Huge spending on data centers is squeezing cash flow
These head-spinning adjustments occur just as Big Tech’s monumental spending on data centers incinerates once-prodigious cash flow. From 2023 to 2025, for example, Alphabet generated $215 billion of free cash flow; it was negative $5.9 billion in the second quarter. Amazon is on track to burn through $13 billion this year and $28 billion in 2027, according to estimates compiled by Visible Alpha.
Big Tech’s AI accounting gains are masking cash burn
Chatbots are solving mathematical conundrums even as they create financial ones. Consider the following: How can SpaceX SPCX.O and Anthropic, two artificial-intelligence lab operators that lose money, account for half of the $280 billion rise in net profit at S&P 500 .SPX companies so far this quarter? The answer is largely that their respective backers, Alphabet GOOGL.O and Amazon.com AMZN.O, have booked enormous paper gains as the pair’s valuations ballooned. This , when combined with dwindling cash and deeper corporate entanglements, makes the financial logic hard to follow.
The AI boom is flattering an already impressive three-month stretch of U.S. earnings. More than half the companies in the blue-chip S&P 500 Index have reported a collective 47% rise in their bottom lines from a year ago, according to FactSet. Included in the figure, however, is Alphabet’s $98 billion of “other income,” which traces back to a $900 million investment in Elon Musk’s SpaceX more than a decade ago. The rocket-maker’s initial public offering last month imputed a $2 trillion market capitalization, forcing Alphabet to revalue its stake under generally accepted accounting principles. Amazon similarly recalculated the price of its non-voting stock in Anthropic.
Commercial ties between AI backers and startups keep deepening
Further confusion stems from substantial commercial dealings between these companies. Right before its IPO, SpaceX said Google would pay it $920 million monthly for computing capacity, including 110,000 Nvidia NVDA.O chips. Five years earlier, Google agreed to provide cloud services to SpaceX. Anthropic, meanwhile, has agreed to spend more than $100 billion over a decade on Amazon’s AWS technologies and the e-commerce giant is making $20 billion of financing available to the Claude developer in exchange for more equity.
The intricacies only keep growing. Valuations in the machine-learning industry are swinging wildly while giant software companies grapple with their newfound capital intensity. An opaque web of cooperation and competition between the two groups further leaves investors contending with a madcap mix of complexity and hype. It will take more than a simple prompt to solve this costly mess.