AI’s Deflationary Product Cannot Support the $5 Trillion Bill
Key takeaways
- AI’s rapid cost decline is outpacing the productivity gains needed to justify the multi‑trillion‑dollar infrastructure spend, risking a costly misallocation of capital.
- Inference‑token costs falling 47% per quarter
- $750 billion data‑center spend this year
- $1 trillion next year
What happened
Hyperscalers such as Alphabet, Microsoft, Amazon, Meta, and Oracle are pouring more than $5 trillion into AI data‑center infrastructure over the next four years, while AI inference‑token costs are falling 47% per quarter. The capital outlay far exceeds the productivity gains needed to cover depreciation, capital costs, and a 15% hurdle rate, raising concerns that the build‑out could be the largest misallocation of capital in history.
Why it matters
If the AI spend does not translate into proportional productivity and revenue gains, these companies will face free‑cash‑flow deficits and will need to raise external capital, tightening the already strained U.S. fiscal environment. The mismatch also threatens the sustainability of the AI boom and could shift value…
Short-term and longer-term
Short-term. Short‑term traders should monitor the release of quarterly free‑cash‑flow statements from Alphabet, Meta, and Microsoft, as any continued deficits could trigger margin calls or equity dilution.
Longer-term. Long‑term investors need to reassess the sustainability of the AI boom, weighing the projected $5 trillion infrastructure spend against the modest productivity gains reported by executives.