The AI market, currently valued at US$20 trillion ($28.4 trillion), draws parallels to the late 1990s dotcom bubble. While technological revolutions drive progress, current speculative momentum concerns veteran investors. This time, a fundamental physical constraint threatens to significantly impede the technology’s projected growth, setting it apart from past market manias.
A primary concern is the United States’ electrical grid, largely unchanged since the 1970s and ill-equipped for the escalating demands of hyperscalers, chipmakers, and the broader AI complex. Experts highlight global bottlenecks in critical components like transformers and transmission lines, compounded by a shortage of skilled workers. Furthermore, massive water consumption for cooling data centres – with some facilities using 10 million gallons daily – raises environmental alarms, particularly concerning vital aquifers.
Beyond this infrastructure crunch, other factors could challenge the AI market. Persistent US inflation might force the Federal Reserve to tighten monetary policy, while cheaper, commoditised AI solutions from rivals like China could disrupt current pricing. The global supply chain for essential equipment, such as heavy-duty gas turbines, faces seven-year waiting lists. A triopoly dominating 90 per cent of this market reportedly slow-walks expansion while enjoying “premium pricing,” driving up costs significantly.
While the AI revolution’s potential is undeniable, history offers cautionary tales. The internet revolution was real before the Nasdaq index experienced a substantial downturn. Without addressing these fundamental infrastructure and resource limitations, the current speculative fervour risks an unsustainable peak. Investors should track the underlying ‘picks and shovels’ – the essential infrastructure – to gauge the market’s true direction, rather than focusing solely on chip orders.
