As A.I. drives up the Big Tech stocks, caution is building up among experts. Last week, Apollo’s chief economist, Torsten Sløk, pointed out that today’s top 10 companies by market cap are more overvalued than those during the tech bubble of the mid-1990s, where the Nasdaq saw an 800 percent surge before plummeting 740 percent by 2002.
Goldman Sachs recently released a note showing hedge funds unloaded tech stocks at the fastest pace in seven months in the week ended Feb. 23. A separate research note by Morgan Stanley noted a similar trend.
Many experts agree that today’s stock market looks eerily like the one before the dot-com crash of 2000. Bank of America’s Michael Hartnett, for one, called the current stock market a “baby bubble.” Similarly, economist David Rosenberg said markets are responding to companies advertising their AI capabilities just like when 90s tech stocks would appreciate “just because they added ‘.com’ to the name.”
At the center of the A.I. market boom are the “Magnificent Seven” companies, composed of Apple (AAPL), Amazon (AMZN), Alphabet (GOOGL), Meta (META) Platforms, Microsoft, Nvidia and Tesla (TSLA). These seven stocks alone represent 30 percent of the S&P 500’s market cap. The last time the S&P 500 was this top-heavy by market cap weight was 1970, per Axios.
Microsoft, Meta, Alphabet and Apple each described investing in A.I. among their highest priorities in recent earnings calls and reports. Meta CEO Zuckerberg said his company’s major goal moving forward will be “building the most popular and most advanced A.I. products and services.” Microsoft has touted its success in being an early investor of OpenAI. Amazon recently invested $1 billion in A.I. robot start-ups. Tesla, having long promised full self-driving, last month released an updated beta version that leverages better neural network technology.
NYU professor Gary Marcus highlights that A.I. capabilities may be overhyped, pointing to the persistent issue of A.I. hallucination, where systems generate or distort facts. He elaborates in his Substack, “Hallucinations are in their silicon blood… to blithely assume that the problem will soon disappear is to ignore 20 years of history.” This problem, Marcus argues, could hinder A.I.’s ability to deliver anticipated value to companies driving high financial valuations. Marcus points to similar overhyped A.I. products in the past, like Facebook’s A.I. assistant and IBM Watson.
One of the biggest limitations to A.I. capabilities, as described by University of Cambridge researchers, is the “curse of recursion,” which showcases how A.I. trained on A.I.-generated information causes lapses in its knowledge and less useful outcomes. As the internet fills with A.I.-generated content, LLM and text-to-video models that train on web-scraped data will plateau in their generative abilities. Moving outside of web-scraping to source training material for A.I. can be manual and costly.
Despite these hurdles, the transformative potential of A.I. cannot be dismissed. And many esteemed financiers, including Bridgewater Associates founder Ray Dalio and JPMorgan Chase CEO Jamie Dimon, don’t see an A.I. bubble. Alphabet CEO Sundar Pichai described A.I. technology as “more profound” for humanity than electricity or fire. McKinsey Global Institute estimates A.I. could increase worldwide corporate profits by up to $4 trillion a year.
While Nvidia symbolizes the excitement surrounding A.I., broader benchmarks like the MSCI USA IMI Robotics & AI Select Net USD Index have seen more modest gains of 47 percent. Goldman Sachs also wrote in a note to clients that call options on Nvidia are at a two-year high, showing positive speculative positions on the stock.