Investment
Sustainability and Overinvestment
- Unsustainable Capex: A massive
surge in capital expenditure (capex) on AI infrastructure—led by
"hyperscalers"—is currently outpacing the earnings and free cash
flow of these firms.
- Contest-Driven Overinvestment: Intense
competition for market leadership may fuel further overinvestment, as
firms race to dominate an uncertain future market.
- Historical Boom-Bust Cycles: The current
boom bears a striking resemblance to historical episodes like the
19th-century "railway mania" and the late-1990s "dotcom
boom," where technological breakthroughs attracted capital far in
excess of what commercial returns could ultimately justify, leading to
eventual busts and recessions.
Supply-Side
Roadblocks
- Resource Bottlenecks: The buildout
of AI is facing growing shortages in electricity, advanced semiconductors,
and grid equipment.
- Inflationary Pressures: Massive
demand for computing power is already driving up electricity prices and
input costs, which could spill over into broader inflation.
Will the current AI boom be the biggest Boom-Bust cycle? Is it currently still at the starting phase?
Financial
Vulnerabilities and Opaque Financing
- Circular Financing: The AI
sector is characterized by a complex, opaque web of private arrangements,
most notably "circular financing," where chip makers and
hyperscalers take equity stakes in AI labs that, in turn, commit to
multi-year purchases of their products.
- Rising Leverage and Debt: Investment
is increasingly financed by debt rather than just cash flow, leaving firms
vulnerable to a sudden pullback in financing if payoffs disappoint.
- Concentrated Private Credit: Private
credit funds have significantly expanded their lending to the AI and IT
sectors, leading to high concentration risks as multiple lenders often finance
the same software borrowers.
Macroeconomic
and Asset Price Risks
- Stretched Valuations: Equity
valuations for core AI firms are highly elevated, pricing in ambitious
long-term earnings growth that may be difficult to sustain as these firms
mature.
- Wealth Effect Sensitivity: Because
household exposure to stocks has grown relative to wealth and income, a
major correction in AI-related equity markets could trigger a sharper
pullback in consumer spending than in the past.
- Repricing Feedback Loops:
Disappointments in AI payoffs could trigger a sharp tightening of
financial conditions, where compressed risk premia unwind abruptly,
potentially leading to a corporate credit freeze.
Labor Market and Long-Term Growth Concerns
- Labor Displacement: Unlike
previous technologies that provided better tools for workers, AI competes
directly with human cognitive abilities, increasing the risk of
significant labor displacement.
- The "Demand Bottleneck": In a worst-case long-term scenario, widespread automation could divert so much income from labor to capital that the consumer base erodes; if there is insufficient demand to justify further expansion, productivity could stall, leading to a "demand bottleneck".
Technological
and Cyber Risks
- Enhanced Cyber Threats: Frontier AI
models lower the cost and accelerate the pace of cyberattacks, which can
identify and exploit financial system vulnerabilities at scale.
