Tech Stocks Plunge, Safe-Haven U.S. Treasury Trades Return? 10-Year U.S. Treasury Yield Drops to Three-Week Low

The AI narrative has suffered a series of blows and is now reshaping market pricing frameworks in an unexpectedly counterintuitive fashion. Against the backdrop of hot inflation data, U.S. Treasury yields did not rise but instead plunged along with a technology stock rout.

The 10-year U.S. Treasury yield fell 10 basis points during Thursday's Asian trading session to 5.09%, hitting a three-week low. Meanwhile, the MSCI Asia Pacific Stock Index slumped more than 3%, the Korea Composite Stock Price Index plummeted 9% in a single day, while Samsung Electronics and SK Hynix both tumbled 11%. S&P 500 futures fell 0.4%, Nasdaq 100 futures dropped over 1%. Safe-haven funds poured into sovereign bond markets, the dollar weakened, and gold, after three consecutive days of decline, stabilized and rebounded to around $4,140/ounce.

This round of turmoil is not driven solely by inflation data—the cracks in the AI industry’s credit chain triggered by cloud computing company CoreWeave are rapidly transmitting to the equity markets. On Wednesday, CoreWeave’s stock price plunged more than 17%; its 2032 USD bond yield soared above 15%, five-year CDS widened by 135 basis points, and the implied five-year default probability reached 63%. According to Bloomberg, several analysts and investors are re-evaluating the overall financing risk exposure of the AI sector.

CoreWeave’s debt crisis marks an escalation of the AI technology stock sell-off from valuations to credit risk. The company's high-yield bond due 2032 was issued at par just four months ago, but has now dropped below 78 cents, with spreads jumping over 1,000 basis points.

D.A. Davidson analyst Gil Luria characterized CoreWeave as “subprime assets in the AI sector,” warning that its total borrowing cost is roughly 9% while the return on asset investment is only about 1%. If it cannot sustain its debt rollover, equity value risks falling to zero.

“The hotter-than-expected U.S. CPI data and the sell-off of CoreWeave bonds are prompting investors to re-evaluate the financing risk across the entire AI industry,” said Fibonacci Asset Management Global CEO Jung In Yun.

Latest data from AI model company Anthropic has also put pressure on market sentiment.

According to Bloomberg, citing sources familiar with the matter, the company posted $16.8 billion in Q3 revenue, but some analysts believe this figure missed expectations. “The disappointment with Anthropic’s headline numbers reflects deeper concerns over the capital burn needed to sustain this growth,” said Vantage Global Prime senior market analyst Hebe Chen. “In an already tight liquidity environment, the upcoming IPO will only accentuate this concern.”

Inflation beating expectations would normally push Treasury yields higher, but Wednesday’s market action flipped this conventional logic on its head.

Data showed that U.S. CPI year-on-year growth for September hit a five-month high and core inflation also beat expectations, causing traders to raise odds of a possible rate hike in October above 50% at one point. However, long-end Treasury yields, after briefly spiking, swiftly retraced all their gains and closed nearly flat.

The equity derivatives team at Bank of America led by Benjamin Bowler noted in a report that this is not a coincidence but a pattern—two identical market setups already occurred this month: On October 8, news broke that OpenAI’s annualized revenue guidance was nearly $20 billion lower than the market expected, and on that day, the Nasdaq slipped while Treasury yields likewise retreated sharply from morning highs.

BofA's analysis suggests that once the market begins to price in an AI slowdown risk, asset correlations will completely reverse. Bloomberg Markets Live strategist Mark Cranfield also commented, “This time, the risk could be different from the late-July episode—the risk of an AI bubble bursting could hurt economic growth, and therefore this bond safe-haven bid might last longer.”

The macro logic behind this is direct and clear: If AI-related capital expenditure slows, the sudden braking of compute infrastructure spending will severely impact supply chains and simultaneously lower expectations of demand for electricity, commodities, and high-end hardware—this is not just a reckoning with high valuations but also a blow to growth prospects and inflation premiums, ultimately benefitting bonds. This macro dynamic is highly reminiscent of the 2000 dot-com bubble burst, when the Nasdaq crashed and the 10-year Treasury yield tumbled from around 6.5% to the 4% range.

Tail Risk Accumulating: Underlying Currents in A Low Volatility Market

On the surface, U.S. stock volatility remains at very low levels—the S&P 500’s five-day realized volatility is only 7.6%, sitting in the lowest 3% of the past year. But BofA cautions that this calm is largely underpinned by market makers carrying massive Gamma positions, and signs suggest this cushion is weakening.

Macro strategist Michael Ball points out that the S&P 500 has slipped below its first key support area of 7,775–7,750. SpotGamma data shows the “volatility trigger point”—where market makers suppress volatility—sits right around 7,775; if the index breaks below 7,700, the positive Gamma cushion will erode rapidly, dip-buying and mean reversion strategies will fail entirely, and the market risks entering a negative feedback-driven downward spiral.

Meanwhile, with realized volatility in the spot market heavily suppressed by short-volatility, yield-seeking funds, buy-side institutions are aggressively snapping up VIX call options—their net long exposure is approaching the highs seen in August 2024, just before the “flash crash” triggered by the unwinding of volatility arbitrage trades.