BlackRock's Rosenberg: Oil-Treasury Correlation Broke Down in October, AI Trade Becomes Market's Top Risk

Jeffrey Rosenberg, senior portfolio manager of BlackRock's Systematic Alternatives ETF, said, "Oil and Treasury yields have always had an astonishing correlation, but that stopped in October."

Over the same period, AI stocks fell. Rosenberg said that these changes in market behavior in October—the breaking of the ultra-high correlation between oil and Treasuries and the sharp correction in AI stocks—show that in the market's view, the number one risk right now is the AI trade, followed closely by geopolitical risk.

Last week, Federal Reserve Chair Kevin Warsh highlighted the relationship between oil and Treasury yields at the Federal Open Market Committee (FOMC) press conference. He said that the continued rise in oil futures prices is a factor the Fed must consider in fighting inflation. But Rosenberg said what Warsh did not discuss was "the most important point to watch," namely that "this year the correlation between U.S. Treasuries and oil has been 80%, which is an outsized share compared with the normal correlation. So we basically had ample reason to abandon trading oil, but in October this ultra-high correlation came to an abrupt halt."

Over the past month, the AI trade has undergone a pullback, and the Nasdaq has even suffered a sharp selloff. Earlier, Google suffered the largest one-day drop in its history, wiping out $5 trillion in market value. Rosenberg said Treasury yields fell because the market equated weakness in the AI trade with broader weakness in the U.S. economy. The latest U.S. gross domestic product (GDP) data also support this market speculation.

"Whenever and wherever, AI is everything. Economic growth on the investment side and the wealth effect on the consumption side both revolve around AI. Third-quarter GDP came in as a blowout number, related to AI investment. Especially considering that U.S. real income has been declining, while the increase in equipment investment and consumption was far above expectations, it is clear that both were driven by the wealth effect brought about by AI investment, with the latter especially notable," he analyzed.

In the third quarter, U.S. business investment grew 12.7%, the largest increase since 2011 (excluding the pandemic period). Of that, information processing equipment rose 37%, and industrial equipment investment grew nearly 25%. Although not directly an AI trade, this does partly reflect spending closely related to data centers.

However, Warsh's assessment of the economy at the press conference differed slightly from the market's. He said, "The strength of the economy does not depend entirely on this new wave of technology (AI). The data from the last meeting to this meeting, especially the capital expenditure story, shows that capital expenditure is not only strong but also broadening."

Admittedly, market breadth—the indicator used to measure how many stocks in the S&P 500 show positive trading momentum—improved significantly in October after falling to a 2026 low at the end of September, and defensive sectors such as health care and food also rebounded. But Rosenberg said the improvement in U.S. equity market breadth in October shows that any "volatility" in the AI trade will be viewed as volatility in the U.S. growth outlook and will have an inflation-dampening effect. And in a disinflationary environment, "there is only one asset to hedge with, and that is long-term Treasuries, which the market previously hated. This can be said to be the truly important lesson the market learned in October."

That said, Rosenberg remains doubtful about the market dynamics of Treasuries once again becoming a safe hedge asset. He said Treasury behavior has not returned to normal; the normal situation should be bonds hedging stocks. "Inflation is too high, which limits the Fed's ability to provide accommodative policy, and we have not escaped this situation. What we learned in October is that if Treasury yields are high enough (Treasuries have fallen enough), and at the same time if AI stocks move contrary to macroeconomic and inflation trends, Treasuries may strengthen again as they did in October and briefly serve as a hedge." But he added that the AI boom as a whole will last longer, and with the U.S. still facing broader and severe fiscal challenges, Treasury behavior will remain challenged.

At the same time, AI is not only affecting the stock market but increasingly affecting the bond market as well. The AI trade has become a very large part of the bond market. Over the past week, this was strongly reflected between Google's earnings decline and Microsoft's return to the public bond market. Rosenberg said, "This is a surprise for the bond market; this kind of heat has not returned for nearly a decade. As tech giants' capital expenditure needs begin to overwhelm their free cash flow more frequently, the public bond market is absorbing expectations of a large amount of AI debt issuance. AI companies are huge marginal issuers, and more and more money keeps pouring in again and again."

Alyssa Irving, a fixed income fund manager at Wellington Investment Management, told Yicai that AI is not only driving the stock market but is also increasingly important to the credit market. AI has now penetrated almost every area of the credit market, with debt issuance related to data centers growing exponentially.

"The massive infrastructure spending required by AI may still be in its early stages, but its impact on the credit market is becoming increasingly clear: higher capital intensity, increased credit supply, and greater differentiation between sectors and issuers," she said. "For investors, the challenge is to distinguish which companies can turn AI-related investment into sustainable returns and which cannot. As AI investment increasingly becomes an important driver of market performance, prudent analysis of capital allocation, financing and execution risk may be crucial."

She added that while AI infrastructure spending is generally supportive of credit demand, its investment impact varies significantly across sectors: investment-grade credit investors are focused on the balance between growth and rising capital intensity; high-yield credit investors see a growing divergence between AI beneficiaries and issuers hit by AI; and securitized credit investors focus on deal structure and repayment assumptions.

"As a result, investment opportunities in the credit market are becoming more differentiated, with the main theme being market divergence rather than a broad rise in risk. Companies directly tied to AI infrastructure buildout, such as hyperscalers, semiconductor companies and infrastructure providers, are generally benefiting from stronger growth, and the fundamentals of some have improved. However, the picture is more complicated for companies making large-scale investments simply to stay competitive. If the path to monetizing investment is uncertain, high capital expenditure could pressure cash flow and leverage metrics. This divergence could further intensify differentiation within the investment-grade credit market and further highlight the importance of issuer-level analysis and prudent capital allocation," she said.

Rosenberg also stressed that the $40 trillion to $60 trillion global bond market can absorb the expected $4 trillion of AI debt over the next four years, but "the market's capacity is not unlimited." AI companies' huge debt issuance needs will also lead to falling prices and wider spreads.

Rosenberg said more hyperscale debt being pushed into the public bond market is a good thing, because it means bond prices and credit default swap (CDS) spreads will be more transparent. Although Microsoft's AI debt CDS spread is currently near a five-year high of about 55, it remains far from Oracle's CDS spread of nearly 300 by comparison. This offers investors another trading opportunity: given the U.S. inflation environment and fiscal challenges, they may not want to buy bonds directly, but they can use CDS to hedge the AI trade. In bank loans and private credit, investors cannot trade CDS, but once these AI companies enter the public bond market, investors gain this additional trading tool.

He said, "Given the massive AI investment boom, the AI trade will not move in one direction, but will constantly escalate and reset. As the AI boom continues, the change in the AI bond market is a positive part. Investors can find a balance between AI pessimism and optimism."