Nvidia, Tesla, Microsoft, Alphabet, Meta, Apple, Amazon, otherwise known as the “Magnificent 7”, have clinched huge popularity among everyday investors in recent years. These are a cohort of very influential tech stocks and their soaring valuations have contributed to more than half of all S&P 500 gains since 2023. This veneer of dominance has convinced many that AI is not only the “future”, but the safest bet in the market.
Unfortunately, this mentality appeared awfully familiar, and many began to notice. When a handful of giants account for nearly all of the market growth, history informs us about one thing: we are simply no longer looking at a technological revolution alone, but perhaps a potential bubble.
Many investors lately are becoming convinced of the circular “ouroboros economy”, and they do have a reasonable basis.
“It would be like the equivalent of saying, if I gave $100 to you and you gave $100 back to me. Are we creating anything or are we just shifting the money back and forth?” Dylan Jennings, instructor of economics, observed.
Likewise, Nvidia in 2025 reached a record $5 trillion market capitalization by essentially bankrolling its own customers. Through the process of “Vendor Financing” , Nvidia and major cloud hyperscalers invest billions into AI startups like OpenAI, who then use those very funds to rent compute capacity or buy chips from their investors. This so-called investment-for-hardware swap weaves an illusion of breakneck revenue growth that ignores something crucial. OpenAI’s spending commitments have reached $1.4 trillion over the next eight years, while its annualized revenue at the end of 2025 was just a fraction of that, at approximately $20 billion.
“It seems kind of implausible that these companies will meet the spending, revenue, and loan commitments they have announced,” Jennings emphasized.
From the very beginning, Wall Street and the finance world treated AI as a stairway towards superhuman intelligence. However, the view from technical research and the lab is far more grounded.
“AI is appearing more and more like a hype now. Our expectation is higher than the actual value” computer science instructor Hazem Saleh noted, contending that “many companies today inflate their branding by claiming to use AI when they are merely repackaging older automation systems.”
Indeed, this AI branding mirrors what occurred in the late 1990s, when many firms began adding “.com” to their names to spike their stock prices.
It is also worthy to note that training GPT-3 cost roughly $4.6 million, GPT-4 with more than $100 million, and GPT-5 is projected to cost in the billions. This tells us that while the reported capability improvements between models remain relatively stagnant, the jump in cost is huge in comparison. This is also happening against the backdrop of MIT Sloan’s 2024 study showing that around 95 percent of businesses implementing AI have not seen quantifiable productivity gains. When the costs rise exponentially and the returns rise marginally, speculative inflation is the answer.
Speaking of the 1990s, It is pretty well-known that the “dot-com” bubble has been serving as a parallel. In fact, both moments are built on technological promises, both moments saw exaggerated market concentration where a tiny group of companies drove majority gains. Yet another similar parallel is frequently ignored: The Nifty Fifty, a group of established bluechip stocks in the 70s.
Unlike the dot-com which consisted of fantasy startups with no revenues, both the Nifty Fifty and the Mag 7 are composed of very well-known, dominant businesses. Firms like IBM, Polaroid and Coca-cola were called “One Decision” stocks that a no-brainer investor could put their money into. Similarly, the Mag 7 was treated by investors in this manner in recent years.
The Nifty Fifty also experienced extreme market concentration, with its top five performers making up about 25% of the S&P 500 index. Similarly, the top 5 stocks in Mag 7 accounted for roughly 28% of the S&P 500 at the end of last year.
To add on to this, the irrational valuations also serve as a striking parallel, with an average P/E(Price-to-Earning: how much investors pay for $1 earning) ratio of 42x for Nifty Fifty and around 53x for Mag 7: Both well beyond index average.
Even if AI does not collapse in a dramatic bubble, “there has to be some sort of correction, because the current valuations far outpace real revenue.” Jennings argued. AI related companies account for an estimated 40 to 50 percent of recent GDP growth. If AI spending cools down, and it inevitably will unless revenue rises to meet it, then the overall GDP growth could flatten or even turn negative, setting off a chain reaction of layoffs with the tech companies cutting their spending, local businesses losing income from those workers, and service industries contract. “The stock market could realistically face a 20 to 25 percent decline in tech-heavy indices like the NASDAQ.” Jennings speculates. “Much like the dot-com bubble, the technology will survive, but many of today’s highly valued companies may not.”