After the AI Spending Boom, Who Gets the Money Back?
AI demand is still strong, but the 2027 market may favor companies that recover their spending best, not those that spend the most. A structural outlook for the next 3, 6 and 12 months.
Something unusual is happening in the US stock market. Corporate earnings are growing fast, yet AI-related stocks are pulling back. AI demand has not vanished. Demand for semiconductors and data center power remains strong.
Wall Street's question has changed. The issue is no longer how much companies invest in AI, but how much cash that investment brings in.
That standard is likely to separate the winners in the 2027 US market.
Earnings are strong. Why is the market uneasy?
According to FactSet, 83% of S&P 500 companies that had reported third-quarter results as of Oct. 23 beat consensus earnings per share (EPS) estimates. Expected third-quarter earnings growth for the whole index is 29.5% from a year earlier.
But the average margin of the beat was 2.1%, well below the five-year average of 7.5%. Many companies are reporting good results, but fewer are overwhelming expectations.
| Metric | Q3 2026 | Note |
|---|---|---|
| S&P 500 expected earnings growth | +29.5% | Year over year, mix of actuals and estimates |
| Average EPS beat | +2.1% | Five-year average +7.5% |
Source: FactSet
Valuation is not clearly expensive, either. The S&P 500's forward 12-month price-to-earnings ratio is 18.7, below the five-year average of 19.8. But that figure uses future earnings that have not yet been realized as its denominator. If earnings forecasts are cut, the P/E can rise even at today's share prices.
The problem for the market as a whole and the problem for some AI growth stocks need to be treated separately. The fact that the AI industry is growing is a different matter from whether AI stocks will deliver high returns.
Meta and Tesla show two realities
| Meta | Tesla | |
|---|---|---|
| Q3 2026 revenue | $64.601 billion | $28.05 billion |
| Operating income | $12.988 billion | $718 million |
| Free cash flow | -$2.12 billion | +$1.531 billion |
Source: Meta earnings release, Tesla SEC quarterly filing. Accounting structures differ, so comparing absolute amounts has limits.
Meta's revenue rose 26% from a year earlier. Both ad impressions and ad prices increased, a sign that AI is improving the efficiency of its existing advertising business.
But capital expenditure reached $34.91 billion for the quarter, and free cash flow was -$2.12 billion. The drop in operating income cannot be blamed entirely on AI spending, however. It includes a one-time legal settlement charge of $9.89 billion. Excluding it, operating income was about $22.88 billion, higher than a year earlier.
Tesla showed the opposite picture. Free cash flow improved to $1.531 billion, but operating income fell sharply from a year earlier. The key question for Tesla is how long it can keep funding autonomous driving, robotics and AI computing with cash from its auto business.
The two companies share one point: long-term technological potential alone no longer explains today's investment bill. Still, one quarter of cash flow is not enough to pin down future returns on investment.
AI infrastructure: tight supply does not guarantee returns
Power infrastructure company Bloom Energy posted third-quarter revenue of $1.236 billion, up 138.1% from a year earlier. It shows how strong data center power demand is.
But when revenue surges, production capacity, inventory and working capital grow too. Even as orders rise, it can take time for cash to actually arrive.
The same goes for semiconductors. Nvidia has to prove its pricing power on GPUs, TSMC its utilization at leading-edge nodes, and Micron its HBM pricing and supply capacity. Even among AI beneficiaries, gross margins and capital spending burdens differ.
The AI opportunity is not disappearing. It is a time to distinguish companies that have secured demand from those that can supply that demand at high margins.
SpaceX and Tesla: price and terms matter more than the merger
SpaceX listed on the Nasdaq in June. In October it formally disclosed that a special committee had been formed to evaluate a possible merger with Tesla. The merger is not yet confirmed.
The two companies could potentially cooperate in AI computing, autonomous driving, robotics, satellite communications and energy infrastructure.
But combining businesses does not by itself raise shareholder value. Each company's valuation, the exchange ratio, additional capital spending and governance all matter.
In particular, it is important to see how much economic stake existing Tesla shareholders and SpaceX shareholders would each hold after a merger. Technical synergy and the economic benefit that reaches shareholders are not the same question.
High oil prices and rates: another variable for AI returns
September CPI rose 3.7% from a year earlier, and the producer price index (PPI) rose 6.3%. The Fed raised its benchmark rate to 3.75%-4.00% in September.
High oil prices weigh on consumption and corporate earnings, and through inflation they can delay rate cuts. High rates are especially hard on growth stocks, where expectations of future earnings make up a large part of today's price.
That is why AI companies can see their shares fall even when they report good results: the discount rate rises.
What to watch over the next 12 months
- 1 to 3 months: earnings versus rates. Big Tech cloud growth and spending plans, the Fed's policy direction, oil prices and inflation.
- 6 months: profitability gaps among AI companies. Semiconductor pricing power, data center utilization, and paying-customer growth and cash flow at AI software firms.
- 12 months: the real ability to recover AI spending. Earnings contribution from cloud and AI services, commercialization of autonomous driving, and returns and debt burdens in power infrastructure.
The 2027 winners may not be the companies that invested the most in AI. The companies that recover their investment most efficiently are likely to change how the market values them.
Insight Times Editorial Desk
