AI Capital Spending Creates Three Plausible Revenue Scenarios, Not Certain Bubble or Success
Source: Oren Etzioni. "Opinion: The AI capex conundrum – GeekWire." May 7, 2026. www.geekwire.com
The Gist
The author argues that massive AI spending by tech companies could follow three different paths - great success, modest returns, or failure - and we won't know which one until we see more financial data over the next year. Anyone claiming to know for certain which path we're on is just guessing.
Conclusion
The AI capital expenditure boom presents three plausible revenue trajectories (bull, bear, catastrophe cases), and the disagreement between optimists and pessimists will only be resolved by data arriving over the next 12 months, making current certainty claims premature
Premises
- AWS revenue data through Q1 2026 is consistent with three different growth trajectories that imply vastly different returns on Amazon's $200 billion capex plan
- Bulls argue hyperscalers fund AI buildout from cash flow rather than debt, making this different from historical telecom and railway bubbles, with AWS showing 28% growth validating real enterprise AI demand
- Bears counter that hyperscaler debt issuance reached $175 billion in 2026 (six times the five-year average) and GPU assets depreciate in 5 years rather than 30, creating financial fragilities
- The 1990s fiber boom collapsed not from lack of money but because WorldCom claimed internet traffic doubled every 100 days when it actually doubled yearly, showing how wrong growth predictions can destroy markets
- By Q1 2027, AWS quarterly revenue will clearly show whether it's accelerating toward high $40 billions (bull case), tracking flat in low $40 billions (bear case), or declining (catastrophe case)
- The relevant question is not whether AI demand exists, but whether it grows fast enough to absorb $700 billion in annual capital expenditure across the industry
Assumptions
- AWS serves as a representative proxy for the broader AI capital expenditure market
- Historical technology bubbles provide relevant parallels for understanding current AI investment patterns
- Quarterly earnings data over the next 12 months will provide sufficient information to distinguish between the three scenarios
- The financial structure and debt levels of hyperscalers are meaningful indicators of investment sustainability
- Growth rate predictions are the primary driver of capital allocation decisions in AI infrastructure