Internet Bubble "Survivors": Today's AI is more like 1997 than 1999

What point in the history of the internet era does the current AI infrastructure boom correspond to— is it the accelerated start in 1995, the mid-stage sprint in 1997, or has it already edged toward the end of the 1999 bubble?

Wall Street is making radically different judgments about the market stage based on their own experiences.

Dan Niles, founder of Niles Investment Management, believes that the current stage of AI development is closer to 1997—the third or fourth year of building internet infrastructure—rather than 1999, the year before the internet bubble burst.

The assessment from legendary hedge fund figure Tudor Jones is also relatively optimistic. He previously said on media programs that the current stage of AI development is highly similar to the accelerated commercialization period of the internet in 1995, and he estimates that this AI bull run has already covered about 50% to 60% of its course—“it can still last another one or two years.” Jones compares the current market’s feel to 1999, when there was still about a year left before the dot-com bubble stocks topped out in early 2000.

On the pessimistic side, Burry—the prototype of The Big Short—believes that the Nasdaq 100’s actual price-to-earnings ratio is as high as 43 times, while the Philadelphia Semiconductor Index surged nearly 70% over three months. The current trend, in his view, closely matches the period just before the internet bubble burst in 2000.

Short-term overvaluation, long-term undervaluation?

During the internet bubble, Dan Niles was a highly regarded chip-stock and PC hardware analyst, so his judgment carries a certain amount of historical reference value.

In an interview with the Master Investor Podcast, he said, “We are now in the third or fourth year of building internet infrastructure,” which aligns more with 1997 than with 1999, the period just before the bubble peak.

Niles points out that an important catalyst for this round of the AI cycle is the rise of agentic AI, led by Clawd Bot (now renamed OpenClaw). This has driven a rapid expansion in demand for compute power and has caused the value of CPU chips from companies like Intel to draw fresh attention. He said, from a short-term perspective, chip stocks’ valuations are somewhat on the high side, but from a long-term perspective, they are not obviously overvalued—“Intel just returned to its 2000 level last year, and relative to its potential profitability, it is still undervalued.”

However, Niles is cautious about the overall market direction.

He notes that among the backlog orders piled up by hyperscale cloud computing providers, half come from OpenAI and Anthropic, and OpenAI currently does not have the cash flow to support its ambitions.

He also questions the current pattern in which multiple assets are strengthening at the same time: “The stock market is at a historic high, oil prices are up 60% this year, and the yields on the 30-year and 10-year U.S. Treasuries have both hit their highs for the year—these signals cannot all be correct at the same time.” He also states plainly, “Now you should hold a lot of cash.”

Risk warning and disclaimer

        The market is risky; invest with caution. This article does not constitute personal investment advice, nor does it take into account any specific investment goals, financial situations, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article are consistent with their specific circumstances. Invest accordingly at your own risk.