The S&P 500 Index sits only about 3% below its most recent record high, yet 60% of the benchmark's constituents are mired in a bear market.

That's the striking picture painted by Goldman Sachs Group trader Bobby Molavi. He said the "strangeness" lies in the fact that both the tech giants whose capital expenditure is at the heart of the AI trade and the beneficiaries of that spending are seeing their share prices fall together.

"I don't understand how both can be punished at the same time, especially when the base case remains that AI will change everything," said Molavi, who heads execution services for Goldman in Europe, the Middle East and Africa, describing the phenomenon as a "stealth bear market" in a report.

Investors have been dumping AI-related stocks over concerns about the sustainability of massive investment, dealing a heavy blow to technology shares in recent weeks. On the other hand, most S&P 500 constituents have risen over the past month as money rotated away from popular trades toward defensive, under-owned sectors.

Alphabet Inc. last Thursday posted its biggest drop in 20 years, wiping out $500 billion in market value, yet the S&P 500 slipped just 0.3% as healthcare, real estate and consumer staples sectors filled the gap. Chipmakers enjoyed no such cushion.

The Philadelphia Semiconductor Index plunged 7% on Friday, falling below its 200-day moving average for the first time in a year, while its forward price-to-earnings ratio slid back to around 16 times, the low point of the AI cycle.

In Molavi's view, demand is not the problem. The latest earnings showed Google Cloud revenue doubled, Microsoft's Azure grew 48%, and Amazon's AWS rose 39%, setting a five-year record.

Market anxiety is centered on the cost of AI spending and the timeline for returns. The four major tech giants together generated $26 billion in free cash flow, only a fraction of what they once did. They are increasingly turning to debt, equity and cash to fund their AI buildouts.

Molavi wrote that Amazon's long-term debt soared from $66 billion to $173 billion in just nine months. The Goldman trader said last week was when the AI capital expenditure bill came due, with investors starting to focus on returns rather than ambition, and on cash flow rather than capex.