Investors are losing patience with the AI spending boom. Big Tech isn’t slowing down at all. According to TechCrunch AI, Alphabet, Amazon, Meta, and Microsoft have laid out plans in the past two weeks to spend more than $740 billion on capital expenditures this year. Wall Street reads most of that as data centers and chips.
A year ago, spending at that scale lifted stock prices. Now it’s doing the opposite. On Friday the Nasdaq fell 2.9%, Nvidia dropped 4%, Broadcom lost more than 5%, and Qualcomm sank 11%. Alphabet fell again after Thursday’s 12% plunge, its worst day on record, which erased $500 billion in market value.
📉 What’s changed: cash flow went negative
The worry isn’t really about demand. Cloud revenue grew at Google, Amazon, and Microsoft. The problem is what it costs to meet that demand, and the fact that companies once swimming in free cash are now burning it.
Here’s how each one looks, as reported by TechCrunch AI:
- Alphabet: Negative free cash flow of $9.1 billion in Q3, its second straight quarter in the red. Quarterly capex hit $51.4 billion. Full-year spending could reach $205 billion and is expected to rise again in 2027.
- Amazon: Negative free cash flow of $42.4 billion over the past 12 months, compared with $14.8 billion positive a year earlier. Capex guidance is $220 billion. AWS grew 39%, and the stock still fell 7%.
- Meta: Negative free cash flow of $2.1 billion in Q3, with capex guidance up to $145 billion. Ad revenue grew 26%, but shares took a 10% hit anyway.
- Microsoft: The exception, with $18.6 billion in positive free cash flow. Capex came in at $45.2 billion, short of the $50 billion it had projected.
Microsoft says the shortfall wasn’t a deliberate pullback. It blamed delays in capacity from third-party providers. Even so, shares of smaller companies that rent out AI compute sank on the news. That reaction tells you how fragile sentiment is right now.
🔍 Google’s tightrope
What stands out most is Alphabet’s position. AI Mode, the chatbot built directly into Google Search, is a hit. That hit is eating into the business that pays for everything else.
Search advertising makes up roughly half of Alphabet’s revenue, and its growth slowed to 12%, the weakest pace in five quarters. So Google is spending record sums to build technology that cannibalizes its own cash machine. It may not have a choice. If Google doesn’t disrupt Search, someone else will. But investors are now pricing in the cost of that reality.
Meta faces a smaller version of the same problem. Mark Zuckerberg has admitted that Muse, the company’s free personal AI agent, is “very compute-intensive to run.” Free products that cost a lot to serve make for a hard sell on earnings calls.
💸 The debt angle matters more than it looks
This buildout isn’t funded by profits alone. According to S&P Global Market Intelligence, cited by TechCrunch AI, Microsoft, Alphabet, Amazon, Meta, and Oracle raised a combined $326 billion in debt in 2026. That’s triple last year’s total.
This is significant because it changes the risk profile of the whole sector. When AI spending came out of spare cash, a bad bet meant lower margins. When it comes from borrowed money, a bad bet means leverage, and a longer road to recovery if returns disappoint.
🏛️ The ripple effect hits Anthropic
The timing is rough for Anthropic, which filed to go public earlier this week. Its growth depends on securing huge amounts of compute. The Financial Times reported that Blackstone warned investors the company might not get an investment-grade rating after its IPO. That could leave Anthropic leaning harder on hyperscalers and third parties to fund its infrastructure plans, which ties it even closer to the same companies investors are now questioning.
🧭 What practitioners and businesses should do
None of this means AI demand is fading. It means the price of supplying it is under real scrutiny for the first time. A few practical moves:
- Don’t assume compute stays cheap. Providers burning cash will eventually push for better margins. Budget for price increases on API and cloud AI services.
- Diversify vendors. Capacity delays like the ones Microsoft described can hit anyone. Avoid single-provider dependence for critical workloads.
- Track unit economics, not hype. If a free AI feature is a money pit for Meta, it can be one for you too. Measure cost per query before you scale.
- Watch the Anthropic IPO. How the market prices it will be a clear signal of investor appetite for compute-hungry AI labs.
The AI boom isn’t over. But the era of spending without scrutiny probably is. Over the next few quarters, the winners will be the companies that can show returns, not just ambition. Full details are available in the original TechCrunch AI report.