Wall Street is pouring trillions into AI infrastructure, with Morgan Stanley projecting $3 trillion in spending by 2029. But is this the foundation of a new economy or the setup for a collapse that could dwarf the dot-com crash? Can AI revenues scale fast enough to justify the most concentrated infrastructure bet in modern history before funding conditions tighten or expectations collapse?

Azeem Azhar—founder of Exponential View, author, and one of the world's leading analysts of exponential technologies—has spent months developing a rigorous framework to answer this question. Drawing on his experience as both an investor who lived through the dot-com bubble and a technologist studying transformative shifts, Azhar has built a five-gauge dashboard that benchmarks today's AI boom against history's most catastrophic bubbles: the railway manias of the 1870s, the telecom frenzy of the 1990s, and the dot-com crash of 2000.

Azeem's analysis examines five critical dimensions, each calibrated against historical precedent:

• Economic Strain (AMBER): AI infrastructure spending is approaching 0.9% of US GDP in 2025, projected to hit 1.6% by 2030. More than a third of current US GDP growth traces back to data center construction. When railways reached 4% of GDP in 1872, the Panic of 1873 followed. Are we approaching dangerous territory?

• Industry Strain (AMBER-RED): GenAI generates roughly $60 billion in revenues against $370 billion in capital expenditure—a 6:1 ratio that's worse than railways (2:1) and telecoms (4:1) at their peaks. Hyperscalers now invest 68% of operating cash flow in capex, up from 44% pre-ChatGPT. How long can this imbalance persist?

• Revenue Growth (GREEN): GenAI revenues are doubling annually, with projections of 122% compound growth through 2028. This diverges sharply from past bubbles—railways grew 22% before the 1873 crash, telecoms just 16% before 2001. But there's a catch: GPUs depreciate in three years, not the decades enjoyed by railway tracks or fiber optic cables.

• Valuation Heat (GREEN): The Nasdaq trades at a P/E of 32—half the dot-com era's 72. At the 2000 peak, internet stocks implied a P/E of 605. Today looks prudent by comparison, but is that the right comparison?

• Funding Quality (GREEN-AMBER): Tech giants can self-fund about half the projected $2.9 trillion needed through 2028. The $1.5 trillion gap must come from private credit, securitized finance, and government pledges. Consider CoreWeave: $8 billion in debt, two customers, GPUs depreciating 20-30% annually. When does healthy expansion become fragile speculation?

Topics open for community Q&A:

• Is AI different this time? Which historical parallel is most relevant—railways, telecoms, or dot-com—and why the differences matter

• The dashboard methodology: How to weight and interpret the five gauges, and which warning signals matter most

• Critical thresholds: What specific triggers could flip this from boom to bust within the next 2-3 years

• Investment strategy: How investors and executives should position themselves given the current gauge readings

• The depreciation problem: Why GPU lifecycles fundamentally change the risk calculus compared to past infrastructure buildouts

• The funding gap: Where the $1.5 trillion shortfall will come from and what it means for market stability

• Sector-specific risks: Which players are most exposed if momentum falters—hyperscalers, startups, or the emerging debt markets

• The enterprise adoption question: Whether current revenue growth can possibly continue, and what happens if it doesn't

• Macro wildcards: How recession, interest rates, or geopolitical shocks could accelerate bubble dynamics

• Opportunities in uncertainty: Where to find asymmetric bets if you believe in the technology but recognize the risks