War, AI buildout, and rate hikes: why interest rates are stuck above 5% and what breaks first

War, AI Buildout, and Rate Hikes: Why Interest Rates Are Stuck Above 5%

US interest rates have climbed above 5% for the first time since 2007, and the driver is straightforward: war in the Middle East is triggering an energy shock that bleeds into inflation across the entire economy.

The mechanism works through oil. The Strait of Hormuz is a critical chokepoint for global energy supply, and the ongoing conflict has created scarcity. Energy prices rise. Those costs cascade into food prices, wages, and everything else that depends on transportation and production. Inflation accelerates. When inflation expectations rise, treasury yields spike to compensate investors for eroding purchasing power. The result: the Fed is effectively locked out of cutting rates, even as borrowers face mortgage rates above 7% and refinancing pressures mount.

The war-rates relationship is not automatic. During Gulf War One, interest rates spiked from 8.29% to 9.05% in less than a month when Iraq invaded Kuwait on August 2, 1990. But the conflict resolved quickly—the Desert Storm offensive lasted roughly six months—and rates fell back to 8.03% by the end. The Afghanistan war followed a different pattern. Rates initially fell to 4.22% on optimism it would be a quick operation, then climbed to 5.25% six months in as the conflict became a prolonged quagmire. The difference: duration. A contained war creates a temporary supply shock. A grinding, indefinite conflict keeps inflation expectations elevated indefinitely.

AI is making the rate problem worse, but in a counterintuitive way. The sector is creating upward pressure on rates not because it's failing, but because it's succeeding—and succeeding expensively. Roughly $1 trillion in AI infrastructure spending is underway against only a few hundred billion in actual AI revenues today. That capital demand for data centers, power, and equipment is happening now, while productivity benefits take years to diffuse. NVIDIA is actively pulling investment capital away from other sectors by securitizing GPU purchases and lending investment-grade status to data center buildouts, which opens these deals to mutual funds and insurance companies that can only invest in investment-grade assets. This competitive demand for capital drives up rates.

There is also a wealth effect. The stock market's rally on AI enthusiasm has put money in investors' pockets. Even modest gains on tech holdings translate into discretionary spending—car upgrades, appliances, travel—that drives consumer demand and supports inflation.

The Fed's inflation problem is broad. Consumer price inflation sits at 3.4%; the Fed's preferred PCE measure is at 3.7%. Even stripping out food and energy, core inflation is at 3.3%. There is no easy narrative of a temporary energy shock that will naturally resolve. Inflation is spilling into the broader economy.

There is a forked outcome. Apollo describes two scenarios, both of which would lower rates, though through opposite mechanisms. If AI succeeds, productivity gains will diffuse into the economy—law firms, hospitals, courts—creating competition and deflation. Costs fall, demand for capital normalizes, rates fall. If AI fails and the bubble bursts, there will be a flight to treasuries as investors panic out of equity positions. Heavy treasury buying pushes yields down and rates fall. A third, more extreme scenario from Dylan Patel imagines the AI buildout never stops, continuously absorbing capital until a sovereign debt crisis forces a resolution.

The near-term question is what breaks first: the war, the AI buildout, or inflation expectations. The next six months will likely clarify which scenario is underway.