Debating the Scale of AI Compute Capex
LexSokolin · x · 2026-07-13
This thread discusses the actual scale of AI capital expenditure, focusing on the debate over scale estimates following David Cahn's updated "AI's $600 Billion Question."
Key points include:
- One estimation method based on NVIDIA's trajectory multiplies revenue (which already includes networking equipment) by TCO, potentially double-counting some interconnect costs.
- Another approach starts from roughly $75 billion in total hyperscaler capex, but this includes replacement investments and non-AI spending, making it aggressively optimistic to attribute all of it to AI growth.
- Commenters believe the core issue isn't just "how much is invested," but whether these capital expenditures can actually be recouped.
- It also notes that the definition of a 50% margin has shifted across different years, making direct comparisons between early and late stages inaccurate.
More from Venture
- An indie builder open-sourced 50+ AI apps and says users should only pay for tokens — matchaman11 · 2026-07-21
- Augustus raises $180M at a $1B valuation to build an AI-native dollar bank — _rockt · 2026-07-21
- UK robotics startup Humanoid raises $152 million at a $1.35 billion valuation — Polymarket · 2026-07-21
- Former officer says he retired at 29 by placing small machines in local businesses — ahuja_priyank · 2026-07-21
- Presidio acquires LookingPoint to expand AI, cloud and cybersecurity — TechNadu · 2026-07-21
- AI bottlenecks are shifting to memory, optics, yield control and power — thedealdirector · 2026-07-21