Highlights
- OpenAI spending $1.4trn over three years but generating only $20bn profit in 2025.
- AI chips and data centres may become obsolete within 2-3 years, threatening $1.6trn in tech valuations.
- Only 12 per cent of US companies adopting AI for production despite massive industry hype.
The artificial intelligence investment boom that has dominated global markets faces mounting questions over sustainability, as tech giants commit trillions to infrastructure while generating minimal returns, raising fears of a potentially catastrophic economic bubble.
Major technology companies including Microsoft, Amazon, Google, Meta and Oracle are expected to spend approximately $1trn on AI by 2026, with OpenAI alone committing $1.4trn over the coming three years. However, OpenAI's expected 2025 profit of little more than $20bn represents a fraction of its spending commitments, highlighting a troubling gap between investment and returns.
The "magnificent seven" tech companies Nvidia, Microsoft, Amazon, Google, Meta, Apple and Tesla now account for 37 per cent of the S&P 500's performance, with 75 per cent of the index's returns coming from just 41 AI stocks.
The scale of infrastructure being built is staggering. Stargate, announced in January by Donald Trump and OpenAI's Sam Altman, already covers an area the size of Manhattan's Central Park in central Texas.
Meta's $27bn Hyperion data centre in Louisiana approaches the size of Manhattan itself and will consume twice the power of nearby New Orleans. The rampant power demand is straining America's grid, with some data centres waiting years for connections.
This concentration of market value in artificial intelligence has intensified concerns about systemic risk.
Gary Marcus, AI scientist and emeritus professor at New York University, warned of serious consequences "In the worst case, what happens is the whole economy falls apart, basically. Banks aren't liquid, we have bailouts, and taxpayers have to pay for it."
Returns under pressure
The sustainability concerns centre on three critical factors. Firstly, AI infrastructure depreciates far more rapidly than traditional assets.
Fund manager Michael Burry, who predicted America's sub-prime crash, recently bet against AI stocks, reasoning that chips require replacing every three years. The Economist estimates this could reduce big tech's combined value by $780bn to $1.6trn.
Secondly, business adoption remains limited. According to the US Census Bureau, only 12 per cent of companies currently use AI to produce goods and services, with larger corporations showing declining adoption in recent months.
Finally, even AI pioneers doubt the current approach. Ilya Sutskever, OpenAI co-founder who helped pioneer ChatGPT, told the Dwarkesh Podcast last month. "Is the belief that if you just 100x the scale, everything would be transformed? I don't think that's true."
Jensen Huang, CEO of Nvidia, dismissed bubble concerns, told Sky News that "We are long, long away from that." However, analysts estimate big tech needs $2trn in profit by 2030 to justify AI costs, a target that appears increasingly distant as adoption stalls and technological limitations emerge.
