AI is driving up the cost of capital. Image: X Screengrab

Ten-year US Treasury yields are hovering near 4.5% even though bond markets are showing remarkably little panic about long-term inflation.

Oil prices have jumped, conflict in the Middle East has intensified and headlines scream inflation risk daily. Yet the bond market’s own gauges of future price pressures remain relatively subdued. 

Barclays points out that US 10-year breakeven inflation rates still sit around 50 basis points below the peaks reached during the brutal tightening cycle of 2022. The five-year, five-year forward inflation expectation measure — one of the cleanest indicators of medium-term inflation expectations — now trades near 2.2%.

Investors blaming geopolitics alone for higher borrowing costs are missing the bigger shift underway. Artificial intelligence (AI) is starting to push real yields structurally higher.

Bond markets increasingly believe AI will create enormous capital demand, stronger productivity growth and a materially higher neutral rate across the global economy. Real yields are rising because investors no longer see the post-2008 global financial crisis world of permanently cheap money as sustainable.

For 15 years, developed economies were trapped in a low-growth, low-productivity environment. Corporates underinvested, productivity stagnated and central banks flooded markets with liquidity because private-sector demand for capital remained weak.

AI is changing this equation completely. Microsoft plans roughly US$80 billion in capital expenditure this fiscal year, largely tied to AI infrastructure. Amazon is expected to spend more than $100 billion.

Alphabet recently lifted its annual capital expenditure guidance to around $75 billion. Meta Platforms plans as much as $72 billion next year. Nvidia became the emblem of the AI boom after revenue growth exploded more than 260% year-on-year during the buildout phase.

Data centers require gigantic investment in semiconductors, cooling systems, electricity generation, transmission grids, fiber networks and physical construction. Goldman Sachs estimates global power demand from data centers could rise by as much as 165% before the decade ends.

Bond markets understand the implications. Large-scale investment booms push up the cost of capital because demand for money outpaces supply. Governments borrow more aggressively, companies race to secure strategic advantages and productive opportunities multiply simultaneously across sectors.

AI is already triggering all three. Real yields are therefore climbing because investors increasingly believe future growth will be stronger than markets previously assumed. Productivity gains from AI could reshape finance, healthcare, defense, logistics, manufacturing and software engineering at high speed. 

McKinsey estimates generative AI may eventually contribute between $2.6 trillion and $4.4 trillion annually to the global economy. PwC projects AI could add almost $16 trillion to global GDP by 2030.

Bond investors can’t ignore numbers on that scale. Markets have, until now, spent years pricing in permanent stagnation, yet AI is forcing a repricing toward expansion. Another factor that receives far too little attention: AI could prove inflationary during the buildout phase even if longer-term productivity gains eventually suppress costs.

Zoom in, and you’ll see electricity demand is surging while copper markets are tightening. Semiconductor supply chains remain constrained and engineering talent is scarce.

Utility infrastructure requires enormous upgrading and nuclear power discussions have moved back into mainstream policy conversations, largely because existing grids may struggle to support projected AI demand growth.

Shortages create pricing pressure while fiscal conditions worsen the problem. US federal debt held by the public is approaching 100% of GDP, according to Congressional Budget Office projections. Refinancing costs continue to climb rapidly. Annual interest payments on the US national debt recently overtook defense spending for the first time in modern history.

Bond markets notice every part of this equation. Long-duration sovereign debt now faces pressure from multiple directions simultaneously: heavier issuance, rising capital demand, stronger productivity expectations and the growing belief that neutral interest rates may settle far above post-crisis norms.

Investors waiting for yields to collapse once oil prices stabilize may be badly misreading the moment. AI is no longer simply a tech story driving equity excitement. Bond markets increasingly view it as a macroeconomic force capable of reshaping the global cost of money itself.

Financing technological revolutions has never been cheap. And the evidence is showing in real time that AI will not be different.

Nigel Green is CEO and founder of the deVere Group