Market Brief: Friday, October 16, 2026

Today's lesson

Why Reasoning Models Cost More to Serve

Reasoning models generate long chains of thought before answering, multiplying output tokens — and output tokens drive inference cost. Why it matters today: Today's lesson connects the Model Benchmarks & AI Compute Economics track to the compute market readers are watching.

Term of the Day

Tokens per second per user

Tokens per second per user measures how quickly one user receives generated text. It differs from the total output of a server serving many people at once. A server can produce more tokens overall while each person gets a slower stream as the number of active users rises.

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

Value

$4.29 / H100-hr

Last Price Change

Oct 16, 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

83 / 100 score

Last change

Sep 17, 2026

Oct 15, 2026

Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models

The project described in “Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models” adds prospective compute capacity, but permitting, power procurement, construction, and deployment timing prevent it from increasing immediately available GPU supply.

Oct 15, 2026

NVIDIA Expands Advanced Technology Access for Europe’s Scientific Community Through Open Learning Resources, Collaborations and Compute

The source for “NVIDIA Expands Advanced Technology Access for Europe’s Scientific Community Through Open Learning Resources, Collaborations and Compute” does not yet establish a supported change in compute supply, demand, price, availability, power delivery, financing, or deployment timing.

Oct 15, 2026

Oracle renews attempts to secure natural gas pipeline approval for Project Jupiter data center - report

The project described in “Oracle renews attempts to secure natural gas pipeline approval for Project Jupiter data center - report” adds prospective compute capacity, but permitting, power procurement, construction, and deployment timing prevent it from increasing immediately available GPU supply.

Oct 15, 2026

DigiCo signs 71MW lease for Sydney data center with unnamed tech firm

The project described in “DigiCo signs 71MW lease for Sydney data center with unnamed tech firm” adds prospective compute capacity, but permitting, power procurement, construction, and deployment timing prevent it from increasing immediately available GPU supply.

Oct 15, 2026

Colt DCS and RMZ start work on third facility at Mumbai campus, India

The project described in “Colt DCS and RMZ start work on third facility at Mumbai campus, India” adds prospective compute capacity, but permitting, power procurement, construction, and deployment timing prevent it from increasing immediately available GPU supply.

H100 GPU Pricing Report: September 2026

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

Previous issue: Market Brief: Thursday, October 15, 2026

European Cities Deploy NVIDIA-Powered AI Agents to Run Smarter Urban Operations