Scholar recommends policy changes to address potential AI-driven market crash.
As millions of users flock to artificial intelligence (AI) platforms such as Claude and ChatGPT, the AI market has gained billions of dollars in capital investments. But according to market analysts, AI firms have yet to generate proportionate revenue.
In a recent report, Asad Ramzanali of Vanderbilt Law School cautions that this rapid expansion of the AI industry could lead to an economic crash similar to the dot-com crash of the early 2000s or even the 2008 financial crisis. As part of his analysis, Ramzanali recommends policy solutions which might prevent this crisis.
Economic bubbles occur when the market value of an asset exceeds its actual profitability. Such conditions often result from rapid and large-scale investment into new technologies. When an asset fails to generate enough true value relative to the mass of wealth invested into it, the bubble bursts, leading to a plunge in market value, decreased earnings, and large losses for investors.
The greater and more concentrated the investment, the greater the impact on both the industry and the economy as a whole. Ramzanali argues that the current state of AI investment has the potential to create an economy-wide crash similar in scale to previous major financial crises.
Since the launch of ChatGPT in late 2022, the AI market has experienced substantial growth. AI companies have accumulated trillions of dollars in market value, with industry investments totaling $1.5 trillion in 2025 and spending expected to reach $2.5 trillion by 2026. In June 2026, AI companies Anthropic and OpenAI filed for initial public offering with an expected share valuation of up to $1 trillion.
Ramzanali warns that this growth is not sustainable. As demand for AI software rises, so does the cost of infrastructure. So far, the cost of AI data centers reached an estimated $50 to $60 billion per gigawatt of capacity in 2026. Major AI developers are currently operating at a loss, not expecting to turn a profit for several years. A 2025 report by the Massachusetts Institute of Technology’s Networked AI Agents in Decentralized Architecture project even found that 95 percent of generative AI projects fail to generate a return on their investments.
Ramzanali argues that, should AI investment turn into a bubble, the subsequent crash would extend far beyond the tech industry, causing mass bankruptcies, supply shortages, price spikes, and wide-scale job loss.
Ramzanali identifies several common financial schemes in the AI industry that could contribute to an AI crash. Circular equity financing schemes—where AI, computer chip, and cloud computing companies invest capital into each other—can lead to distorted pricing. Debt financing deals, where companies borrow capital to mask their debt, can obscure the risk of investment. Competitive data center subsidies and tax benefits can incite a “race to the bottom,” with state and local governments offering increasingly large tax breaks while creating few permanent jobs.
Ramzanali recommends that the U.S. Congress introduce laws restricting these practices and setting rules on how public dollars can fund AI development. He also suggests that the U.S. Department of Justice be mindful of accounting discrepancies in the AI industry and enforce existing criminal statutes to prevent fraud.
Financial fraud can often contribute to an economic crisis. Although federal statutes criminalize various forms of financial falsification and misuse, such laws are not always enforced. Lack of effective penalties can encourage behaviors that exacerbate future market crashes.
Beyond the economic and commercial consequences of the AI bubble bursting, a financial crash would chill research and investment into AI technology. Ramzanali suggests replacing corporate investment with government-funded research into AI for public purposes to preserve AI technology following a market crash.
An AI crash would bankrupt large swaths of the specialized data center infrastructure that AI software relies upon. Ramzanali recommends that Congress authorize federally funded research and development centers to purchase these data centers and establish a public option for cloud computing. These government-run programs would promote competition and supply chain resilience by providing public access to AI technology.
To protect workers in the event of an AI crash, Ramzanali suggests that Congress adopt policies that support the unemployed, create jobs, and reduce staff cuts in the technology sector. These policies would include expanding unemployment insurance, removing work requirements for federal social safety nets, and instituting mass employment programs like those in the 1935 Second New Deal.
Ramzanali also emphasizes that many AI companies are producing not just AI models, but also the data centers and computer chips that power them. This kind of vertical integration distorts markets and could accelerate an AI crash.
As a result, Ramzanali calls for statutory requirements separating AI software development and physical data infrastructure, similar to the Glass–Steagall legislation of 1933 which structurally separated commercial and investment banking. This proposed policy would not limit commercial transactions or agreements, but instead force companies to divest from either the hardware or software side of the industry.
Ramzanali concludes that, given the unique and complex nature of AI as both a technology and an economic product, no single federal agency is equipped to manage the industry. Rather, he proposes that Congress could create a new federal agency with the specific resources and authority to regulate digital markets and resolve issues specific to the industry.
In just a few short years, the United States economy has become increasingly dependent on the AI industry. Ramzanali believes that if Congress implements his recommended policy choices, the United States will have a better chance of preventing or reducing the harm of an economically debilitating AI market crash.
