Summary of Key Points
This article highlights that the newly established AI task force at the Federal Reserve (Fed) is only focusing on the impact of AI on employment and inflation, while overlooking the more critical risk of financial stability. The AI industry is currently engaging in large-scale borrowing for infrastructure projects (such as data centers and chips), but the return on these investments is highly uncertain, and the debt levels far exceed current revenues. The credit market poses risks similar to the 2007 shadow banking crisis. If the debts cannot be repaid, the Fed may be forced to provide assistance, which could undermine its independence. On the other hand, if AI becomes successful, it could lead to an unprecedented employment shock. The author urges the Fed to prioritize financial stability in its assessment of AI.
Detailed Analysis
1. The Fed’s “Blind Spot”: Why Is Financial Stability Being Ignored?
The Fed has three main objectives: to promote employment, stabilize inflation, and maintain financial stability. The new task force is solely focused on the first two, completely neglecting the third. Although Powell has discussed the impact of AI on bank cybersecurity, this is just a minor risk; the broader systemic crisis posed by the industry’s borrowing spree has not been addressed. It’s like a doctor treating only the symptoms of a cold while ignoring the underlying pneumonia—a classic case of missing the bigger issue.
2. The “Debt Hole” in AI Infrastructure: Can the Borrowings Be Repaid?
AI companies are borrowing heavily to build data centers and purchase chips. Sequoia Capital has estimated that the infrastructure investments made since the launch of ChatGPT would need to generate at least $3 trillion in profits to recoup their costs, but Anthropic, a leading AI company, only earns $60 billion per year—50 times less. Bain & Company warns that by 2030, the AI industry will need to earn an additional $2 trillion annually to meet its computational demands. This is akin to taking out a $1 million loan to open a milk tea shop, with a monthly mortgage payment of $10,000, but only earning $2,000 in revenue—how long do you think that would last?
3. The “Hidden Bomb” in the Credit Market: More Dangerous than the Stock Market
The funds for AI are no longer coming from tech giants themselves; they are sourced from the capital market, creating a cycle of financing: chip manufacturers invest in AI labs, which in turn use the money to purchase chips; cloud service providers fund startups, which rent their servers. This is like a snowball effect—more investment leads to higher prices, which in turn drives more investment. However, if doubts arise about the profitability of AI, the entire system could collapse. Moreover, the chips used as collateral become obsolete in 3-5 years, depreciating faster than the debts can be repaid. The real risk lies in the credit market: if institutions lose confidence in AI’s ability to repay, short-term financing could dry up, forcing companies to sell assets at reduced prices, similar to the 2007 shadow banking crisis (where it was not individuals but institutions that stopped lending to each other).
4. The Fed’s Dilemma: Rescue or Not Rescue?
If AI-related debts default, the Fed might be forced to assist non-bank institutions (such as private lending funds) and data center operators, just as it did with money market funds in 2020. However, the current reputation of the AI industry is not as strong as that of Wall Street in 2007, and providing support could lead to public dissatisfaction, potentially compromising the Fed’s independence. If AI succeeds, it could result in a massive job loss, an unprecedented employment crisis. It’s like flipping a coin: either financial chaos or a wave of unemployment—either outcome is problematic.
5. Historical Lessons: Even the Most Advanced Technologies Can Fail Due to Mismanaged Debt
Past booms in railroads and the internet have shown that while technology can transform the world, excessive investment with inadequate returns can lead to widespread losses for creditors and shareholders. The same principle applies to AI. No matter how transformative the technology, if the debt exceeds the potential earnings, it can still fail. Therefore, the Fed must not solely focus on employment and inflation; financial stability must be a central consideration in its assessment of AI. The fundamental principle of finance remains unchanged: mismatch between debt and cash flow will inevitably lead to problems.
The core message of this article is that, despite the immense potential of AI, the issue of debt cannot be ignored. The Fed’s current focus is misplaced, and it must urgently prioritize financial stability. Otherwise, it will face either a credit crisis or an employment disaster—both with detrimental consequences.