Is the AI Bubble Real in 2026? Dotcom Crash Lessons Explained Summary

The current trajectory of artificial intelligence has moved beyond mere technological curiosity into a massive industrial buildout that rivals the greatest infrastructure projects in history. Three years since the launch of ChatGPT, the landscape has shifted from experimental software to a capital-intensive arms race. Leading the charge are the AI chip companies, most notably NVIDIA and AMD, who provide the specialized hardware necessary for model training. These firms have seen their market capitalizations skyrocket, with NVIDIA briefly becoming the world’s most valuable company. Supporting this are the infrastructure providers—Amazon (AWS), Microsoft (Azure), and Oracle—who invest billions in data centers to rent compute power to the next tier of the ecosystem.

At the peak of this structure are the model developers like OpenAI, Anthropic, and xAI. While these companies are the faces of the revolution, they are currently operating as massive "money furnaces." For instance, OpenAI is projected to lose $8.5 billion in 2025 alone, with estimated total spending reaching $115 billion by 2029. This aggressive spending is facilitated by the belief that being the first to achieve Artificial General Intelligence (AGI) will yield returns that dwarf these initial costs. However, the sheer volume of capital being deployed is staggering, with AI industry spending reaching nearly 20% of the total capital expenditures across all US industries.

Despite the excitement, voices of caution are growing louder. Michael Burry, famous for his 2008 housing market short, has recently taken positions against key AI firms. Furthermore, organizations like the International Monetary Fund (IMF) and the Bank of England have issued warnings regarding soaring valuations. The core concern is whether the revenue generated by AI applications can ever justify the $7 trillion in capital expenditures estimated by firms like McKinsey for the next five-year cycle.

One of the most controversial aspects of the current AI market is the emergence of circular financing deals. This occurs when a hardware vendor, such as NVIDIA, invests capital into a startup or data center provider, which then uses that exact capital to purchase chips from the vendor. While not illegal, this practice can lead to "propped-up" demand and may overstate a company's true profit margins. By investing in its own customers, a company effectively discounts its products without reflecting that hit in its operating margins, as investments are accounted for differently than revenue costs.

These arrangements draw haunting parallels to the vendor financing seen with companies like Nortel during the twilight of the Dotcom bubble. Critics argue that if the end-user demand for AI does not grow fast enough, the entire circle could collapse. We are already seeing AI companies borrow billions of dollars using their existing chip inventory as collateral for more loans. This leverage adds a layer of systemic risk to the industry, where the failure of one major node could trigger a cascade of defaults across the financing chain.