Real Rates, Oil and the AI Build-Out: Insights from Jurrien Timmer - September 21, 2026

Oil, bonds and stocks as one trade

The oil disruption that began in late February has not faded. Six months later, the market is still feeling its effects, with Brent crude around 104 and forward prices near 84, which Jurrien said suggests supply remains very tight. That pressure extends beyond the energy market. Oil and the S&P 500 remain almost perfectly negatively correlated at minus 59%, while stocks and bonds remain positively correlated. In practical terms, Jurrien said it behaves like a single trade: as oil rises, bond yields rise and, because bonds are moving with stocks, equities tend to come under pressure. He argued that the Federal Reserve has little choice but to take those inflation pressures into account, even though it cannot directly influence oil prices. Its primary policy tool remains short-term interest rates, which he described as a blunt instrument for addressing a supply-driven shock. Higher rates are also being felt unevenly across the economy. While some large companies remain relatively insensitive to financing costs, prospective homebuyers continue to face mortgage rates around 7%, adding to affordability challenges.

A rare moment for real yields

Another development Jurrien highlighted is the recent move in bond yields. The 10-year Treasury yield reached 5.04% before easing back to 4.96%, completing a technical breakout he said he had been monitoring for some time. The Federal Reserve also raised rates at its most recent meeting, effectively reversing one of the insurance rate cuts introduced a year earlier amid concerns that AI-related disruption could weaken the labour market. So far, he noted, those concerns have not materialized. More significant, in his view, is what has happened to real yields. The real 10-year Treasury yield, as measured through TIPS, reached 2.67%, moving above the Congressional Budget Office's estimate of five-year potential real GDP growth. That crossover matters because debt tends to be more sustainable when growth exceeds funding costs. Jurrien noted that the last periods when real yields moved above estimates of real growth included the global financial crisis, the dot-com era and the 1994 rate shock. He cautioned that a single data point does not establish a trend. He also noted that current growth estimates may not yet reflect any future productivity gains from AI. Even so, he described the development as noteworthy and worth monitoring.

The AI capital-raising question

While interest rates remain in focus, the AI investment cycle continues to reshape capital markets.

Demand for computing power remains extremely strong, and hyperscalers continue to raise capital to fund AI investment. Jurrien said corporate debt issuance and net equity issuance have combined for roughly US$3.2 trillion year over year, with net equity issuance turning positive for the first time since 2021. He also noted that the figure does not include potential future public offerings from OpenAI and Anthropic, which he said could increase the total further. For now, he argued, the economics remain attractive. According to Jurrien, current AI-related investments are generating returns in the range of roughly 20% to 35%, making companies less sensitive to whether borrowing costs are 6%, 7% or 8%. The larger question, however, is not the technology itself but the financing required to support it. He suggested that one potential risk is that markets eventually struggle to absorb what he described as a "fire hose" of debt and equity issuance. The substantial spending on data centres and computing infrastructure will ultimately need to generate adequate returns, and he noted that the level of returns required to justify that investment remains uncertain.

A different set-up than the dot-com era

Those questions have naturally led to comparisons with previous technology booms. Jurrien observed that the market's recent price pattern bears a notable resemblance to the dot-com period, though he emphasized that historical analogies are only illustrative and that history does not repeat itself exactly. He argued that the underlying fundamentals differ substantially from the late 1990s. During the dot-com era, the technology sector's forward price-to-earnings ratio reached roughly 66. Today, he noted, it is closer to 20.6. According to Jurrien, investors in the dot-com period were pricing in earnings that ultimately failed to materialize. Today, he sees a different picture, with earnings growth remaining strong while valuations have not kept pace. While he cautioned that this does not guarantee a positive outcome, he argued that the current environment does not resemble the valuation bubble of the late 1990s.

Rotation and the MAG7

Market leadership has also shifted in recent months. Earlier in the summer, both the cap-weighted and equal-weighted S&P 500 reached new highs while market breadth climbed to 74%. More recently, that measure has fallen to roughly 53%, with the MAG7 once again providing much of the market's support. He attributed part of that shift to higher interest rates and elevated oil prices weighing on the broader market. Even so, he characterized the move as part of an ongoing process of market rotation rather than evidence of a broader breakdown. Many hyperscalers, he argued, remain relatively insulated from higher borrowing costs because of the returns they are generating on AI-related investments. Despite higher yields and valuation headwinds, the S&P 500 remains only modestly below its highs, which he viewed as a relatively resilient outcome under current conditions.

Conclusion: watching the plumbing