What Justifies Trillion-Dollar Valuations for Foundation Models
paraschopra · x · 2026-07-16
The author argues that valuations fundamentally depend on expected lifetime profits, which in turn rely on moats. They are skeptical of the high valuations of current **foundation model companies** for the following reasons: - Much of their revenue comes from **API calls**, a use case that can easily switch to cheaper, similarly performing open-source models. - In contrast, the historically high valuations of FAANG companies were driven by stronger moats like **network effects** and **switching costs**. - **Nvidia's** valuation is easier to justify because of its robust **CUDA ecosystem**, and chips aren't easily replaced quickly once purchased. The core question remains: if the differences between frontier models shrink and they compete increasingly like commodities, what exactly supports their trillion-dollar valuations?
More from Venture
- Mercor is being framed as a $20 billion machine in frontier AI — Glum-Adagio7489 · 2026-07-21
- Yuandong Tian says AI researchers are becoming the stars closest to “what’s next” — FinanceYF5 · 2026-07-21
- Recursive Superintelligence raises $650M at a $4.65B valuation — FinanceYF5 · 2026-07-21
- AI Visibility Tracking Startup Hall Acquired by Tracksuit — gaganghotra_ · 2026-07-21
- Five U.S. tech giants’ hidden AI debts rise to $1.65T — NordStreamYacht · 2026-07-21
- Andera raises $37M Series A to automate audit work with AI agents — Jackyhuang · 2026-07-21