A breakdown of the Magnificent Seven and AI hyperscalers raises concern for the stock market and the economy.
It’s time to consider and stress-test what was once thought as unthinkable: Can a sustained breakdown of the “Magnificent Seven” and AI hyperscalers crash both the U.S. stock market and the economy?
Now that the Magnificent Seven has definitively weakened and violated a key support level, the next question is whether the decline can be halted at its 200-day moving average. More ominously, the relative performance of these stocks is forming an inverted saucer top after violating a relative support level (bottom panel).
Chart showing the performance of Magnificent 7 (MAGS) ETF and its ratio against the S&P 500 from March 2025 to February 2026.
AI crash = economic slowdown
The market has started to become nervous over soaring hyperscaler capital expenditures.
Bloomberg Chief U.S. Economist Anna Wong outlined the risks of a hyperscaler blowup. Wong entered 2026 with a bullish outlook on the U.S. economy, and my view runs parallel with hers. I wrote in January: “For U.S. equity investors, early 2026 is a time to reap the benefits of Trump’s 2025 policies. Last year was tumultuous for policy, but policy uncertainty is fading, and the stimulative and pro-cyclical elements of the OBBB Act are becoming evident in early 2026. In addition, the Economic Surprise Index, which measures whether economic releases are beating or missing consensus expectations, has been steadily positive since mid-2025.”
Now Wong has turned more cautious. One that the AI bubble might be deflating is the loss of monopolistic-like pricing power by the leading companies. Wong said that such an event boosts adoption because of an erosion in pricing power, but it would not be positive for the stock prices of hyperscalers as the market starts to discount lower earnings growth potential. A recent Wall Street Journal report confirms the pricing pressure and lengthening sales cycle narrative.
Indeed, the market is getting more nervous over soaring hyperscaler capital expenditures. While the accelerated depreciation provisions of the “One Big Beautiful Bill” Act incentivize capex, all that spending eventually shows up as a depreciation expense in future earnings statements. For investors, the question becomes whether sales and margins can keep up with the acceleration in future depreciation — or does it become a headwind for earnings growth?
Hyperscaler free-cash-flow estimates are estimated to crater in the coming quarters.
Bar chart showing the capital expenditure of the top five hyperscalers from 2012 to 2026, totaling $602 billion, with Microsoft at $160 billion, Amazon at $155 billion, Alphabet at $125 billion, Meta at $120 billion and Oracle at $42 billion.
Already, hyperscaler free-cash-flow estimates are expected to crater in the coming quarters because of the frantic pace of anticipated capex. Where will the profits come from, especially when AI models mature and become commoditized, which drives down margins?
Stacked bar chart showing trailing 4-quarter free cash flow for ORCL, META, GOOGL, AMZN and MSFT from 2016 to 2028, with forecasts from 2026.
BCA Research estimated that, “to regain pre-capex-boom ROEs, hyperscalers need roughly +250bps in revenue growth or +100bps in margins.”
Bar chart showing return on equity for Hyperscalers and S&P 500 Information Technology from 2020-2027, with forecasted values highlighted.
Wong added that if the AI correction spills over into the credit markets and widens yield spreads, she projects a total headwind of 1.3–1.4% hit to GDP growth, which cuts the Bloomberg GDP growth estimate by about half.
The latest BoA Global Fund Manager Survey shows that AI hyperscaler capex is perceived to be the second-highest source of systemic credit risk (annotation is mine).
Bar chart showing private equity/private credit as the most likely source of a systemic credit event at 43%, and AI hyperscaler capex at 30% for February 2026.
Investment implications
What does this mean for stock prices? Jurrien Timmer at Fidelity pointed out that the Magnificent Seven companies represent such a large index weight “that if they should fall they could well take the S&P 500 (cap-weighted) index with it.” When megacap stock prices fall, the benchmark index tends to be weak.
The table below shows the math of a hyperscaler crash by explaining how much the S&P 493 would need to gain to offset Magnificent Seven losses under different conditions. For example, if the group were to fall 5%, the S&P 500 would need to rise by 2.7% to keep the index even.
Table showing how much the S&P 493 needs to rise to keep the S&P 500 flat if the Magnificent Seven basket declines by a certain percentage.
How likely is Magnificent Seven be compensated by gains in the rest of the market? As the chart below shows, the market has been undergoing a growth to value rotation in both the U.S. and international stocks since October. As growth stocks weakened during that period, value stocks rose and the S&P 500 showed little or no gains.
An analysis of global equity markets by region shows that between October 2025 and February 2026, when the S&P 500 went nowhere, other regions outperformed the MSCI All-Country World Index (ACWI), with the exception of China.
Chart comparing various indices and regions against ACWI (MSCI All Country World Index) from March 2023 to February 2026.
In other words, it’s entirely possible that the S&P 500 can see a sideways consolidation period when investors abandon megacap stocks.
There is a catch. There can’t be a recession during these periods of benign leadership rotation. The chart below of the Nadaq composite, which is a proxy for innovative companies with growth characteristics, shows that all recessions have led to bear markets, though not every bear market has signaled a recession.
FRED chart showing the Nasdaq Composite Index from 1971 to 2025, with shaded areas indicating U.S. recessions and black circles highlighting bear markets within those recessions.
A prolonged war in the Mideast could see an oil price surge that raises recession risk.
The projections of Bloomberg’s Wong indicate that, in the worst case, the U.S. economy is likely to experience a growth scare but no recession. The only realistic scenario under which a recession might occur is an economy weakened by AI equity correction and credit contagion that encounters a second macro shock, such as an oil-price spike. Past studies by economist James Hamilton suggested that oil price spikes have always resolved in recessions.
Mideast tensions are high, and a prolonged war could see oil prices surge, raising the risk of a recession. During these periods of geopolitical uncertainty, it’s worthwhile to recall Carl von Clausewitz’s view that “war is merely the continuation of politics by other means.” As U.S. forces gather in the Middle East, the White House hasn’t articulated a set of achievable and realistic objectives in Iran.
Is it regime change? That will be extremely difficult to achieve without boots on the ground. Destabilize the Iranian regime and trigger an uprising? The opposition is fragmented and lacks organization. Destroy Iran’s nuclear program or missile programs? A military strike could seriously degrade Iranian facilities, but they will be reconstituted in the absence of regime change. Target Iran’s Supreme Leader Ali Khamenei? Removing Khamenei in a decapitation strike could fragment the country or radicalize the Iranian regime. Going to war without a clear set of objectives raises the risk of a prolonged regional conflict that permanently elevates oil prices.
Given this uncertainty, investors have become nervous about the profit potential of the Magnificent Seven AI hyperscalers. I stress-tested a scenario of a pullback in equity and credit risk appetite sparked by AI anxiety. I found that while it may resolve in a stock market correction, a recession is unlikely. The only exception would be if the economy gets hit with a second shock, such as an oil price spike that weakens growth and sparks a recession — but everything has to go wrong first.
Comments