The managing director of the IMF and a new report from the Bank of England have joined a growing chorus of institutions issuing a stark warning: we are in an AI bubble, and it could be set to burst. The Bank of England’s Financial Committee stated, “equity market valuations appear stretched, particularly for technology companies focused on artificial intelligence.” In plain language, tech companies might be wildly overvalued, and given their massive weight in the stock market, we could be due for a major correction.
This script is not new. Historians can recall the tulip mania in the Netherlands in the 160s at the time a tulip bulb cost as much as a house. Savvy investors can also recall the dot-com hype when the stock market crashed erasing almost 80% of investor value. History is replete with manias where the revolutionary potential was larger than the reality of money. The current AI mania, driven by a shaky mix of hyperbole and fear of missing out (FOMO), is demonstrating all the classic symptoms
The "New" Decades Old Technology
To understand the fuss, one needs to separate science from spectacle. Artificial Intelligence itself is not new, and Generative AI—the impetus behind models like ChatGPT—is not new either. The theoretical foundation dates back to Alan Turing during the 1950s. Research on neural networks began in earnest form in subsequent decades. The 1980s saw innovation of early generative models, like Boltzmann machines. The real jump occurred in 2014 with the emergence of Generative Adversarial Networks (GANs) that could create highly realistic-looking images. Later, in 2017, a Google research team released a paper that presented the "Transformer" model. The model is the "T" in GPT (Generative Pre-trained Transformer), and it revolutionized the field by making the models capable of processing language with unprecedented coherence and fluency. What we see today is not the creation of a technology, but its mass consumer acceptance tipping point and commoditization, driven by mind-boggling computing power and even more mind-boggling hype.
The Anatomy of the AI Bubble: Circular Logic and Broken Fundamentals
The warnings are not based on speculation but on alarming economic data and unsustainable patterns. The Staggering Investment vs. Revenue Chasm: Analyses show that tech giants have invested approximately $560 billion in AI over the past two years, while generating only about $35 billion in AI-related revenue. This unsustainable burn rate is compounded by a recent MIT study finding that 95% of AI pilot projects fail to yield any meaningful results. The Circular Investment Loop (The Modern "Emperor's New Clothes"): This is one of the most telling signs of a bubble peak. A complex web of "strategic investments" is creating an illusion of value and demand. Nvidia, the dominant AI chipmaker, invests in its own customers, like OpenAI. OpenAI then uses that money to buy billions of dollars worth of chips from Nvidia. The cycle repeats: OpenAI strikes a deal with AMD to buy tens of billions in chips, and in return, becomes one of AMD's largest shareholders. When the deal is announced, AMD's stock surges, creating paper gains for OpenAI.
The Unsustainable Math: A report from Bay & Company estimates that by 2030, the cost of just the computing power needed for AI will reach $2 trillion, while revenues are projected to be only $200 billion. Revenues are not expected to exponentially rise to reach healthy figures while companies and individual users are becoming more susceptible about it's value.
What Happens When the AI Bubble Bursts?
The bubble will not pop. it will most likely deflate. When it does, the fallout will follow a familiar pattern. As the "Dot-Bomb" Scenario did during the dot-com bust, nearly all of those pure-play AI startups with weak business models will vanish. Trillions of market value will be lost, and then the heavy layoffs will begin. AI is not going anywhere. The AI industry is many years old driven by scientific progress. If companies like Google and Amazon survived and strived after the dot-com crash, so will AI firms that have strong fundamentals and useful applications.
The Conditioning Effect
Beyond the financial reckoning, a deeper struggle is unfolding. Just as the hype around institutional blockchains and the overregulation of fintech can be seen as methods to control the flow of money, the AI gold rush could also be a race to capture the world's data and condition user behavior.
The goal for the dominant players may not just be profitability, but to funnel all users towards a controlled set of Large Language Models (LLMs). These LLMs can be subtly tailored to respond in ways the algorithm designers see fit—shaping public opinion, reinforcing certain ideologies, and filtering information. This conditioning raises a critical question for the near future: how will regulations impact algorithm design? Will governments mandate "alignment" that serves state interests, or enforce "neutrality" in an inherently biased system? The current battles over data privacy and intellectual property are just the opening skirmishes in a much larger war over who controls the foundational models of human knowledge and communication.
The Bottom Line
The field of AI is not new, and it will not die. Generative AI is a powerful culmination of research dating back decades. However, the current investment frenzy, fueled by FOMO and propped up by circular financing, is building a castle on a foundation of sand. The proponents of "AI will solve everything" are hyping the market to fund a global data center build-out and consolidate control over the digital realm. But as the fundamentals—profit, revenue, and productivity gains—continually fail to support the valuations, a correction is inevitable. The real opportunity will not be chasing the hype, but waiting for the hard ground of true profitability and sustainable business models to emerge. The technology is real and here to stay, but the bubble, and the battle for control of it, s just beginning.
