Gigawattonomics: Revenue Per Watt Dictates AI Compute Capex Returns

BenBajarin · x · 2026-08-11

In a paid post, Ben Bajarin introduces the "Gigawattonomics" model to evaluate the return on invested capital for AI factories (data centers). The model posits that while power availability determines the deployment schedule of compute capacity, revenue per watt dictates whether that capacity earns an adequate return.

The analysis emphasizes that when the power envelope is fixed, the platform that turns each watt into the most useful monetizable output generates the best economic return. Therefore, "revenue per watt" serves as the grounding metric for compute capex. According to the model, NVIDIA systems carry higher capex per gigawatt compared to custom-silicon and AMD alternatives, but maintain competitive advantages through conversion efficiency.

Original post →

More from Venture

Venture channel →