"Internet bubble 'survivors': Today's AI is more like 1997, not 1999."

```

Exactly where does the current AI infrastructure boom stand on the historical timeline of the internet era—is it at the accelerated starting stage of 1995, the mid-stage sprint of 1997, or approaching the bubble’s end in 1999?

Wall Street is offering sharply different judgments about the current market phase, based on their own experiences.

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

Legendary hedge fund manager Paul Tudor Jones also takes a rather optimistic view. Previously, he said on a media program that the current stage of AI development closely resembles the period of rapid commercialization of the internet in 1995, estimating this AI bull market has completed about 50% to 60% of its course and "could last another year or two." Tudor Jones compares today's market feel to 1999—when internet bubble stock prices wouldn't peak until early 2000, still about a year away.

On the pessimistic side, Burry, the inspiration for "The Big Short," notes the actual P/E ratio of the Nasdaq 100 has reached 43, the Philadelphia Semiconductor Index has soared nearly 70% in three months, and the current trajectory closely matches the eve of the internet bubble burst in 2000.

Short-term Overvaluation, Long-term Undervaluation?

Dan Niles was a highly regarded chip stock and PC hardware analyst during the internet bubble period; his judgment carries historical reference value.

In an interview with Master Investor Podcast, he states, "We are now in the third or fourth year of internet infrastructure construction," which aligns more with 1997 than with the bubble’s peak in 1999.

Niles notes that a key catalyst in this round of AI developments is the rise of agentic AI represented by Clawd Bot (now renamed OpenClaw), driving rapid expansion in computing power demand and bringing renewed attention to CPU chips from companies such as Intel. He says, chip stocks are clearly overvalued in the short term, but from a long-term perspective, not obviously overvalued—"Intel just returned to its level of 2000 last year, and, relative to its potential profitability, remains undervalued."

Nevertheless, Niles takes a cautious attitude toward overall market trends.

He points out that half the backlog orders of hyperscale cloud computing companies come from OpenAI and Anthropic, while OpenAI currently does not have the cash flow to support its ambitions.

He also questions the current pattern of multiple asset classes rising together: "The stock market is at historical highs, oil prices are up 60% this year, and yields of both 30-year and 10-year U.S. Treasury bonds hit their yearly peaks—these signals can't all be right at the same time." He states plainly, "Now one should hold a lot of cash."

Risk Warnings and DisclaimerThe market involves risks; investment must be prudent. This article does not constitute personal investment advice, nor does it take individual users’ specific investment goals, financial situation, or needs into account. Users should consider whether any opinions, views, or conclusions in this article fit their particular circumstances. Investment based on this is at one's own risk. ```