The US stocks tech sector has fallen recently, and the Philadelphia Semiconductor Index has already slipped into a technical bear market.

During the drop, all kinds of bearish charts have started popping up—like the new one below:
Latest Bank of America forecast: the cash flows held by AI giants (Microsoft $MSFT, Google $GOOG , Amazon $AMZN, Meta $META ) will be $191 billion in 2025, $19 billion in 2026, and then turn to negative $26 billion by 2027.
Over two years, the cash flow gradually gets burned through, indicating that capital expenditures for AI infrastructure have grown too fast—spending on GPUs, data centers, and power has been much faster than the pace at which money is being made.
This chart also has an even scarier detail:
Each line represents a forecast made for a different year—the newer the line, the lower it goes. Over the past four years, the market has been overestimating these companies’ cash flows, adjusting them once per year, but each time they get revised downward.
Of course, when the market is down, everything looks bearish. This chart can be understood as “AI money burning can’t keep going”—several major AI companies are facing trouble.
If next year the AI revenue side still hasn’t broken out into a profitable business model, then in theory these companies would have to hit the brakes and cut spending, and their stock prices could also face revisions.