How did AI manage to put the whole world into a bubble?

Author: Sleepy.txt

"The only way to win is not to play the game."

In October, Michael Bury wrote this quote on social media. It comes from the 1983 film *WarGames*, in which a supercomputer reaches this conclusion after repeatedly simulating nuclear war.

A few days later, Bury disclosed his third-quarter holdings.

This investor, famous for his precise shorting of the 2008 subprime mortgage crisis, has staked nearly 80% of his managed fund's assets, approximately $1 billion, in one direction—shorting Nvidia and Palantir. In his view, the most effective way to avoid participating in this irrational "long" frenzy is to short it. Burry's bet isn't just against a few overvalued companies, but against the most powerful consensus of our time. Because in this consensus, AI is not just a technological revolution, but also a belief of capital. But how was this consensus formed? And how was it pushed to its climax? As this machine of belief continues to operate, what price are we paying for it? Behind every financial frenzy lies a story that is repeatedly told and believed by countless people. In this wave of AI, the way this story is written is textbook-worthy. It was accomplished by the combined efforts of three forces: technology leaders are responsible for writing the "myth," Wall Street provides the "rationality," and the media carries out the "evangelism." The first writers of this story were the evangelists of the singularity. Technology leaders, represented by Sam Altman, CEO of OpenAI, and Demis Hassabis, co-founder of Google DeepMind, have successfully transformed artificial general intelligence (AGI), a distant concept that once existed only in science fiction and academia, into a "new god" that is within reach, tangible, and capable of solving all of humanity's major problems. Altman, during his global speaking tour, repeatedly stated that AGI will be humanity's "greatest technological leap to date," and that the wealth it will bring will "far exceed all our imaginations." Hassabis, in more philosophical terms, defined it as a tool to help humanity understand the ultimate mysteries of the universe. Their language, imbued with a religious fervor for the "future" and "intelligence," has successfully imbued this technological wave with a near-sacred meaning that transcends commercial considerations. If tech leaders provided the script for the myth, then Wall Street and economists provided the "rational" endorsement for it. Against the backdrop of slowing global economic growth and frequent geopolitical conflicts, AI was quickly chosen as the "growth antidote" that could restore capital's faith in the future. Goldman Sachs released a report at the end of 2024 predicting that generative AI would contribute 7% to global GDP growth within a decade, approximately $7 trillion. Almost simultaneously, Morgan Stanley offered a more ambitious definition, stating that AI is "at the heart of the Fourth Industrial Revolution," with a productivity effect comparable to the steam engine and electricity. The real value of these figures and metaphors lies in turning imagination into assets and belief into valuation. Investors are beginning to believe that giving Nvidia a P/E ratio of 60 is not insane; they are not buying a chip company, but the engine of the future global economy. Since the launch of ChatGPT in November 2022, AI-related stocks have contributed 75% of the S&P 500's returns, 80% of its earnings growth, and 90% of its capital expenditure growth. This technological narrative has become almost the sole pillar supporting the entire US stock market. Finally, media and social networks have become the ultimate amplifiers of this myth. From the stunning debut of Vincent's video model Sora to every model update from giants like Google and Meta, each milestone has been amplified, looped, and amplified again. Algorithms have pushed this belief into everyone's timeline. Meanwhile, the discussion of "AI replacing humans" has spread like a shadow. From engineers to teachers, from designers to journalists, no one can be sure if they belong to the next era. When fear and awe spread simultaneously, a grand, almost unquestionable creation myth was written, paving the way for one of the largest capital assemblies in human history. As the "gospel" spread to every corner of the world, a group of financial engineers, masters of structural design, began to act. Their goal was to transform this abstract belief into a functioning machine—a self-circulating, self-reinforcing capital system. Rather than a bubble, it was more like a meticulously constructed financial engine, far more complex than the derivatives designs of 2008. At the core of this machine were a few tech giants. They wove capital, computing power, and income into a closed loop, within which funds flowed, amplified, and flowed again, like a perpetual motion system driven by algorithms. First, tech giants like Microsoft have invested heavily in AI research institutions like OpenAI. This company, which has a history of betting on infrastructure since the cloud computing era, has invested over $13 billion in OpenAI. In just a few years, OpenAI's valuation has soared from billions to nearly $100 billion, becoming a new legend in the capital market. The first consequence of this massive funding is more expensive training. To build GPT-4, OpenAI used over 25,000 NVIDIA A100 GPUs, and the computing power requirements for the next generation of models are growing exponentially. These orders naturally flowed to the market's sole monopolist, NVIDIA. Nvidia's data center revenue jumped from $4 billion in 2022 to $20 billion in 2025, with a profit margin exceeding 70%. Its stock price soared, making it the world's most valuable company. Those holding a large portion of Nvidia's stock are major tech giants, including Microsoft, and institutional investors. The rise in Nvidia's stock price further bolstered their balance sheets. The story doesn't end there; training is just the beginning, deployment is the main battleground for expenses. OpenAI needs to host its models in the cloud, and its biggest partner is Microsoft. Billions of dollars in cloud service fees flow into Microsoft's books annually, translating into growth for its Azure business. A perfect closed loop was thus born. Microsoft invested in OpenAI, OpenAI purchased GPUs from Nvidia and cloud services from Microsoft, Nvidia and Microsoft's revenue growth boosted their stock prices, and the rising stock prices made Microsoft's investment look even more successful. In this process, funds simply circulated among a few giants, yet created huge "revenue" and "profits" out of thin air. The on-paper growth corroborated each other, and valuations rose one after another. The machine began to feed itself. It didn't even need real demand from the real economy to achieve "perpetual motion." This core engine quickly expanded to various industries. Fintech and the payments industry were among the first to be integrated. Stripe is a prime example. This payments company, valued at over $100 billion, processed a total of $1.4 trillion in payments in 2024, equivalent to 1.3% of global GDP. A year later, it announced a partnership with OpenAI to launch an "instant checkout" feature in ChatGPT, allowing the payment system to be truly embedded in the interactive scenario of language models for the first time. Stripe's role in this wave is quite nuanced. It is both a purchaser of AI infrastructure, continuously buying computing power to train more efficient fraud prevention systems and payment recommendation algorithms; and a direct beneficiary of AI commercialization, creating new transaction entry points by combining with language models, thereby boosting its own valuation. PayPal followed suit. In October 2025, the long-established payment giant became the first wallet system to be fully integrated with ChatGPT. But the ripples didn't stop in finance. Manufacturing was one of the first traditional industries to feel the impact; in the past, it relied on automated hardware, but now it's paying for algorithms. In 2025, a German automaker announced it would invest €5 billion over three years to drive AI transformation, with most of the funds going towards purchasing cloud services and GPUs to reshape the nervous system of its production lines and supply chains. This is not an isolated case. Managers in industries such as automotive, steel, and electronics are all trying to improve efficiency in similar ways, as if computing power is the new fuel. Retail, logistics, advertising—almost every industry you can think of is undergoing similar transformations. They're buying AI computing power, signing cooperation agreements with model companies, and repeatedly emphasizing their "AI strategy" in earnings reports and investor conferences, as if those three words themselves bring a premium. The capital market has indeed rewarded them: valuations have soared, fundraising has become easier, and the narrative has become more complete. And all of this almost points to the same few companies. No matter which industry the funds flow from, they ultimately return to core nodes like Nvidia, Microsoft, and OpenAI, flowing to GPUs, to the cloud, and to models. Their revenues are thus climbing steadily, their stock prices are rising continuously, which in turn reinforces the belief in the entire AI narrative. **Costs** However, this machine is not without foundation. Its fuel comes from real economic and social resources, which are gradually extracted, transformed, and burned into the roar of growth. These costs are often obscured by the clamor of capital, but they do exist and are quietly reshaping the framework of the global economy. The first cost is the opportunity cost of capital. In the world of venture capital, funds always chase the direction with the highest returns. The AI ​​gold rush has created an unprecedented capital black hole. According to PitchBook data, in 2024, about one-third of global venture capital flowed into AI; by the first half of 2025, this proportion had climbed to a staggering two-thirds in the United States. This means that capital that could have supported key areas such as climate technology, biomedicine, and clean energy is being disproportionately absorbed into the same narrative. When all the smartest money is chasing the same story, the soil for innovation is being hollowed out. The focus of capital doesn't always mean increased efficiency; it often means the loss of diversity. In 2024, global venture capital investment in clean energy was only one-fifth of that in AI. Climate change remains considered humanity's most pressing threat, yet funds are flowing into computing power and models. The situation for biotechnology is no different. Several entrepreneurs admitted in interviews that investors showed little interest in their research because "the AI ​​story is more attractive, and the return cycle is shorter." This frenzy of capital has approached a dangerous tipping point. The year-on-year growth rate of capital expenditure in the US tech industry has now almost caught up with the peak of the dot-com bubble of 1999-2000. Back then, everyone was talking about a "new paradigm," companies expanded aggressively before becoming profitable, and investors rushed to bet on a "world-changing" vision. Until the bubble burst, Nasdaq lost two-thirds of its market capitalization, and Silicon Valley entered a long winter. Twenty-five years later, the same sentiment has been reignited, only this time the protagonist is AI. The capital expenditure curve has risen sharply again, with giants racing to invest tens of billions of dollars in building data centers and computing clusters, as if the expenditure itself could guarantee a certain future. The repetition of history is unsettling; the outcome may not be exactly the same, but this extremely concentrated capital momentum means that once the turning point arrives, the cost will be borne by the entire society. The second cost is the intellectual cost of talent. This AI boom is creating an unprecedented intellectual drain globally. Top engineers, mathematicians, and physicists are being drawn from the forefront of solving fundamental human problems in the same direction. In Silicon Valley, the scarcest resource today is not funding, but the top scientists in large-scale model teams. The salaries offered by companies like Google, Meta, and OpenAI dwarf those of all other science and engineering disciplines. Industry data shows that an experienced AI research scientist can easily earn over a million dollars annually; while in a university lab, a top physics professor often earns less than one-fifth of that. Behind this salary gap lies a shift in direction. The world's brightest minds are withdrawing from long-term fields such as basic science, energy innovation, and biological research, concentrating on a single, highly commercialized track. The flow of knowledge has never been faster, but the channels through which it flows are becoming increasingly narrow. The third cost is the strategic cost to the industry. Swept up in the AI ​​wave, companies in almost all traditional industries have fallen into a passive anxiety. They are forced to join this expensive AI arms race, investing huge sums of money and building AI teams, even though the vast majority of them do not have a clear roadmap for return on investment. According to data from Dell'Oro Group, global data center capital expenditure is expected to reach $500 billion in 2025, most of which is related to AI; Amazon, Meta, Google, and Microsoft alone plan to invest more than $200 billion. But this investment frenzy has long since transcended the boundaries of the technology industry. A large retail company announced in its earnings call that it will invest tens of millions of dollars over the next three years to purchase AI computing power to optimize recommendation algorithms and inventory systems. According to MIT research, the returns on most investments in these projects are far from covering the costs. For these companies, AI is not a tool, but a statement. Such investments are often not driven by proactive strategic needs, but by a fear of "falling behind the times." However, viewing this AI wave merely as a story of financial bubbles and resource misallocation is rather one-sided. Because regardless of whether the market tide rises or falls in the future, some profound and irreversible structural changes have already quietly occurred amidst this clamor. "Intelligence," and the computing power that drives it, is replacing traditional capital and labor as the new fundamental factor of production. Its position is like that of electricity in the 19th century and the internet in the 20th century—irreversible and indispensable. It is quietly permeating all industries, rewriting cost structures and the competitive order. [Image source: Sparkline] The competition for computing power has become the oil race of our time. The ability to control advanced semiconductors and data centers is no longer just a matter of industrial competition, but a core issue of national security. The US chip law, the EU's technology export ban, and policy subsidies from East Asian countries constitute a new geoeconomic front, and a global race for computing power sovereignty is accelerating. Meanwhile, AI is setting a new benchmark for all industries. Whether a company has a clear AI strategy has become crucial to winning the trust of the capital market and surviving in future competition. Whether we like it or not, we must learn to communicate with the world in the language of AI—it's a new business grammar and a new rule of survival. Michael Bury isn't always right; he's misjudged the situation multiple times over the past decade. This latest gamble may prove his foresight once again, or it may make him a tragic figure caught in the reshuffling of history. Regardless of the outcome, the world has been permanently changed by AI. Computing power has become the new oil, AI strategy has become a must-answer question for corporate survival, and global capital, talent, and innovation resources are converging in this direction. Even if the bubble bursts and the tide recedes, these changes will not disappear; they will continue to shape our world, becoming the irreversible backdrop of this era.