BofA sees higher gold prices, likely to hit $5,000/oz in 2026
Investing.com -- Recently, U.S. stocks experienced their sharpest volatility in months, exposing strains in the artificial intelligence trade and prompting renewed debate over whether a speculative bubble is forming.
The pullback followed a strong earnings report from Nvidia that failed to lift the broader market, raising questions about the durability of this year’s AI-driven gains.
However, Evercore analysts argue that today’s setup differs meaningfully from the conditions leading into the 2000 dot-com bust.
They highlight that cloud capital expenditure (CapEx) has surged substantially, quadrupling from $150 billion in 2023 to a projected $600 billion in 2026, a pace that has stirred comparisons with the telecom buildup of the late 1990s.
But the broker highlights four distinctions that, in its view, separate the current cycle from the earlier bubble.
Stay ahead of AI-driven market swings and see how analysts are positioning by upgrading to InvestingPro - get 55% off today
1) The first is financial strength. “90% of Cloud CapEx is from well-capitalized companies like Amazon, Microsoft, Google, Meta, Apple, and sovereigns,” Evercore analysts write, noting that these tech giants are expected to account for more than 90% of spending by 2026.
2) The second difference is demand. Accelerating inference workloads are now driving investment, a contrast to the “dark fiber” overbuild of 1999. Evercore says cloud bookings are rising in line with CapEx and its checks show inference demand is “about to cross over training as the largest AI compute workload at hyperscalers.”
3) Sentiment is another distinction. While today’s market is asking “Are we in an AI Bubble?”, the analysts recall that investor mood in 1999 was far more euphoric.
"In 1999, we recall 9 out of 10 investors being euphoric about the Internet investment cycle," they wrote.
4) Lastly, valuations also look different compared to the 2000 dot-com bust. Leading AI names trade at next-twelve-month (NTM) earnings multiples in the 20s to 40s, far from the 20–40x sales once in communications-chip stocks then.
Nvidia, for instance, trades at 22x its 2026 earnings estimate, near levels that marked past cycle bottoms.