The AI Bill Comes Due: Why Investors Are Now Grading Big Tech on Cash

For three years the market rewarded big technology companies for spending on artificial intelligence. Every increase in capital spending was read as confidence. That changed this week. Investors made clear they now care less about how much the giants are spending and more about whether the cash coming in can cover it.

The week in numbers

• Alphabet fell about 12% on Thursday, October 29, its worst single day on record. Roughly half a trillion dollars of market value disappeared. The company spent $51.4 billion on capital projects in the quarter, mostly data centres and AI chips. Free cash flow came in at about −$9 billion, its second straight negative quarter. It bought back no stock. Instead it raised money through debt and new share sales, and its finance chief said spending would rise significantly again in 2027. This happened even though Google Cloud revenue roughly doubled from a year earlier.
• Amazon reported $77.3 billion of property and equipment purchases in the quarter. Its free cash flow over the trailing twelve months swung to an outflow of $42.4 billion, from a $14.8 billion inflow a year earlier. It reaffirmed roughly $220 billion of capital spending for 2026 and said 2027 would be “meaningfully more.” The shares fell.
• Meta had its credit outlook cut to “negative” from “stable” by S&P Global late Friday. The rating itself (AA−) was unchanged. S&P cited negative cash flow and uncertainty about spending, and warned that leverage could breach its downgrade threshold as early as 2027.
• Microsoft beat earnings estimates but still fell after saying 2026 spending would be well above 2025. It remains cash-flow positive and sold $25 billion of bonds during the week.

Taken together, tech stocks lost more than a trillion dollars of value over the week’s final sessions. The iShares Semiconductor ETF dropped 9.1% on Friday alone.

Why free cash flow is suddenly the number that matters

Earnings and cash are not the same thing. When a company buys a $1 billion data centre, accounting doesn’t charge that $1 billion against profit right away. It spreads the cost over years as depreciation. So reported earnings can look healthy while the bank balance drains.

Free cash flow is cash from operations minus capital spending. It removes that cushion. It’s the money actually left over to pay dividends, buy back shares, pay down debt, or survive a downturn. When it turns negative, the company has to fund the gap from its cash pile, by borrowing, or by issuing new shares.

That changes the investment case in three ways:

1. Shareholders stop getting paid back. Buybacks have been a major support for big-tech share prices. A company issuing shares instead of retiring them dilutes existing owners. (See stock splits and buybacks.)
2. Balance sheets start to matter. These companies have been treated almost like bonds: ultra-safe, cash-rich and low-debt. S&P’s warning on Meta says that status is not guaranteed.
3. Returns are pushed further out. Money spent today has to earn a return later. The further away and less certain that return, the less investors will pay for it now. This is especially true with interest rates rising rather than falling, as the central banks showed this week.

Is this a bubble bursting?

Not necessarily, and it’s worth separating two different claims.

The demand for AI computing appears real. Cloud businesses are growing fast, and Alphabet’s cloud revenue doubling is not the sign of a fad. The economics are the open question. The spending is front-loaded and enormous, while the revenue arrives gradually and its eventual profit margins are unknown. Railways, telecom fibre and the early internet all saw genuine, transformative demand and investors who lost money because too much capacity was built too fast, with too much borrowed money.

There’s also a reasonable optimistic reading. Economist Ed Yardeni argued this week that funding constraints are “healthy for the durability of the AI boom.” In other words, a market that forces discipline now makes a disorderly bust later less likely.

What it means for ordinary investors

• Check your concentration. The largest tech companies make up a big share of broad US index funds. An S&P 500 or Nasdaq-100 fund is, in practice, a substantial bet on exactly these spending decisions. It’s worth knowing how much of your portfolio rides on them.
• Look past “beat or miss.” Revenue and earnings beats did not protect these stocks this week. Capital spending, free cash flow and financing activity are now in the headline. (Our guide on how to read an earnings report covers where to find them.)
• Watch the suppliers too. One company’s capital spending is another’s revenue. Chipmakers and equipment makers benefit from the spending the market is now punishing. The 9% Friday drop in semiconductors shows investors are asking how long that spending lasts.

What to watch next

This week brings Apple (Monday), AMD and Super Micro (Tuesday) and Qualcomm and Arm (Wednesday). Apple is the interesting contrast. It has spent far less on AI infrastructure and is the best-performing of the largest tech stocks this year, up about 28%. If the market keeps rewarding restraint, that gap may widen.