Learn AI compute, then follow the market

Market Brief

Market Brief: Wednesday, September 16, 2026

Start-of-day AI compute market report.

Published Monday through Friday with the prior day’s source-backed signals across GPU pricing, cloud capacity, power constraints, data centers, and AI infrastructure. Track the prior day’s source-backed news and changes shaping GPU supply, cloud availability, power constraints, data centers, and compute market pressure.

Market Brief

Morning Tape

Report date
Wednesday, September 16, 2026
  • NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut
  • Emerald AI, Google and NVIDIA Launch Alliance to Advance Flexible AI Data Centers
  • AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories

Today in Compute College

Lesson & Term of the Day

Today's lesson

Human review and model-based graders

Combine human judgment and model-based grading without mistaking either one for ground truth. Why it matters today: Today's lesson connects the AI Engineering track to the compute market readers are watching.

Term of the Day

Open model

An open model is an AI model with weights, code, or usage rights made available under terms that allow broader deployment than closed proprietary models. Open models can be run by enterprises, developers, clouds, or neoclouds. The exact rights depend on the license.

Market signals

Compute market signals

H100 IndexWhat it costs to rent one H100 GPU for an hour on demand — the market’s anchor price.

Value
$4.12 / H100-hr
Last Price Change
Sep 16, 2026

Spot Compute PriceThe cheaper, interruptible GPU price — a read on whether spare capacity is available.

Value
$2.46 / H100e-hourH100e-hour means an H100-equivalent GPU hour, a normalized unit used to compare different GPUs and providers.
Last Price Change
May 27, 2026

Model Serving CostRoughly what it costs to run a model and generate its output at scale, per million tokens.

Value
$2.78 / 1M tokens
Last Price Change
Aug 8, 2026

GPU AvailabilityHow easy or hard it is to actually get GPU capacity right now, scored 0–100.

Value
78 / 100 score
Last change
May 27, 2026

Signals

Top market signals

These are the highest-impact source-backed items behind this morning’s brief.

Positive for compute demand · Near term · high confidence · Sep 16, 2026

NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut

The change described in “NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut” can increase inference or self-hosting demand, raising accelerator-hours consumed even without an immediate change in public GPU prices.

Positive for future capacity · Medium term · high confidence · Sep 16, 2026

Emerald AI, Google and NVIDIA Launch Alliance to Advance Flexible AI Data Centers

The project described in “Emerald AI, Google and NVIDIA Launch Alliance to Advance Flexible AI Data Centers” adds prospective compute capacity, but permitting, power procurement, construction, and deployment timing prevent it from increasing immediately available GPU supply.

Positive for compute demand · Near term · high confidence · Sep 15, 2026

AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories

The change described in “AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories” can increase inference or self-hosting demand, raising accelerator-hours consumed even without an immediate change in public GPU prices.

Market effect uncertain · Medium term · medium confidence · Sep 14, 2026

Perplexity Portable Computer Is Now Available on Windows, Powered by NVIDIA RTX

The source for “Perplexity Portable Computer Is Now Available on Windows, Powered by NVIDIA RTX” does not yet establish a supported change in compute supply, demand, price, availability, power delivery, financing, or deployment timing.

Market effect uncertain · Medium term · medium confidence · Sep 16, 2026

Lambda signs the White House Ratepayer Protection Pledge, reinforcing commitment to community benefits and grid reliability

The source for “Lambda signs the White House Ratepayer Protection Pledge, reinforcing commitment to community benefits and grid reliability” does not yet establish a supported change in compute supply, demand, price, availability, power delivery, financing, or deployment timing.

Market effect uncertain · Medium term · medium confidence · Sep 16, 2026

TotalEnergies partners with Mistral to develop AI models for oil and gas exploitation

The source for “TotalEnergies partners with Mistral to develop AI models for oil and gas exploitation” does not yet establish a supported change in compute supply, demand, price, availability, power delivery, financing, or deployment timing.

Power & Data Center Watch

  • Emerald AI, Google and NVIDIA Launch Alliance to Advance Flexible AI Data Centers (NVIDIA Blog).
  • Lambda signs the White House Ratepayer Protection Pledge, reinforcing commitment to community benefits and grid reliability (Lambda Blog).
  • Goodman raises $455m for its Hong Kong data center partnership (Data Center Dynamics).

Market report

Latest GPU pricing report

For a durable pricing artifact, use the monthly report rather than a single Market Brief issue.

H100 GPU Pricing Report: August 2026

Canonical CT-H100 median, reviewed H100 provider rows, freshness dates, and composition caveats in one citable report.

Recent Briefs

Brief archive

Use the archive to move between dated Market Brief issues without relying on the sitemap.

Previous issue: Market Brief: Tuesday, September 1, 2026

Nscale Closes Approximately $3 Billion in Financing for Both Ward County, Texas and Madison, North Carolina AI Deployments Learn more