Is the AI bubble about to pop? | The Take

Key Concepts

• AI Bubble: The current surge in investment and valuation of AI companies, with concerns about overvaluation and potential market correction.
• Market Valuation Concerns: Doubts about whether the current high valuations of AI companies are justified by their actual performance and future potential.
• Market Correction: A significant decline in stock prices, often following a period of rapid growth and speculation.
• Enabling Technologies: Technologies that facilitate the development and deployment of other technologies, such as the chips produced by Nvidia for AI.
• Large Language Models (LLMs): AI models trained on vast amounts of text data, capable of understanding and generating human-like text.
• Circular Financing: A situation where money flows in a loop between companies, potentially inflating valuations without underlying value creation.
• Moat: A competitive advantage that protects a company's market share and profitability.
• Dotcom Bubble: A speculative bubble in the late 1990s and early 2000s related to internet-based companies, which eventually burst.
• Job Displacement: The potential for AI to automate tasks currently performed by humans, leading to job losses.
• Normal Technologies: Technologies that are stable, predictable, and integrated into everyday life, allowing for business building and planning.

The AI Boom: Cracks in the Hype?

The artificial intelligence (AI) boom is currently facing scrutiny, with investors expressing concerns about market valuations and the growing volatility of AI stocks. This has led top bank CEOs to issue warnings of a potential market correction. The central question is whether the current AI surge is a sustainable revolution or an unsustainable bubble.

The Concentration of Value and Market Dynamics

A significant aspect of the current AI landscape is the immense value captured by a small number of companies that have released groundbreaking technologies. These include AI developers like OpenAI (creators of ChatGPT and Sora) and Anthropic, along with crucial enablers like Microsoft and chip manufacturers such as Nvidia. The market has been repeatedly told that these technologies will fundamentally change the world, leading to sustained investment and boosting the valuations of a relatively small portfolio of AI-related stocks.

This concentration of value is evident in market performance. Approximately 80% of US stock market gains in 2025 have been attributed to a handful of AI-related companies, including Nvidia, Meta, and Alphabet. A notable graphic illustrates Nvidia at the center, pouring cash into other AI companies, which then use that same cash to purchase Nvidia chips. This circular financing model is viewed with suspicion by some, suggesting a degree of self-dealing and speculation rather than pure value creation. As Paul Ford, president of an AI startup, notes, "when billions of dollars are getting moved without actually getting moved, um you got to start looking a little more closely, right?"

Warning Signs of a Potential Bubble

Several indicators suggest that the AI boom might be peaking:

• Hype vs. Reality: The tech industry has a history of hyping new technologies, as seen with blockchain, often creating a sense of urgency and fear of missing out.
• Concentrated Value: The bet that all value will be captured by a few dominant companies like Nvidia is risky. Historically, internet technology tends to distribute value more broadly. If this doesn't happen, it could stifle future innovation.
• Circular Financing: The self-reinforcing loop of investment between AI companies and chip manufacturers like Nvidia raises questions about the sustainability of these valuations.
• Speculative Investment: Companies are investing heavily based on the promise of future "magic tricks" from AI, which may not materialize as expected.
• International Competition: The emergence of competitive AI systems from China, developed at a fraction of the cost, challenges the narrative of an insurmountable "moat" for US tech giants. The market reacted negatively when DeepSeek announced an LLM trained more quickly and cheaply, highlighting the vulnerability of US dominance.
• Investor Skepticism: Prominent investors like Michael Bur, who correctly predicted the 2008 housing crash, are betting against AI giants like Nvidia and Palantir, deeming them overvalued.

The Global AI Landscape: US vs. China

The current conventional wisdom in the US frames AI as an arms race. However, China's ability to develop competitive AI systems at a lower cost poses a significant threat to the valuations of US tech giants and could accelerate a potential AI bubble burst. The market's reaction to DeepSeek's announcement, where the market "tanked for a day," underscores this concern. While open-source models from China may be harder to access, they are often very good, effective, and cheaper to use.

The US approach is characterized by a "more fuel in the tank, whatever it takes" mentality, focusing on building data centers and relying on capitalism. China, in contrast, adopts a more strategic approach, emphasizing competition and cost-effectiveness. The narrative of "true great American genius" behind AI is being challenged by the rapid shrinking of the technological moat and the emergence of capable, cheaper alternatives.

What's at Stake if the Bubble Bursts?

The potential bursting of the AI bubble carries significant implications:

• Financial Losses: A bubble burst could lead to substantial financial losses for investors, with trillions of dollars held by American households potentially wiped out. This is particularly concerning as roughly two-thirds of Americans own stocks, and retirement accounts could be deeply affected.
• Economic Impact: While not as fundamental as the 2008 housing crisis, a significant AI bubble burst could still trigger downstream effects across the economy. The tech industry itself is deeply embedded in the broader economy, and a collapse could impact various sectors.
• Job Market Disruption: The hype around AI replacing jobs is a significant concern. Layoffs have already begun, partly due to overhiring in previous years. A bubble burst could exacerbate this trend, leading to further job losses across multiple industries. The irony is that jobs seem to be at risk regardless of whether the bubble bursts or not, as companies invest in automation.
• Innovation and Growth: If value remains concentrated in a few companies, it could limit future innovation and growth, dictating what the public sees and experiences based on the visions of a few powerful entities.

Lessons from the Dotcom Bubble

The current AI surge bears striking similarities to the dotcom bubble of the late 1990s. Both periods were characterized by immense market belief, driven by greed and fear of missing out. The dotcom bubble saw a significant market correction, with the NASDAQ falling by 80% and taking 15 years to fully recover, wiping out about $5 trillion in market value.

Key lessons from the dotcom era include:

• The cyclical nature of markets: Markets experience periods of irrational exuberance followed by corrections.
• The importance of diversification: Investors need to hedge their bets by balancing speculative investments with more stable assets like fixed income and bonds.
• The enduring value of useful technology: While speculative bubbles burst, fundamentally useful technologies tend to persist and evolve. The internet, despite the dotcom crash, became a foundational part of modern life.
• The potential for a "cleared-up" market: A bubble burst can sometimes lead to a more open and rational market, allowing for genuine value creation to emerge.

The Future of AI and "Normal" Technologies

The rapid evolution and improvement of AI technology are undeniable. Three years after ChatGPT's public release, new models are launching at an exponentially increasing rate. However, the focus is shifting from AI as a potential "galactic star baby" to its practical applications as software.

Paul Ford expresses a preference for "normal technologies" – those that are stable, predictable, and allow for building businesses and planning for the future. The hope is that AI will eventually become integrated into this familiar landscape. While the exponential changes in AI capabilities might be slowing, the development of practical, "classic tech industry products" built on these models is accelerating. This means more useful applications and less experimental ones.

The challenge lies in adapting to the speed of this integration. Building four apps in a weekend, which previously would have taken months, highlights the rapid pace. The conversation needs to shift from the existential potential of AI to its role as a tool. The goal is for AI to fit back into the existing world of software, enabling productive conversations about its long-term development and application.

Conclusion

The AI boom is at a critical juncture. While the underlying technology holds immense potential, concerns about overvaluation, concentrated market power, and the potential for a significant market correction are valid. The parallels with the dotcom bubble serve as a cautionary tale, emphasizing the importance of realistic expectations, diversified investments, and a focus on sustainable value creation. The future of AI hinges on its ability to transition from a hyped phenomenon to a "normal" technology that integrates seamlessly into society, driving innovation and progress without causing widespread economic disruption or job displacement. The conversation is far from over, and readiness for these discussions is crucial for shaping the long-term impact of AI.