The AI Bubble and the Ghost of Netscape: Is 2027 the Year It All Pops?

Everyone is tired of hearing the word “Singularity.” When does it arrive? Many people point to 2027, citing a leaked OpenAI document from 2024 that suggested the company would create an Artificial General Intelligence by then — a machine that thinks and learns like a human, with autonomous consciousness and emotions.

But a growing chorus of economists and tech-sector analysts are warning that 2027 might not mark the birth of the AI era. It might mark the day the AI bubble bursts.

The man with a date on the calendar

Victor Gao, Vice President of the Center for China and Globalization, is one of the few voices willing to put a specific timeline on the crash. Between May and July of this year, Gao stated publicly and repeatedly that the United States is sitting on a powder keg. He predicts a global financial crisis will erupt between late 2026 and the first half of 2027 — one potentially ten times more destructive than the bursting of the dot-com bubble in 2000.

The trigger? AI.

“A financial crisis of almost unprecedented scale will unfold over the 12 to 18 months following January 1, 2026,” Gao said. “Where will the climax occur? I believe it will happen right here, in the U.S. Where will the endgame play out? That, too, will happen in the U.S.”

He is not alone in this concern, but he is unusually precise about timing.

The numbers behind the nervousness

The anxiety has a factual basis. Let us consider a few data points.

AI-related stocks currently account for roughly 45% of the S&P 500’s total market capitalization. According to research by the investment bank Jefferies, if the AI sector were stripped out entirely, the S&P 500 would have risen by only 2% this year. The U.S. stock market isn’t just benefiting from AI — it is being carried by it.

And it isn’t just the stock market. Oxford Economics economist Michael Pearce estimates that about one-third of current U.S. economic growth is driven by the AI sector.

What makes the numbers look even more fragile is how the money is actually moving.

The circular financing trick

Much of the AI industry’s eye-popping investment figures are inflated by a practice called circular financing — or as it was known during the dot-com era, “vendor financing.”

Here’s how it works: Company A invests in Company B. Company B uses those funds to purchase products and services from Company A. The money goes in a circle, but each transaction shows up on both companies’ books as revenue and investment. A simple analogy: imagine a coffee machine supplier who lends you money to open a café, and you use the loan to buy coffee machines from that same supplier. On paper, both parties are growing. In reality, the money is just changing hands between the same players.

In the AI industry, this translates to chipmakers and cloud providers investing in AI model companies, who then turn around and buy chips and rent cloud capacity from the investors. These deals can reach hundreds of billions of dollars, but much of it is simply shuffling funds across the balance sheets of a handful of AI giants.

If a single link in this chain loosens, the domino effect could be severe.

Caption: When Company A invests in Company B, and Company B buys from Company A, the “growth” is a loop. Photo created by Claude

What the other people are saying

Gao is far from the only prominent voice raising alarms, though analysts disagree on timing.

Wang Yuquan, founder of Haiyin Capital, believes the bubble will burst in 2029 rather than 2027. Moody’s Analytics has offered a grimmer forecast, suggesting a correction could come as early as late 2026,if AI revenue growth falls short of expectations, panic selling could trigger a 25% stock market crash, wiping out $20 trillion in market capitalization.

Ray Dalio, founder of Bridgewater Associates, frames the danger window between the U.S. midterm elections and the next presidential inauguration — November 2026 to January 2029. His reasoning: once “billionaire tax” proposals surface in Congress, major shareholders will cash out to avoid heavy taxation, triggering a sell-off.

Michael Burry,the investor who predicted the 2008 subprime crash and inspired The Big Short — believes the U.S. stock market is on the verge of a major reversal. He compares the current situation to 2000. Critics point out that Burry has been calling for a crash for years, earning the nickname “the boy who cried wolf.” His counterargument is simple: just because the wolf hasn’t arrived yet doesn’t mean there is no wolf.

Bank of America analyst Michael Hartnett notes that on the final trading day of May 2026, the S&P 500 hit a record closing high — but among the twenty stocks that also reached all-time highs that day, only seven were not directly linked to AI. That kind of concentration mirrors what analysts saw at the peak of the dot-com bubble twenty-six years ago.

Rewind: what actually happened in 2000

To understand why experts keep reaching for the dot-com comparison, it helps to know what actually happened.

The spark was a young programmer named Marc Andreessen. While still a student at the University of Illinois, he helped build the world’s first image-capable web browser ,initially called Mosaic, later renamed Netscape. The browser monopolized the market with over 90% share.

On August 9, 1995, Netscape went public. The company was sixteen months old and had never turned a profit. The shares were priced at $28 — doubled from the original plan at the last minute — and rocketed as high as $74.75 on opening day before closing at $58.25. The company’s market cap hit nearly $3 billion in a single day.

For context: it took General Dynamics 43 years to reach the same valuation.

https://youtu.be/4aN5TbGW5JA?si=CFTgrJa_sFYyG9LX

The Pygmalion Effect — a concise explainer of the dot-com bubble’s mechanics

The Netscape IPO sent a signal to Wall Street: the internet was the next steam engine. Thousands of startups flooded into the market. In 1999 alone, 457 U.S. companies went public, and the Nasdaq rang the opening bell for new listings almost daily. Of these hundreds of companies, 308 belonged to the tech sector.

The era had its own valuation logic. Nobody measured companies by P/E ratios — Price-to-Earnings. The metric that mattered was what people jokingly called the P/D ratio: Price-to-Dream. You could slap an “i-” or “e-” prefix on any company name, or add “.com” to the end, and raise tens of millions of dollars. It didn’t matter if you were selling pet food or toilet paper online. The market didn’t care about your business plan. It cared about your domain name.

Paper millionaires

The stock surge created a new social class: “paper millionaires.” Startup employees holding stock options worth millions — on paper. But there is a vast difference between wealth and cash. A real millionaire can deposit money in a bank and spend it. A paper millionaire’s fortune sits in a hypothetical bank, and it can only be realized if someone else is willing to buy at the same inflated price.

Consider an internet company with $10 million in actual assets but a $500 million market valuation. It appears to have conjured $490 million from thin air. But that $490 million doesn’t buy groceries unless a buyer materializes.

Meanwhile, online brokerages were pulling ordinary people into the market. The number of online brokerage accounts surged from 7.3 million in 1998 to 17.4 million by 2000. When even your grandmother was convinced that buying internet stocks was a guaranteed win, the frenzy was already nearing its end.

So is this time different?

The four most expensive words in financial history, according to the saying, are “this time is different.”

Defenders of the current AI boom make a fair point: unlike the dot-com startups of the late 1990s, today’s AI leaders are enormously profitable. NVIDIA, Microsoft, Google, and Meta generate real revenue at scale. The technology is already embedded in industries from healthcare to defense. This is not a market built on domain names and dreams.

But the bears have their own fair point: profitability in a handful of companies doesn’t make an entire sector immune to a correction. If AI revenue growth disappoints expectations — even modestly — the concentrated weight of AI stocks in the S&P 500 means any sell-off would be felt across the entire economy, not just in tech.

And the circular financing structures, the unprecedented concentration of market gains in a tiny number of AI-linked companies, and the eye-watering U.S. national debt (now past $39 trillion, or over 100% of GDP) all create the conditions where a small trigger could cascade into something much larger.

What can ordinary people do?

If you’re not a hedge fund manager or a venture capitalist, what does any of this mean for you?

First, understand concentration risk. If your retirement portfolio is heavily weighted toward the S&P 500, you are more exposed to AI stocks than you probably realize，roughly 45% exposed, in fact. Diversification across sectors and geographies is the oldest piece of financial advice, and it has never been more relevant.

Second, remember what “paper wealth” means. The dot-com era taught us that valuations are not cash. Unrealized gains are just that — unrealized. If the music stops, the people who moved to safety before the rush will fare better than those who assumed the party would go on forever.

Third, don’t try to time the crash. Nobody, not Gao, not Burry, not Dalio, can tell you the exact date the bubble pops — if it pops at all. What matters is not predicting the day, but being positioned to survive whatever comes.

History doesn’t repeat, the saying goes, but it rhymes. The melody coming from Wall Street right now sounds awfully familiar.