Andrew Chen says AI value is shifting from models to hardware, then to vertical AI
andrewchen · x · 2026-07-25
Andrew Chen says AI value capture is moving from models to hardware to vertical apps
Andrew Chen sketches a fast-evolving narrative:
- first, GPT wrappers led people to think models would capture all the value;
- then frontier open weights shifted attention to hardware as the main value owner;
- next, he expects “good enough” local AI on lower-end hardware to make vertical AI the layer that captures the most value.
The post is not a product announcement, but a compact thesis about where AI business value may accumulate as deployment becomes cheaper and more distributed.
More from Companies & People
- DeepSeek-style cost discipline could curb bloated agent environments, poster argues — teortaxesTex · 2026-07-25
- FT profiles Moonshot AI founder Yang Zhilin as a rising China AI star — EleanorOlcott · 2026-07-25
- Some CEOs are planning to move off OpenAI and Anthropic by year-end — fkasummer · 2026-07-25
- Flock Safety’s license plate readers helped locate a suspect car in 12 hours — aleximm · 2026-07-25
- Fluidstack hosts San Francisco dinner on the future of AI infrastructure — MxMnr · 2026-07-25
- AI cuts one GTM workflow from millions of dollars to $30 — briannekimmel · 2026-07-25