Cloud hosts may earn higher margins selling open-source model tokens than GPU hours
zephyr_z9 · x · 2026-07-28
A reposted comment argues that cloud providers hosting open-source models and selling tokens can make materially higher gross margins than selling raw GPU hours.
The post says open-source model adoption is already putting pressure on pricing and margins for frontier model vendors, while upstream semiconductor demand is mostly not a substitute effect — except for DeepSeek, most open models still consume substantial inference compute. It also claims the growth in open-source model ARR has been strong since June, but overall adoption is currently constrained by compute supply.
The final analogy frames frontier closed models as “Apple A” and “Apple B,” while open-source models are likened to Android: the commenter expects Android-like models to take the majority of the market.
More from Infra
- Agentic AI system reportedly cost $1.2M a month before being cut to $100K — emmanuelvivier · 2026-07-28
- Wistron’s Early Nvidia Bet Turned It Into One of AI’s Biggest Supply-Chain Winners — pstAsiatech · 2026-07-28
- RTX 5090 prices are up €1,061 in Europe as local inference demand stays hot — egudegi · 2026-07-28
- GitHub repo tracks daily IP ranges for OpenAI GPTBot, Googlebot and major clouds — tom_doerr · 2026-07-28
- US power grid proposes making data centers pay fully for the electricity they use — pstAsiatech · 2026-07-28
- A blunt rebuttal to the idea that data-center debt is the new subprime crisis — JOBhakdi · 2026-07-28