 |
| This image is AI-generated |
As newly appointed Federal Reserve Chair Kevin Warsh prepares to lead the upcoming Federal Open Market Committee (FOMC) meeting on July 28–29 in Washington, D.C., market participants and businesses are keenly focused on any indications of how interest rates may evolve in the months ahead.
Following his first meeting as the new Fed Chair in June
2026, Kevin Warsh signalled a hawkish policy stance. The Committee explicitly
removed his previous rate-cutting language from the policy statement and reaffirmed
its commitment to price stability. As per the analysis done on Warsh’s previous
speeches, Warsh never once dissented from an FOMC decision, while he served on
the Board from 2006–2011. Furthermore, Bloomberg studied 13 speeches in which
he specifically expressed concern with respect to the upside risks to inflation,
showcasing his hawkish approach. With inflation currently at 4.2%, its highest
level in three years, financial markets are already pricing in approximately a
42% probability rate hike at the upcoming meeting. AI infrastructure is
arguably the most capital-intensive corporate investment cycle in modern
history. Unlike previous software-led technology disruptions, the economics of
AI depend not only on innovation and demand, but also on the robustness of its physical
infrastructure
Any increase in interest rates would directly raise the cost
of debt, creating additional financing pressures for companies reliant on
external capital. This is particularly significant for hyperscalers, which have
collectively earmarked approximately $700 billion in capital expenditure for AI
infrastructure in 2026. With operating cash flows heavily consumed by capex,
hyperscalers have turned to the bond markets at an unprecedented scale. Higher
borrowing costs extend beyond incremental interest expense. They increase the
weighted average cost of capital (WACC), forcing companies to apply higher hurdle
rates when evaluating future projects. Projects that appeared economically
attractive in a low-rate environment may no longer generate sufficient returns.
The Big Five hyperscalers raised over $121 billion in new
debt during 2025, with over $90 billion of that concentrated in the final three
months of the year. JP Morgan and Morgan Stanley project the technology sector
may need to issue as much as $1.5 trillion in new debt over the coming years to
finance AI and data centre construction. It is not surprising that for the
first time, hyperscalers are collectively expected to hold more debt than cash
on their balance sheets.
Credit risks are already mounting across the AI
infrastructure ecosystem, driven by negative free cash flow, expensive debt
financing, and growing reliance on off-balance-sheet financing structures (such
as special purpose vehicles). These financing pressures are not restricted to
the hyperscalers but translate across the broader value chain including
semiconductor manufacturers, data centre developers, utilities, equipment
suppliers, construction firms, and investors.
The AI infrastructure buildout has notable similarities to
the telecommunications investment boom of the late 1990s. Following the Telecommunications
Act of 1996, industry deregulation triggered a wave of capital expenditure,
which peaked at approximately $121 billion in 2000 ( ~$213 billion in today's terms).
Much of this investment was driven by the widely cited belief that internet
traffic was doubling every 100 days, a projection that originated from a
best-case capacity planning model developed by a WorldCom engineer in 1997. At
the same time, the Federal Reserve tightened monetary policy, raising the
federal funds rate from 4.75% in January 1999 to 6.50% in May 2000. While
internet demand continued to grow, it fell short of the significantly bullish
projections. Even four years after the bubble burst, an estimated 85–95% of the
fiber-optic network deployed remained unused as "dark fiber," and at
least 23 US telecommunications companies filed for bankruptcy. The idea is not
that AI buildout will fail but drawing lessons from the past, businesses should
be aware that there can cases of severe capital misallocations
The experts argue that today's wave is "more
disciplined." The lead financiers (Microsoft, Google, Amazon, Meta) have
investment-grade-elite ratings, real cash flows, and profitable core businesses.
However, the argument rests entirely on the premise that these companies are
deploying funds from operating cash flows rather than raising external capital.
While the leading hyperscalers continue to generate substantial operating cash
flows, those cash flows are no longer sufficient to finance the entire scale of
AI investment. as external financing becomes an increasingly important source
of capital, the industry's sensitivity to monetary policy is also rising. The
greatest risk to the AI infrastructure boom is shifting from technology risk to
financing risk.
The views expressed are personal and do not represent views of the organization I work for
No comments:
Post a Comment