AI Compute Economics: 1GW Requires $450B Infra, Consumes 87.6TWh/Year
demian_ai · x · 2026-08-16
Based on The All-In Podcast, the economic model for 1GW of productive AI compute involves roughly $100B in lab revenue, $50B to compute providers, and $30B flowing to Nvidia systems.
Adding 6-8 GW requires $300-400B in capex ($50B per new GW). Scaling revenue from $100B to $1T requires 10 GW total, implying $450B in incremental infrastructure and $300B in silicon.
10GW running continuously consumes 87.6 TWh per year, highlighting energy as the next major bottleneck.
More from Infra
- 14 open-source LLM observability tools to know in 2026, from Langfuse to Laminar — TheTuringPost · 2026-10-16
- Local AI is a routing problem, not a privacy feature: where each task should run — bigdata · 2026-10-16
- Moving agent work from GPUs to CPUs for durable, predictable actions — cramforce · 2026-10-16
- First TPUv7 agentic benchmark on vLLM claims 1.42x perf-per-dollar over GB300 NVL72 — vllm_project · 2026-10-16
- Phison CEO calls CXMT memory scare a 'ghost story', warns of materials shortage — BenBajarin · 2026-10-16
- Ben Bajarin: The Future of Compute Is Fungible as Hyperscalers Design for Workload Flexibility — BenBajarin · 2026-10-16