[Today’s Signal] AI Is Absorbing Capital — And Long-Term Rates Are Rising Even as the Fed Stands Still

AI is becoming a major driver of capital demand worldwide. Vast sums are flowing into data centers, GPUs, and power grids to fuel the AI revolution. The question now is who can secure the necessary long-term capital and afford its rising cost. Despite the Federal Reserve's relatively stable policy, long-term U.S. Treasury yields have surged above 5.3%, their highest level since 2007.

This divergence between monetary policy and capital markets is driven by factors like fiscal deficits, heavy Treasury issuance, inflation concerns, and reduced foreign demand for U.S. debt. Added pressure comes from record private-sector capital demand to fund AI infrastructure. Major technology companies like Amazon, Alphabet, Meta, and Oracle have issued billions in corporate bonds, with projected issuance reaching $400 billion by 2027.

Goldman Sachs projects hyperscalers, including Microsoft, will raise about $250 billion this year alone. When viewed more broadly, bond issuance by AI-related firms is even larger. For example, Reuters estimates that hyperscalers have raised about $220 billion for AI infrastructure this year. Governments and AI companies are now competing for a limited supply of long-duration funding, leading to "capital scarcity."

This shift is evident in corporate bond demand, with investor orders for hyperscaler bonds falling from five times offered in February to less than two times in July. As investors demand higher yields to absorb new supply, companies' financing costs are rising. The distinction between policy rates and the cost of capital has become critical.

Even if the Fed does not raise rates, higher Treasury yields and wider corporate bond spreads can increase long-term financing costs for data center operators. AI companies' ability to secure long-term capital becomes as crucial as the Fed's rate decisions. Alphabet's recent $3.89 billion bond issuance in Australia exemplifies this trend, as major technology companies are expected to invest over $730 billion this year.

Nvidia's $105 billion guarantee for an OpenAI data center project in Ohio highlights how far the competition for AI capital has extended. The world's largest supplier of AI chips is assuming part of the financing risk for a customer facility that will use its GPUs, and Goldman Sachs is facilitating over $500 billion in AI infrastructure financing via a new financial platform.

While circular financing structures are common in capital-intensive industries, the $105 billion guarantee raises concerns about potential circular financing. The success of this structure depends on whether the resulting cash flow from AI services can support the high interest expenses, GPU depreciation, and power costs. In Korea, the AI capital boom is also driving semiconductor profits.

The Korean government expects a 3.5% growth in 2026, its strongest pace in five years, fueled by the AI semiconductor upcycle. Moody's recently raised its forecast for Korea's 2026 growth from 1.8% in February to 3.5%, nearly doubling its projection in six months.