AI bubble may deflate slowly as $10 trillion needed by 2032
Economist Kyla Scanlon warns the AI boom could unwind gradually, needing $10 trillion in revenue and facing U.S. data‑center power constraints.
Economist and author Kyla Scanlon says the AI sector is more likely to unwind gradually than to implode suddenly, noting it must generate roughly $10 trillion by 2032 to make today’s massive infrastructure spend worthwhile [1]. She adds that many AI firms are financing growth with heavy borrowing, observing “the companies are borrowing quite heavily; they’re spending a ton of money” [1]. At the time of the interview, about three‑quarters of S&P 500 companies were trading below their 50‑day moving average, a sign of broader market softness [1]. Scanlon also points to a shortage of power for U.S. data centers, a bottleneck that could curb AI compute capacity and pressure operators to manage costs more tightly [1]. Together, these factors suggest a slow deflation of the AI bubble, with capital efficiency and energy constraints shaping the next phase of the industry.
Questions and answers
Why does Scanlon say the AI bubble will deflate slowly?
She expects a gradual unwind because revenue growth must catch up to the massive capital already invested, and financing pressures are high [1].
What is the $10 trillion figure about?
It represents the revenue AI companies would need to generate by 2032 to justify current infrastructure build‑out costs [1].
How does the U.S. power shortage affect AI development?
Limited electricity for data centers can restrict compute capacity, raising costs and slowing AI model training and deployment [1].
Companies & investors in this story
Sources
- 1.Q&A: Why economist and author Kyla Scanlon thinks the AI bubble will pop slowly — BetaKit, Oct 23, 2026
- 2.Anthropic, DeepSeek and drones: 10 takeaways from the 2026 unicorn boom — e27, Oct 27, 2026