Tom Lee, co-founder of Fundstrat Global Advisors, suggests the artificial intelligence market is currently in a phase comparable to 1997 rather than the 1999 dot-com peak. Speaking at TOKEN2049 in Singapore, Lee projected that the peak of the current AI cycle could arrive around 2033, drawing parallels to the long-term growth of the internet and mobile networks.
Lee estimates total infrastructure spending on AI at approximately $30 trillion. This figure aligns with the historical cost of building out the internet and mobile sectors, which he values at about $35 trillion in 2026 dollars. He notes that the global market sees roughly $7 trillion in new liquid assets annually, which he believes is sufficient to support this level of investment.
In contrast, Ray Dalio of Bridgewater Associates characterizes the current AI market as a classic bubble driven by debt. Dalio argues that rising interest rates will force investors to liquidate positions, potentially triggering a market correction. Recent market volatility in Asia on October 13, which saw a decline of over $800 billion in equity value, highlights these growing concerns.
The two analysts differ on the nature of AI financing. Lee views current expenditures as early-cycle investments, while Dalio warns that credit-fueled spending is pushing markets toward a collapse. Meanwhile, McKinsey reported in April 2025 that AI data centers alone would require $5.2 trillion in investment by 2030, a figure significantly lower than Lee's broader infrastructure estimate.
