Most of US growth now depends on AI – and economists suspect a bubble

What happens when artificial intelligence becomes both the main driver of the economy and the economy’s potential Achilles heel? Maybe we’ll find out soon.

Harvard economist Jason Furman recently said that AI investments would account for nearly 92% of US GDP growth in the first half of 2025. In short, the entire US economy put its eggs in one algorithmic basket.

And that basket may be about to break. Today the Bank of England warned that market valuations for AI companies are becoming increasingly irrational.

“Based on a number of measures, equity market valuations appear stretched, especially for technology companies focused on artificial intelligence (AI),” the bank’s analysts wrote. “Combined with increasing concentration within market indices, this makes equity markets particularly vulnerable as expectations around the impact of AI become less optimistic.”

If you need more proof of the market’s exuberance, Elon Musk’s xAI is currently raising a whopping $20 billion, apparently earmarked for what amounts to the world’s most expensive Nvidia purchases.

The company, which is valued at $200 billion (up from $6 billion a few months ago), plans to use the money for its ‘Colossus 2’ data center. In a delightful bit of circular logic, Nvidia itself is reportedly investing up to $2 billion in the deal, essentially paying for priority access to its own chips.

Ruchir Sharma, chairman of Rockefeller International, said Fortune that “America has, in a sense, become one big bet on AI,” and warned that “AI better help the US, otherwise the economy and markets will lose the one leg they currently stand on.”

The Bulls’ defense: “This time it’s real”

So what is it: a transformative revolution or the second coming of the dot-com crash? There are arguments for all of these positions, and renowned doomers and accelerationists debate them every day.

The bulls have convincing ammunition. Unlike the vaporware companies of 1999, today’s AI giants are printing money. The ‘Magnificent Seven’ technology companies appear to be hugely profitable. Nvidia shares are up 1,700% in the past two years. OpenAI aims for a turnover of $12.7 billion by 2025. Microsoft, Google, Meta, AMD, Oracle and all the major companies involved in AI are also exceeding expectations.

The infrastructure being built is tangible too: data centers bustling with activity, power generation facilities (including those quirky nuclear deals) and software that is already transforming the way businesses work. Nearly 90% of developers use AI today, with generative AI adoption more than doubling within a year.

UBS Chief Investment Officer wrote in an analysis that this is just big but healthy momentum. “There is little evidence of a market bubble at the moment, and we would like to take advantage of the AI-driven momentum in the stock market with a broadly diversified portfolio,” he said.

Financial analyst Steven Fiorillo also argues that despite what doomers preach, this bullish market is backed by results and not hype. “I have a newsflash for every bear, AI is not a bubble,” he posted on X last week.

“In the end, MSFT, AMZN, GOOGL and META generated $493.31 billion in cash from their operations, allocated $291.35 billion in CapEx and generated $201.96 billion in FCF (free cash flow) in the TTM (twelve months later). These numbers indicate that the dotcom era and the AI ​​era are very different and that there is no AI bubble can be found.”

This view is also shared by analysts at Bank of America, who argue that volatility is a signal that markets are healthy and that there are insufficient signs of a market bubble.

The doomers’ rebuttal: “History rhymes”

But the bears have history on their side, and history has a nasty habit of rhyming. The Bank of England’s comparison with dotcom valuations is based on hard numbers.

The top ten S&P 500 companies now control more than a third of the index’s total valuation, a concentration not seen in half a century. When so much wealth is tied to so few companies, a stumbling block becomes a systemic risk.

However, according to AI researchers at MIT, 95% of organizations fail in their investments in generative AI. Are we witnessing real transformation or collective delusions? The practical bottlenecks – power shortages, limitations in chip supply, the sheer physics of cooling all those servers – suggest that even if AI is revolutionary, the revolution could be slower than stock prices imply.

Tech CEOs themselves are increasingly vocal about bubble concerns, even as they continue to raise astronomical sums.

“Are we at a stage where investors as a whole are too excited about AI? My view is yes,” Sam Altman, CEO of OpenAI, said in August. “When bubbles form, smart people get overexcited about some truth.”

Jeff Bezos of Amazon is also convinced: “This is a kind of industrial bubble,” he said on Friday during the Italian Tech Week in Turin. “There will be a reset, there will be an audit at some point, there will be a recording.”

The truth probably lies in that murky middle ground. AI is undoubtedly transformative: any technology that can drive 92% of economic growth is not just hype. But the question, as always with the next big thing, is how much of the hype is indeed there.