THE METER
SECTION 03
ISSUE 001
Inference Becomes a Metered Utility
Projection: OpenAI and Broadcom are co-designing an inference accelerator around kernels, memory movement, networking, and serving; the IEA identifies rapid load growth and grid constraints. Buyers may procure guaranteed useful-work capacity rather than named chips or tokens. That utility model requires comparable task, latency, and reliability meters that do not yet exist.
Why this idea is here
What the evidence establishes.
OpenAI and Broadcom announce a custom inference accelerator co-designed around LLM kernels, memory movement, networking, and serving patterns; The IEA reports fast 2025 datacenter electricity growth, rising AI-focused load, and constraints in grids, transformers, turbines, and approvals. These are source-backed premises for this projection; they do not by themselves prove broad adoption or the eventual outcome.
Source ledger
Read the sources.
- S01OpenAI and Broadcom unveil LLM-optimized inference chip
official hardware announcement / published 2026-06-24 / retrieved 2026-07-09
- S02Data centre electricity use surged in 2025
intergovernmental research release / published 2026-04-16 / retrieved 2026-07-09