Tom Lee Says AI Is in 1997, Not 1999: Peak Bubble Could Be 2033

Tom Lee, co-founder of research firm Fundstrat Global Advisors, says the AI bubble is years away. He pegs AI spending at $30 trillion, similar to the 1990s internet buildout, and sees a peak near 2033. Lee spoke at TOKEN2049 Singapore, as Ray Dalio, founder of hedge fund Bridgewater Associates, calls AI a classic bubble. Lee says near-universal bubble talk is itself a reason to doubt one. Why Does Tom Lee See the AI Bubble Peaking in 2033? Lee argues that many of todays fund managers never managed money through the dot-com bubble or the 2008 financial crisis. Ive never seen a bubble when everyone thinks its a bubble. Tom Lee, co-founder of Fundstrat Global Advisors, at TOKEN2049 Singapore. His anchor is cost. Lee puts the internet, mobile network and handset buildout at about $35 trillion in 2026 dollars. He estimates AI infrastructure spending at roughly $30 trillion. By contrast, McKinsey estimated in April 2025 that AI data centers would require $5.2 trillion in investment by 2030. Lee notes that Cisco did not peak until March 2000 and fell 50% twice before then. The stock later dropped nearly 90% from that peak. By that clock, Lee places AI in 1997. Wireless grew exponentially for 25 years, he says, and its stocks peaked 10 years in, so he lands on 2033. Lee adds that about $7 trillion in new liquid assets arrives each year, capital he says can fund AI spending. Does Debt Change the Math for AI Investors? Dalios AI bubble warning centers on borrowing. As rates climb, he argues, holders of paper profits cash out, and that selling typically pops a bubble. Dalio says markets are approaching that stage. Doubts about AI already rattled Asia on October 13, when an Asian equity selloff erased more than $800 billion. The two views split on financing. Lee reads AI spending as early-cycle investment, while Dalio sees debt pushing markets toward a burst. Both cases rest on the same spending, and on whether cash flow or credit pays for it.