AI Companies Are Earning More. Why Are Their Stocks Falling?
Earnings are growing, but the price the market will pay has fallen. Why investors should separate AI industry growth from AI shareholder returns.
AI correction: when does the rebound start? Part 1 of 4
Earnings are good, but they did not clear expectations by much
According to FactSet, 24% of S&P 500 companies had reported third-quarter 2026 results. Of those, 83% beat analysts' earnings-per-share (EPS) estimates. That is above the five-year average of 78%.
The size of the beat is a different story. Earnings topped estimates by an average of 2.1%, well short of the five-year average of 7.5%. Many companies cleared the bar, but not by much. The 29.5% EPS growth rate and 12.7% revenue growth rate are also not final. They combine reported results with estimates for companies yet to report.
| Metric | Figure |
|---|---|
| Share of companies reported | 24% |
| Beat EPS estimates | 83% |
| Average EPS surprise | +2.1% |
| Q3 EPS growth | +29.5% |
Why stocks are weak despite high growth
A stock price reflects not only current earnings but also future profits and the multiple applied to them. If the return investors demand rises, the price-to-earnings ratio can fall even as expected EPS grows.
The S&P 500's 12-month forward P/E is 18.7. Inverted, that gives an earnings yield of about 5.35%. The 10-year Treasury yield is about 4.99%, so the gap is only 0.36 percentage point. This is a simple earnings-yield comparison, not a rigorous equity risk premium. Still, it helps show how much valuation pressure the market carries.
With long-term rates high, AI companies face a dilemma. If they raise capital expenditure, investors worry about slowing cash flow. If they cut it, investors worry about slowing growth. The key question is no longer the absolute size of the spending. It is how quickly companies recover cash relative to what they invest.
| Metric | Figure |
|---|---|
| Forward P/E | 18.7x |
| Earnings yield | 5.35% |
| 10-year yield | 4.99% |
| Simple gap | 0.36 pt |
AI industry revenue is not the same as shareholder profit
A GPU sale is revenue for the chip supplier but capital spending for the cloud company that buys it. Cloud usage fees are revenue for the cloud provider and a cost for the AI software company. The value chain can sustain its returns on investment only if the money that end businesses and consumers pay for AI keeps growing.
A memory shortage can raise memory makers' pricing power, but it also raises costs for data center operators. Falling inference costs can lead to more AI usage. But if competition pushes service prices down faster, revenue or profit may come in below expectations.
For that reason, technological progress, industry revenue, an individual company's free cash flow (FCF) and per-share value are separate questions in AI investing.
This content is information for investment decisions and is not a recommendation to buy or sell any financial product.
Insight Times Editorial Desk
