Is the Global AI Craze a Super Bubble Ready to Burst?

In the past two years, artificial intelligence (AI) has undoubtedly become the most frenzied sector globally. From the explosive popularity of ChatGPT to advancements in autonomous driving, from AI coding to AI in pharmaceuticals, the world is buzzing with excitement. The capital market has been betting wildly on this “new frontier”; in just the first quarter of 2026, global venture capital poured a record $300 billion into AI startups, accounting for nearly 80% of total global venture capital.

Tech giants like Microsoft, Google, Amazon, and Nvidia are investing heavily in computing power and data centers. Many startups can secure enormous funding simply by labeling their business plans with “AI.” Everyone is exclaiming that AI will reshape human society completely, just like electricity and the internet did in the past.

However, amidst this frenzy, a pressing question arises regarding the future of AI: Is it a genuine industrial revolution or a super bubble that is expanding?

1. Historical Echoes: AI Resembles the 2000 Internet Bubble

Looking back through economic history, nearly every significant technological revolution has been accompanied by a massive capital bubble. The story of the railroads in the 19th century, the automotive boom in the 20th century, and the internet bubble in 2000 share striking similarities: the technology is real, the future is promising, but capital often inflates expectations for decades in a single swoop.

The AI industry has clearly entered an overheated phase. The most direct manifestation is the significant disconnection between valuations and reality. Many AI companies are projected to lose billions in the coming years yet still achieve astonishing valuations in the hundreds of billions. Investors have shifted their focus away from traditional business logic such as profits and cash flow, falling into a worship of parameters and a belief in computational power.

The market is flooded with numerous “AI shell companies” that lack core technology, merely wrapping themselves in a layer of large model API interfaces. Even traditional companies selling skincare products or providing SaaS software are forcibly branding themselves as “AI-enabled” to deceive investors into high valuations. This absurdity mirrors the internet bubble of 2000, where simply adding “.com” to a company name could skyrocket its stock price.

However, there is a critical difference between today’s AI bubble and that of the past: who is footing the bill? In 2000, the internet bubble was inflated by numerous unprofitable shell companies relying on retail investors to drive up prices. Today, the core players in AI are the world’s most profitable tech giants with strong cash flows. This is not a group of broke gamblers borrowing money; it’s a collection of the wealthiest companies in the world engaging in an AI arms race with hundreds of billions of dollars.

As a result, there is a strong risk-averse foundation that prevents an immediate collapse like in the past, but that does not mean danger is absent.

2. Calm Before the Storm: Four Danger Signals Piercing the Bubble

If we penetrate the surface of the frenzy, we find that the underlying logic of the AI industry is facing several unavoidable hard issues:

1. The $600 Billion “Recoupment Black Hole”

The biggest risk is that the rate of cash burn far exceeds the rate of revenue generation. Tech giants are investing over $400 billion annually in AI infrastructure. Analysts have calculated that to cover the enormous costs of GPU chips, data centers, and electricity bills, the entire AI industry needs to generate approximately $600 billion in revenue each year to break even. However, the reality is that current AI applications are not even close to achieving this target. The entire industry is severely overextending future earnings.

2. Physical Limits: The End of Computational Power is Electricity

Many believe that the bottleneck for AI is Nvidia’s GPUs, but the real constraint is the global power grid. AI is notoriously power-hungry. Some new super data centers will consume as much electricity as a nuclear reactor. However, the aging power grid cannot keep up with this demand. If GPU clusters purchased for billions of dollars are rendered idle due to power shortages, the massive investments will become worthless in just a few years due to the 60% annual depreciation of AI hardware.

3. Technical Concerns: Human Data is Nearly Exhausted

The AI industry currently believes in the “Scaling Law”—that simply piling on computational power and feeding data will make AI smarter. However, the harsh reality is that high-quality human text data on the internet is limited and is expected to be “used up” by 2028. At that point, AI will only be able to train on “AI-generated data,” leading to severe “model collapse”—content homogenization, decreased creativity, and increased hallucinations. Relying solely on burning cash to pile on parameters may soon hit an intellectual ceiling.

4. Black Swan Events: When “Intelligence” Becomes as Cheap as Tap Water

In early 2025, the Chinese startup DeepSeek released an extremely low-cost open-source model that nearly matched the performance of top Western models. This caused a seismic shock on Wall Street, with Nvidia’s market value evaporating by nearly $600 billion in a single day. Why? Because capital panicked. The market suddenly realized that achieving top-tier AI does not necessarily require exorbitant computational power. If algorithm optimization can replace brute-force chip stacking, and if future AI computational power becomes as cheap as tap water, then the data centers and high-priced API business models built on trillions of dollars will be completely overturned.

3. After the Bubble Bursts: Who Will Be Left Standing?

If this super bubble truly bursts and reshuffles the industry in the next two to three years, what will the world look like?

First, there will be a brutal cleansing. AI startups lacking core technology, stable revenue generation, and relying solely on financing will perish in droves. Venture capital will evaporate, and many inflated valuations will plummet overnight. However, tech giants like Microsoft and Google will seize the opportunity to acquire bankrupt companies’ computational resources, technologies, and talent at rock-bottom prices, ultimately leading to a few dominant players monopolizing the market.

Second, the economy will experience a painful transition period. In the initial phase of introducing new technology, companies will see productivity decline as they restructure management systems and train employees (the “productivity J-curve”). Only those who endure this painful period and complete their digital transformation will be able to experience a surge in the next decade.

Finally, there is a question that concerns everyone: Will your job be replaced? In recent years, the tech layoffs have not significantly impacted employment due to AI. AI is fundamentally a probabilistic prediction tool; it makes mistakes, and human judgment remains irreplaceable in high-risk areas like healthcare, finance, and law. In the future, AI will not directly replace doctors or programmers but will eliminate those who cannot use AI effectively. The truly valuable skills will be those of hybrid talents (Human-in-the-loop) who can harness AI, validate its results, and collaborate with it. Social stratification will depend on your ability to “control the machine.”

Conclusion

We must understand that the technologies that truly change the world have almost always emerged alongside bubbles. Railroads did, the internet did, and AI will too. The essence of a bubble is that capital, in a frenzy, builds foundational infrastructure for the future society without regard for costs. What may seem like excess data centers, fiber optics, smart grids, and computational hardware today will become the cornerstone of human civilization’s leap in ten years.

Thus, in the face of the surging wave of artificial intelligence, the most critical question is not whether it is a bubble, but rather—when the bubble recedes and the moment of world reshaping arrives, who will still be at the table?