The AI Boom is becoming another Infrastructure Super‑Cycle with familiar dangers.

Every few generations, the global economy launches into a massive, capital‑intensive infrastructure boom to support a transformative technology. We’ve seen this pattern for 200 years: railways, electrification, and the digital networks of the late 20th century. Each era began with optimism, required enormous physical investment, and exposed the financial fragility of secondary stakeholders: municipalities, utilities, land boards, water authorities, and regional operators. Many over‑extended, failed to meet long‑duration obligations, and ultimately required public bailouts.

Today’s AI boom is following the same script.

AI is not just algorithms – it’s infrastructure

Behind every AI model sits a sprawling physical system:

• land corridors for data centres and transmission
• water rights for cooling
• wavelength and spectrum for high‑capacity networks
• transportation corridors for equipment and labour
• regional utilities facing unprecedented power loads

This is a multi‑jurisdictional, multi‑asset, multi‑decade build‑out, exactly where financial stress tends to emerge.

The weakest link problem

The largest players, national governments, hyperscalers, major utilities, can absorb shocks. The smaller ones cannot.

AI infrastructure depends on:

• municipalities issuing long‑term land leases
• water authorities guaranteeing supply
• regional telecom operators financing fibre and spectrum
• transportation agencies upgrading corridors
• quasi‑public utilities maintaining grid inter-ties

These secondary owners often borrow heavily against future demand and take on obligations lasting 20 to 50 years. When forecasts fail, they are the first to default and their failure threatens the entire ecosystem.

Capital‑market pressure makes this cycle even riskier

The U.S. government is running historic deficits, putting sustained pressure on global capital markets. Large deficits mean:

• higher Treasury issuance
• tighter liquidity
• upward pressure on interest rates
• more competition for private capital

Higher rates make long‑duration infrastructure financing more expensive, increase refinancing risk, and raise the likelihood of private‑sector bankruptcies. Public budgets, already strained, face additional pressure.

In short: the first generation of AI infrastructure could be undermined not by technology, but by capital‑market stress.

Entrepreneurs still face ordinary business risk

All of this comes on top of the usual hazards facing any private entrepreneur:

• competition
• execution challenges
• market‑adoption uncertainty
• regulatory shifts
• supply‑chain volatility
• financing constraints
• technological obsolescence

The AI boom layers infrastructure fragility and capital‑market pressure on top of standard business risk making early‑stage survival harder than the narrative suggests.

The benefits of AI won’t appear immediately

A final historical lesson: the direct economic benefits of a new technology do not appear until the technology is widely disseminated and accessible. For railways, electrification, and digital networks, this took 5 to10 years after the infrastructure was completed.

AI will follow the same curve. The productivity gains arrive only after compute, power, networks, and access become ubiquitous not during the build‑out.

Should we not learn from history?

We are once again building the physical backbone of a transformative technology. We are once again relying on stakeholders with uneven financial strength. We are once again layering long‑duration obligations onto institutions that may not sustain them. And we are doing so in a capital‑market environment where rising interest rates amplify every risk.

The lesson from two centuries of infrastructure cycles is simple: The success of a technological revolution depends not on the strongest players, but on the weakest and the benefits only arrive long after the infrastructure is complete.

If we want the AI boom to be sustainable, we must:

• strengthen governance
• ensure realistic financing models
• coordinate across jurisdictions
• build resilience before stress arrives
• recognize capital‑market risk
• support entrepreneurs carrying business risk

The technology will succeed. The question is whether the infrastructure behind it will be financially sustainable or whether we will repeat the same mistakes that have accompanied every major infrastructure boom for two centuries.