The AI Capex Paradox: Record Earnings Mask a Weakening Setup

Corporate earnings continue to provide a solid foundation for the S&P 500, but the market backdrop is becoming less favorable. AI-related capital spending is lifting near-term profits for suppliers while raising questions about future free cash flow for the companies funding the buildout. At the same time, Magnificent 7 leadership has become less reliable, equity supply is increasing, and the pre-midterm period could create a window for a broader sentiment reset. Therefore, the market presents a situation where strong earnings and correction risk can exist at the same time.

S&P 500 earnings remain impressive. In Q1 2026, index earnings grew roughly 28% year over year, far above the 13% growth analysts expected at the end of the quarter. Around 85% of companies beat earnings expectations, while index-level net margins reached a record 14.8%. AI-infrastructure names were an important part of this strength, contributing roughly half of the latest earnings growth.

However, the same AI investment cycle that supports current earnings also creates a more complicated setup for the market. Hyperscalers are spending aggressively on GPUs, servers, networking equipment, power, cooling systems, and data centers. For semiconductor and infrastructure suppliers, this spending becomes revenue and profit today. For the companies making the purchases, most of the spending is capitalized rather than immediately expensed under GAAP accounting rules, with the cost recognized gradually through depreciation.

That creates a favorable near-term accounting window for index earnings. The companies selling into the AI buildout record revenue now, while the companies funding the buildout recognize the cost over time. This does not mean the reported earnings are low quality, nor does it mean the AI cycle is unsustainable. But it does mean investors may become more sensitive to the gap between near-term EPS strength and future free cash flow pressure.

If price momentum weakens, the market could start questioning the durability of earnings, even while the current numbers still look strong. With mega-cap leadership fading, AI capital spending raising uncertainty around future returns, equity issuance increasing, and the market moving into a historically weaker pre-midterm window, the S&P 500 could experience a sentiment reset without necessarily entering a new bear market.

Magnificent 7 Leadership Has Become Less Dependable

One of the clearest signs of a changing market structure is the reversal in mega-cap leadership. From May 1 to late June, the Magnificent 7 declined about 9%, while the remaining S&P 493 stocks rose about 7%. This was not a broad risk-off move, since the rest of the index remained resilient. Instead, it suggests that the previous leadership group is being repriced.

The valuation signal is also important. According to Apollo’s data, the P/E premium investors are willing to pay for the Magnificent 7 has compressed to its lowest level in more than a decade.

Source: Apollo

The underlying reason is not simply weaker price performance. It is the market’s reassessment of free cash flow. Heavy AI-related capex makes parts of the group look less like high-margin, asset-light platform businesses and more like capital-intensive infrastructure operators. If investors begin to view these companies through that lens, the valuation multiple should be lower unless returns on invested capital become clearer.

This is why the Magnificent 7 label is becoming less useful. Investors are no longer treating the group as a single trade. They are separating the companies that directly monetize AI spending from those that must fund the capex cycle.

For hyperscalers such as MSFT, AMZN, GOOGL, and META, free cash flow is expected to fall sharply this year and next. The payoff from today’s spending is expected to arrive only after 2028, when the group could generate more than $700 billion in free cash flow. But the AI landscape is changing quickly, which makes projections that far into the future less reliable. The market is unlikely to keep paying a premium for the group unless the path from spending to free cash flow becomes more visible.

Source: BCA Research

Rising Equity Supply Is Becoming a Headwind

Another pressure point is the increase in equity supply. The headline amount may still look manageable, but the timing and concentration of issuance make it more relevant for U.S. equities.

The driver is a mix of cyclical reopening in capital markets and the enormous funding needs of the AI investment cycle. After several years of muted net issuance following the 2021 boom and the Fed’s aggressive hiking cycle, companies are returning to public markets through IPOs and secondary offerings.

Alphabet announced an $80 billion equity raise to fund AI infrastructure. SpaceX completed the largest IPO on record, raising about $85.7 billion. OpenAI was reported to be considering a delay of its listing into 2027, while Anthropic’s filing was pointed toward the autumn.

At the market level, the supply number is still relatively small in 2026, with JPMorgan estimating around $200 billion. But that figure is expected to jump by more than 500% in 2027. The total includes IPOs, secondary offerings, and other share sales, as hyperscalers are expected to raise hundreds of billions of dollars through secondary offerings to finance AI spending plans.

The risk is not just the size of the issuance. It is the motive and timing. Public investors are being asked to fund a capital-intensive AI infrastructure buildout at the same time that mega-cap valuations are being repriced lower. Historically, elevated deal activity has often appeared near periods of peak equity prices. That does not guarantee a market top, but it does suggest that rising supply could make further multiple expansion more difficult.

The Midterm Calendar Adds Timing Risk

The political calendar adds another potential headwind. Midterm years have historically been the weakest part of the four-year election cycle. The two most recent midterm years, 2022 and 2018, produced S&P 500 returns of -18% and -4.4%, respectively. These were also the two weakest calendar years for the index since the 2008 financial crisis.

S&P 500 Total Returns; Source: Slickcharts

The more useful message from history is not that midterm years must deliver negative returns. It is that markets often struggle before election uncertainty is resolved. Since 1970, the S&P 500 has historically weakened ahead of midterm elections, then started to rally roughly 22 trading days before the vote, according to BlackRock. After the election, the average six-month return was 14.1%.

Source: BlackRock

This makes the midterm period a timing risk rather than a standalone bearish catalyst. Markets tend to have more difficulty when policy uncertainty is still unresolved, then recover once investors can better assess potential changes in regulation, taxes, and fiscal policy. A correction into November would therefore fit the historical pattern, especially if sentiment is already stretched and leadership is weakening.

If earnings remain intact and election uncertainty clears, that kind of pullback could ultimately become a reset rather than the beginning of a deeper downturn.

Conclusion

The current market setup is not defined by collapsing fundamentals. Earnings are strong, margins are high, and AI infrastructure continues to support index-level profit growth. The problem is that the market is becoming more sensitive to what comes after the earnings surge.

AI capex is creating near-term revenue for suppliers, but it is also pressuring future free cash flow for the companies funding the buildout. The Magnificent 7 are no longer moving as a single leadership trade. Equity supply is rising as companies return to market and AI funding needs increase. The midterm calendar adds another layer of uncertainty during a historically weaker part of the cycle.

Taken together, these factors suggest that the S&P 500 could face a correction or sentiment reset even if the broader earnings story remains intact. The risk is less about an immediate profit recession and more about valuation discipline, capital intensity, and investor confidence in the future return on AI spending.