AI Bubble: The Catalyst for a New Round of Quantitative Easing in the US?

```

The US AI industry is at a critical point where prosperity and risk coexist.

Currently, the prosperity of US AI is highly concentrated on hardware and cloud services. The incremental capital expenditures of cloud computing giants between 2023 and 2025 have already surpassed the total of the previous seven years, and it is expected that the four largest hyperscale cloud vendors will reach $305.5 billion in capital expenditures by 2025, with AI-related spending contributing 90% of the incremental capital expenditures in the S&P 500.

Meanwhile, foundational model companies represented by OpenAI remain deeply loss-making, commercialization of AI applications is seriously delayed, and the structural imbalance of "strong hardware, weak software" is increasingly pronounced.

Although the current bubble is still in the brewing stage and fundamentally different from the internet bubble of 2000, with the S&P 500 Shiller PE ratio approaching 44, a historic high, and redemption pressure surging in the private credit sector, bubble risk is rapidly accumulating.

On April 7, China Post Securities released a report stating that the rollout of downstream applications in 2026 will become the core observation window, and whether scalable commercialization is achieved is the key metric. If a bubble burst triggers a recession, the Federal Reserve will most likely accelerate rate cuts and restart quantitative easing. At that point, commodities, global risk assets, and the US dollar’s trajectory will undergo systematic re-evaluation.

AI Prosperity is Highly Concentrated on Hardware, Application Layer Lags in Realization

The US AI industry is globally leading. In 2025, 79% (about $159 billion) of global AI primary market financing will flow to US companies. By March 2026, OpenAI will have completed $122 billion in funding, with a valuation reaching $852 billion, maintaining its position as the world's top startup.

AI technology companies have become the absolute core driver of US stock market gains. The market value of the seven tech giants accounts for more than one-third of the S&P 500’s total market cap, contributing 53% of the annual return to the index in 2024, with Nvidia alone contributing 5.6 percentage points.

However, behind the prosperity lies a clear structural imbalance. The value of the industry chain is highly concentrated in hardware and cloud services, the foundational model layer suffers long-term losses, and at-scale commercial realization in the application layer is falling short of expectations.

Taking Ramp enterprise spending data as an example, by February 2026, the penetration rate of US companies subscribing to paid AI models and tools has reached 47.6%, but the gap between AI’s actual revenue generation and the scale of computing power investment continues to widen.

Debt Risk Emerges, Bubble is at the Brewing Stage

The expansion of capital expenditures by leading cloud vendors has surpassed the boundaries that endogenous cash flows can support. Alphabet, Amazon, Meta, Microsoft, and Oracle’s capital expenditures have accelerated again starting in 2024, with shareholder buybacks and dividends giving way to AI strategic expansion.

More noteworthy is the explosive growth in the AI-related debt financing market in 2025, with more than $110 billion in debt issued in just September and October. Oracle's total debt has exceeded $96 billion, and its credit default swap (CDS) prices have surged sharply.

Pressure is also mounting in the private credit sector. Software-related assets account for about 20.8% of the US private credit market. Enterprise-level AI tools launched by companies like Anthropic are raising concerns about the sustainability of traditional software business models and shaking confidence in the private credit market.

In Q1 2026, the top six private credit institutions submitted a combined $10.1 billion in redemption requests, doubling from the previous quarter, and some institutions may have initiated deferred redemption mechanisms.

Despite this, the current bubble has not yet reached an unsustainable tipping point. The net profit margins and revenue growth rates of current tech giants remain robust, fundamentally different from the concept-driven financing model of the 2000 internet bubble.

Risks in the private credit market are also relatively contained, lacking the large-scale derivative transmission chains seen during the subprime crisis, making it more likely a “storm in a teapot.”

If the Bubble Bursts, the Economy Will Face Multiple Shocks

If the AI bubble bursts, the shock will be transmitted to the real economy through at least three channels.

Wealth effect channel is the primary transmission chain. Federal Reserve data shows that in Q3 2025, US households and non-profit organizations directly and indirectly held $66.5 trillion in equity assets, accounting for 32.8% of total assets.

Whether personal investors or ordinary residents holding index products via 401(k) pension plans, all are highly exposed to the AI sector. Studies show that the negative effect of wealth shrinkage on consumption is about 50% greater than the positive effect of wealth growth, amplifying the downside pressure on the consumer market after the bubble bursts.

Capital expenditure cliff is the second path. AI-related capital expenditures have become the core driver of US economic growth, with research estimating their contribution to GDP at 1.1 percentage points in the first half of 2025. Referring to historical patterns when the fiber optic bubble burst around 2000, broadband operators' capital expenditures plummeted by 35% within a year.

If the current AI bubble burst results in a similar drop in capital expenditures, using the projected $450 billion in capital spending by hyperscale cloud vendors in 2026 as the base, the corresponding reduction would exceed $157.5 billion, directly weakening the growth momentum of actual US GDP.

Upstream industry chain resonance is the third transmission channel. AI investment fever is highly tied to the semiconductor and electricity sectors, with Nvidia’s 114% year-on-year revenue growth in 2024 as evidence.

Once investment cools, demand and price volatility for high-performance computing chips closely tied to AI will be even more drastic, and storage chip prices will also face correction risk. AI has also boosted labor productivity, causing employment growth to slow, further providing a macro basis for monetary easing.

Federal Reserve Response Framework: RMP is Just the Prelude, Recession Will Restart QE

Looking back, the Fed responded to the burst of the internet bubble in 2001, the 2008 subprime crisis, and the 2020 pandemic with large-scale asset purchases after policy rates approached zero.

Currently, Kevin Walsh, nominated as the next Fed Chair, was a Fed governor from 2006–2011 and is known for his “rate cuts + balance sheet reduction” policy stance.

But the collective decision-making mechanism of the FOMC limits the Chair’s personal influence. The current ON RRP balance remains near zero, and the SOFR-ON RRP spread is rising, indicating that money market liquidity has shifted from surplus to marginal tightness, leaving little room for further balance sheet reduction.

In December 2025, the FOMC unanimously approved the launch of the Reserve Management Purchases (RMP) program, marking a clear divergence from balance sheet reduction advocates.

Based on this, China Post Securities macro team distinguishes three scenarios:

If the AI bubble leads only to a sharp pullback in tech stocks and does not spill over into the credit market, the Fed may use forward guidance for expectation management, tolerate stock price adjustments to suppress core inflation;

If risk spills over into the credit market and credit spreads widen, the Fed will accelerate rate cuts and expand reserve management purchases;

If the economy enters recession and unemployment rises sharply, aggressive rate cuts to near zero and restart of large-scale QE will follow, with long-term Treasuries and agency MBS as the main tools.

The Global Transmission of a New QE: Commodities May Be the Biggest Beneficiary

If a new round of QE takes place, its effect will extend from the US to global asset markets.

In the US, systemic declines in long-term rates will improve financing conditions, support demand and employment, and interest-sensitive sectors such as real estate will benefit first. Globally, expanded dollar liquidity will improve overall capital market liquidity and lift risk asset pricing.

Looking at asset performance in the four previous QE rounds, commodities benefit most clearly: QE tends to weaken the dollar, directly raising nominal prices of dollar-denominated commodities; the low-rate environment increases commodities’ appeal in asset allocation; and economic recovery boosts real demand, with multiple factors resonating.

Currently, commodities are in a decades-low valuation range relative to US stocks. Historical data shows that such extreme valuation divergence often leads to a reversal in market structure.

Risk Warning and DisclaimerThe market involves risk, investment should be cautious. This article does not constitute personal investment advice and does not take into account individual users’ specific investment goals, financial condition, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investing based on this article is at your own risk. ```