Market Brief: Friday, August 28, 2026

Today's lesson

What is a Megawatt of AI Compute?

A megawatt of AI compute is a power-based way to describe possible data-center and accelerator capacity. Why it matters today: Today's lesson connects the Power & Data Centers track to the compute market readers are watching.

Term of the Day

Orchestration

Orchestration is the coordination layer that decides which models, tools, data sources, and workflows are used for an AI task. In complex systems, orchestration may route requests to different models based on cost, speed, quality, or security needs.

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

Value

$5.22 / H100-hr

Last Price Change

Jul 17, 2026

Spot Compute PriceiThe 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 CostiRoughly 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 AvailabilityiHow easy or hard it is to actually get GPU capacity right now, scored 0–100.

Value

78 / 100 score

Last change

May 27, 2026

Market effect uncertain · Medium term · medium confidence · Aug 26, 2026

NVIDIA NVLink Fusion Expands With NVHBM Custom High-Bandwidth Memory

The source for “NVIDIA NVLink Fusion Expands With NVHBM Custom High-Bandwidth Memory” does not yet establish a supported change in compute supply, demand, price, availability, power delivery, financing, or deployment timing.

Positive for future capacity · Medium term · medium confidence · Aug 27, 2026

LG CNS deploys direct-to-chip cooling for Naver Cloud at data center in Goyang City, South Korea

The project described in “LG CNS deploys direct-to-chip cooling for Naver Cloud at data center in Goyang City, South Korea” 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 · Aug 27, 2026

Lambda closes $926 million senior secured term loan B facility, backing GPU deployment for an investment-grade customer

The change 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 · medium confidence · Aug 27, 2026

Data center load made up 9% of PJM wholesale costs so far in 2026: market monitor

The project described in “Data center load made up 9% of PJM wholesale costs so far in 2026: market monitor” adds prospective compute capacity, but permitting, power procurement, construction, and deployment timing prevent it from increasing immediately available GPU supply.

Market effect uncertain · Medium term · medium confidence · Aug 27, 2026

AWS to deploy 2 million additional Nvidia GPUs

The source for “AWS to deploy 2 million additional Nvidia GPUs” does not yet establish a supported change in compute supply, demand, price, availability, power delivery, financing, or deployment timing.

H100 GPU Pricing Report: July 2026

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

Previous issue: Market Brief: Thursday, August 27, 2026

Rum Group signs up to $13.7bn lease agreement with unnamed US cloud customer