Investors are increasingly hedging against the default risks of tech giants. The cost of such protection is hitting multi-year records, and credit markets are diverging more sharply from stock prices amid the AI race. This is a classic signal that "smart money" is beginning to doubt the sustainability of the current rally.
Warning signs are coming from multiple fronts. These include the rising cost of default insurance, the divergence between credit and stocks, and the collapse of the most speculative securities.
Record cost of protection against big tech risks
The first and perhaps most striking signal is the surge in the cost of credit default swaps (CDS) for leading players in the sector. Five-year CDS spreads for Oracle, Amazon, Google, and Microsoft have risen to around 75 basis points. This is nearly the highest level in at least the last seven years.
Even excluding Oracle, the picture is not much milder: for the same group of companies, spreads have climbed to about 49 basis points—the highest since at least 2018. Both indicators have more than doubled since the start of 2025 and are now notably above the peaks of the 2022 bear market.
Analysts attribute this to record borrowing by big tech. Amazon, Google, Nvidia, Meta, Oracle, and SpaceX placed a record $182 billion in investment-grade bonds in 2026—a 1,300% increase year-over-year. Such debt levels, in my opinion, are beginning to worry creditors, who are demanding a higher risk premium.
Credit and stocks are diverging more sharply
The second signal is the growing divergence between credit markets and stock markets. Credit spreads for the largest data center operators have widened to 153.5 basis points over government bonds, compared to around 118 basis points in February. This is the highest since the launch of such a Goldman Sachs basket.
Meanwhile, the stocks of the same companies have barely risen. Over the same period, their basket has gained only about 3%, highlighting the growing gap between sentiment in stock and bond markets. Demand for the bonds themselves is also weakening: the coverage ratio for placements has fallen from about 5 in February to less than 2 in July.
The reason lies in the scale of future spending. By 2030, data center operators could spend around $5.5 trillion on AI, roughly half of which would be financed through the bond market. Thus, any misstep by the Fed on rates would leave the market with no safety margin. I believe this makes the sector extremely vulnerable to monetary policy tightening.
Speculative AI stocks are crashing
The third signal, experts say, is the collapse of the most volatile stocks. Goldman's basket of high-risk securities with strong momentum risks falling 23% in a month—the worst result in 17 years. The basket includes the market's most "pumped" stocks: Nvidia, Super Micro Computer, Palantir, D-Wave Quantum, and Navitas Semiconductor.
According to the service, these securities are particularly favored by retail investors. Their collapse hits private traders hardest. Analysts attribute the sell-off to growing doubts: investors are increasingly questioning whether data center operators are building excess capacity for AI.
All three signals paint a unified picture. Credit markets, which historically sense problems first, are already pricing in the risks of the AI race, while stocks are holding up for now. And if this tension continues to build, those betting on AI through the stock market will eventually feel the repercussions.
My analysis shows: the current situation resembles a classic "moment of truth" for an overheated sector. Credit markets have already voted with their feet, and investors focused on long-term growth should reconsider their risks in the technology sector.