AI model pricing splits into electricity, GPU rental and profit, as hardware takes a bigger share
davidmanheim · x · 2026-07-27
Model providers’ pricing is broken into electricity, amortized hardware costs or rented GPUs, and profit margin. The point of the thread is that as electricity stays relatively stable, a larger share of every extra dollar in AI company margins is effectively captured by hardware providers and other non-electricity costs.
The reply argues that hardware suppliers want as much of the cost as possible to sit in hardware, because electricity prices barely move compared with other inputs. That means the economics of AI token pricing are increasingly shaped by scarce hardware and rent-seeking around compute rather than by power alone.
Related event: AI Inference Services Cost Up to 15x More Than Renting GPUs(4 posts)→
More from Infra
- fmgo: call Apple's on-device Foundation Models from Go with no CGO and no Swift — Super_Run_8466 · 2026-09-23
- Huawei unveils Peerium architecture: nested BSP unifies million processors into one computer — Dr_Singularity · 2026-09-23
- Grok explains why DeepSeek picked DualPipe + ZeRO-1 over ZeRO-3 on 2048 H800s — TheZachMueller · 2026-09-23
- AI costs fall 47% per quarter, 4x faster than DNA sequencing: Epoch AI — daveholtz · 2026-09-23
- M5 Ultra LLM test: 4x faster prompt processing, but double the power draw — DigitalguyCH · 2026-09-23
- $500 of Dell OptiPlexes become a diskless netboot lab where AI agents can't brick the hardware — colinmcnamara · 2026-09-23