Has the AI boom turned into an AI bubble?

when Jansen HuangMeet the CEO of chip maker NVIDIA Donald Trump He had reason to be cheerful in the White House last week. In Asia, most NVIDIA chips are widely used to train generated artificial intelligence models. Earlier this year, it pledged to increase production in the U.S. and on Wednesday, Trump announced that chip companies that promised to make products in the U.S. would be exempt from some major new tariffs on semiconductors his administration is preparing to impose. The next day, NVIDIA’s stock reached a new all-time high with a market cap of $4.4 trillion, making it the most valuable company in the world, ahead of Microsoft, which is also largely involved in AI.

Welcome to the AI boom, or should I say the AI bubble? It has been more than a quarter of a century since the bursting of the giant internet bubble, during which hundreds of unprofitable internet startups have issued shares to the Nasdaq, and many tech companies’ shares have risen to the stratosphere. In March and April 2000, tech stocks plummeted; many Internet startups then went bankrupt, but by no means all. Over the past few months, Wall Street has discussed whether the current surge in technology follows a similar trajectory. exist Research papers Titled “25 Years; Lessons from the Burst of the Technology Bubble,” published in March, a team of investment analysts from Goldman Sachs believes that this is not: “While enthusiasm for technology stocks has risen sharply in recent years, this does not represent a bubble because the bubble is proven by strong profit fundamentals.” Analysts point to the revenue capacity of the so-called seven companies: Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla. Between the first quarter of 2022 and the first quarter of this year, NVIDIA’s revenue accounted for Quintuplpl, and its after-tax profit increased by more than tenfold.

Goldman Paper also offers helpful history lessons. It noted that between 1995 and 2000, the Nasdaq High Technology index rose five times, and at the peak of the market, widely used valuation measures (price versus price-earnings ratio), or “p/e” – increased by 150, 150, and could not find a level before or after that. By contrast, the five-year period from March 2020 to March 2025 is relatively tame. Indeed, the Nasdaq roughly doubled, and the P/E ratio increased significantly. But it doesn’t have three numbers nearby.

Written widely During the Internet boom, I discovered some of Goldman Sachs’ persuasive analyses. Many people have either forgotten or are too young to reach extremes in the com era. From the logic of speculative hysteria – from the 17th-century “Tulipmania” in the Netherlands to the rise of Pets.com – FOMOlarger investment theories eventually combine to eliminate caution, common sense and financial gravity. In March, there are many FOMO Within the trend range of Wall Street, but not yet reached the level of the late 1990s. However, the five-month echoes are getting louder.

Consider Palantir Technologies, the Pentagon, the CIA and icenot to mention many commercial companies. A few days before Huang visited the White House, Palantir released a positive income report. By the end of this weekend, the market had revenues of more than 600 times over the past twelve months, with sales of about 130 times over the same period, according to Yahoo Finance database. Even in the late 90s, such figures caused eyebrows.

The eye-catching IPO is another feature of the Dot-Com era and is also making a comeback. In late July, Figma was a company that used software by Internet developers and added AI capabilities to its product suite and issued shares on the New York Stock Exchange for $33 per share. When the transaction started, the price jumped to eighty-five. The day closed at $115.50 – the issue price rose 25%. Looking at this market action, I remembered on August 9, 1995, when Netscape Nabscape Netscape made its public Netscape. Its stock price was $28, rising to $75 and closing at $58.25. From a percentage point of view, this leap is smaller than the first day of Figma stock, but is often described as the beginning of the internet bubble.

It should be noted that due to Figma’s IPO, its stock has dropped below $80. This can be interpreted as a sign of prevalence of sanity, but other private AI companies are encouraged to enter the stock market given that stocks are still trading at more than twice the issue price. Renaissance Capital is a research company specializing in IPOs, listing eight prominent candidates: OpenAI, Humans, Cohere, Databricks, Symphonyai, Waymo, Waymo, Scule AI and Confusion. Almost all of these companies are unicorns: fundraising deals with venture capitalists and other early investors, which are worth more than $1 billion. But, according to research firm TracXN, there are about 7,000 smaller and lesser-known AI companies nationwide, with more than a thousand of them already receiving Series A A funding from outside supporters to fund their operations.

The ready availability of early funds means the necessary conditions for the necessary tonal bubbles. Three more: investors’ excitement for groundbreaking technology – a huge snippet of generated AI that clearly has the potential to impact the economy; a Wall Street production line composed of investment bankers eager to earn IPO fees; and adaptive policies. Last month, the Trump administration announced thatArtificial Intelligence Action Plan“Aim to remove barriers to deploying new technologies and prevent states from introducing “heavy” regulatory AI laws. Meanwhile, the Fed appears to be preparing for the Reduce interest rates Next month, this could give the market another boost.

However, there are some important differences between now and the nineties, one of which is that the online economy is no longer a huge open plain where enterprising people can suggest building castles in the sky. This is a bastion of monopoly capitalism, where large technologies dominate the horizon. In any case, in the Point-COM era or its early stages, small startups can reasonably hope to take advantage of the first turnaround, traction early and create lasting commercial franchises. In the AI economy, it seems that many rewards may be attributed to top companies with the ability to build and maintain large AI models and leverage their market capabilities and finances to block or acquire potential competitors. A strong antitrust policy may prevent this from happening, but, as Wall Street Journal Report Last week, the government’s commitment to adopting such policies is now threatened by lobbyists and power brokers who have close ties to the president. If investors believe that monopoly is the future of an AI-driven economy, the results of the stock market are likely to mean further gains from existing industry giants rather than based on a wide range of bubbles.

Of course, all of this is uncertain. The AI boom is still in the stage of building infrastructure, namely training large language models, building data centers, etc. AI applications are just beginning to spread throughout the economy, and no one knows how transformative and profitable the technology is. In this environment, many investors are following the strategy of buying shovel manufacturers and large mine owners with a long-standing strategy. But history tells us that even such strategies are far from risk. In an interesting analysis, it was published on the financial news platform seeking Alpha, an analyst identified as KCI Research compared NVIDIA with Cisco Systems, one of its stocks that went parabolic in 1998-99. Just as Nvidia’s GPUs (Graphics Processing Units) are now widely considered a must-have component of AI infrastructure, Cisco’s routers and other network devices are also seen as basic components of Internet building. For some time, the demand for them seemed almost limitless. Like Nvidia, Cisco is an innovative and profitable company. However, in April 2000, its stock fell nearly 40%, and a year later it fell about 80%. For a quarter of a century, it still hasn’t recovered IT hits in early 2000, although it’s close lately.

NVIDIA-CISCO comparison is a useful reminder, reminding Pioneer stock analyst Benjamin Graham, who is a mentor to Warren Buffett: In the short term, the stock market is a voting machine, but in the long run, it is a weighing machine that can weigh the cash flow generated by the company. Ironically, the NVIDIA-CISCO analogy also inadvertently demonstrates how long it can last in the short term and the danger of predicting its end date. The analysis was published last February. Since then, NVIDIA’s share price has risen by another 55%. ♦