Big Tech's AI spending has become the gravitational centre of the stock market, and on Friday that gravity reversed. For three years the promise of AI spending lifted technology shares on almost every announcement. This week investors added up the bill, looked at the cash leaving the building, and flinched.

By Blumefield | October 31, 2026

The Day the Market Did the Math

The selloff did not follow a disaster. The tech heavy Nasdaq closed 2.9 percent lower and the S&P 500 fell 1.8 percent, their worst single sessions since June, and the companies in the Nasdaq 100 shed more than a trillion dollars in combined market value in a matter of hours.

What makes the move unusual is that the largest technology firms had just reported strong quarters, with cloud revenue growing faster than it has in years. Investors were not reacting to weakness. They were reacting to the size of the checks these companies now plan to write, and to the dawning sense that the spending has outrun the cash coming in to fund it. A weak fourth quarter forecast from Intel added to the gloom, but the real story sat in the capital expenditure lines of the industry's healthiest balance sheets.

The volatility gauge known as the VIX drifted up to the low twenties, hardly panic territory. This was not a crash. It was a repricing, the moment a market that had treated every spending announcement as good news started asking a colder question about when, and whether, the money comes back.

A $828 Billion Bet With Four Cash-Flow Holes

The scale of the commitment is the heart of the matter. Taken together, Amazon, Alphabet, Microsoft, Meta and Elon Musk's SpaceX, which houses the xAI operation, are on course to spend as much as $828 billion on capital expenditure this year, according to their own earnings releases and calls. Amazon accounts for roughly $220 billion of that, Alphabet up to $205 billion, Microsoft about $170 billion, Meta up to $145 billion, and SpaceX up to $88 billion. Most of it is going toward data centres, chips and the power to run them.

Numbers that large would be easier to accept if the money were flowing out of surplus. It is not. Four of those five giants reported negative free cash flow in the most recent quarter, with Microsoft the sole exception. Meta, long a machine for generating cash, posted negative free cash flow of $2.1 billion, against a positive $10.6 billion in the same period a year earlier. Amazon has burned through more than $42 billion of free cash flow over the past year. For investors who bought these names precisely because they minted cash, the reversal is jarring.

The counterargument lives in the cloud results, and it is not trivial. Google's cloud business more than doubled last quarter, Microsoft's Azure grew 48 percent, and Amazon Web Services expanded by about 39 percent. Demand for computing is real and accelerating. The question the market asked on Friday is whether revenue growing at those rates can ever catch spending growing even faster, and how many years the gap has to stay open before patience runs out.

From Cash Machines to Borrowers

The clearest sign that something has shifted is where the money is now coming from. Companies that once had more cash than they knew what to do with are turning to debt and equity markets to keep the buildout going. Microsoft, which earned nearly $134 billion in profit in its latest fiscal year and generated almost $183 billion in cash from operations, still chose to sell $25 billion in bonds this week, its first bond offering in more than nine years. Meta moved earlier in the week to raise as much as $95 billion through a new share issue.

This is a different financial posture, and it changes the risk. When a company funds expansion from operating cash, a slowdown simply means less profit. When it funds expansion with borrowed money and fresh stock, a slowdown means interest payments and dilution landing on top of disappointing returns. The AI spending boom has quietly become a leveraged bet, and leverage is what turns a correction into something worse.

Smaller players show the strain more plainly. Oracle does not expect to return to positive free cash flow until the second half of its next fiscal year, though it projects the figure climbing above $30 billion by the fiscal year ending in May 2029. Anthropic's move toward a public listing this week drew attention to more than $265 billion in future minimum purchase commitments, a reminder that the obligations behind the AI spending surge stretch years into the future. Some banks, including Evercore and Bank of America, now see industry capital expenditure reaching $1 trillion in 2027. As one economist put it this week, when borrowing on this scale competes for capital, it can push up the cost of money for everyone else.

The Depreciation Question Nobody Can Settle

Beneath the headline numbers runs a quieter and more technical argument about accounting, and it may matter most of all. The optimistic case assumes the data centres and chips being bought today will earn their keep for a long time, which lets companies spread the cost across many years and keep reported profits healthy. Skeptics counter that artificial intelligence hardware ages quickly, that today's cutting edge accelerator is tomorrow's surplus, and that generous useful-life assumptions flatter the books by understating how fast these assets wear out.

Amazon's chief executive Andy Jassy addressed the tension directly on the company's earnings call, acknowledging that it would be more convenient if the opportunity unfolded slowly enough to spread the AI spending over 25 years, before adding that the world does not work that way and the window to build is now. It was an unusually candid admission that the timeline is compressed and the risk is real. The company is choosing speed over comfort, betting that the cost of moving too slowly exceeds the cost of moving too fast.

No one can prove either side yet, which is exactly why the market is uneasy. If the hardware lasts and demand compounds, the spending looks visionary. If the gear depreciates faster than the models can monetise it, a great deal of reported profit across the sector is borrowed from a future that may not arrive. The depreciation debate is not an academic footnote. It is the difference between the biggest technology investment in history and the biggest write-down.

What Breaks If the Spending Does Not Pay Off

The deeper worry is concentration. A handful of companies now carry an outsized share of the major indices, which means their capital decisions ripple through the retirement accounts of people who have never read an earnings call. One economist captured the fear this week with a blunt image, warning that if these technology giants sneeze, the rest of the market could catch pneumonia. When five firms are responsible for $828 billion of AI spending and a growing slice of index returns, their confidence becomes everyone's exposure.

For now the bull case is intact. Cloud demand is accelerating, the leading firms remain profitable on an operating basis, and even the heaviest spenders can point to real revenue behind the investment. Friday's selloff was a warning rather than a verdict, closer to a 10 to 15 percent correction than a collapse. Our earlier analysis of Big Tech's spending reckoning traced how quickly sentiment can turn once investors start counting cash instead of capacity.

What happens next depends on evidence that is still months away. Investors will watch free cash flow rather than revenue, debt issuance rather than announcements, and any sign that the cloud growth funding this buildout is slowing. The era in which AI spending was applauded on sight has ended. From here the market will want proof, and the companies betting hundreds of billions on the future will have to deliver it before the patience, and the financing, runs dry. For continuing coverage of the markets reshaping technology, follow Blumefield.