs of a speculative bubble, for the following reasons:
With the exception of manufacturers involved in the production and sale of AI chips and memory, the vast majority of AI-related companies, including SpaceX and Alphabet, which recently released their financial results, have reported negative free cash flow. This suggests that the enormous investment required to advance AI development is putting increasing pressure on the financial positions of many companies in the sector.
Concerns over the potentially uncontrollable development of AI are becoming increasingly difficult to ignore. This is not simply reflected in Jacob Coxon’s warning in September that AI could potentially lead to human extinction before 2030. More recently, numerous hackers have reportedly exploited AI to launch cyberattacks, affecting institutions and their customers across the globe.
These incidents highlight a fundamental risk: Even without reaching artificial general intelligence or AGI, AI can already become a powerful weapon capable of causing serious harm when placed in the hands of malicious actors. This raises an important question: Can the industry really sustain its current pace of rapid development without encountering increasingly significant risks and constraints?
Another critical question is whether financial markets can continue providing unlimited capital to AI-related companies. Over the past two years, AI companies have absorbed trillions of US dollars in funding from global financial markets, becoming one of the contributing factors behind rising global bond yields.
The key issue is whether sufficient capital will remain available to support the next phase of AI investment. Given the sheer scale of funding already committed, the assumption that financial markets can continue financing AI development at the current pace deserves serious scrutiny.
Over the past one to two years, AI companies worldwide have been investing heavily in the construction of AI data centers. But whether the expansion of global electricity infrastructure can keep pace with this rapid development appears to be a factor that many financial market participants have overlooked.
In August, Elon Musk and others highlighted the potential for a significant mismatch between computing capacity and available power supply. It was suggested that approximately 15 gigawatts of AI computing capacity in equipment manufactured in 2027 could remain unusable that year due to power constraints.
This raises a critical concern: Even if AI companies have sufficient funding to continue expanding, their newly constructed data centers may not have access to adequate electricity to operate. If this becomes a reality, AI companies will have little choice but to confront these infrastructure constraints and slow down their investment plans.
Such a development would inevitably force investors to reassess the valuations of AI-related stocks, particularly if the expected returns from massive capital expenditure are delayed or fail to materialize.
Taking these four factors together, I remain firmly convinced that AI-related stocks are exhibiting the characteristics of a significant speculative bubble. The sharp declines across AI-related stocks last week could well be an indication that this bubble has already begun to burst.
Andrew Wong is a veteran independent commentator
