About JouleCloud
When an AI request can run in more than one place, JouleCloud uses live electricity prices and GPU availability to choose where it runs.
Electricity prices and available power change by location and hour.
Some AI requests can run on different GPUs without changing the service.
AI work can move. Electricity prices change by place and time.
Many AI requests can run on GPUs in different regions, and some can run at flexible times. Electricity prices and grid conditions vary by location and hour. JouleCloud uses those differences when routing requests.
Routing with live electricity prices
JouleCloud uses live electricity prices plus GPU health and availability to choose among places that can run a request. Our research found that lower prices aligned with cleaner power in all 13 market-years we studied. Read the evidence →
Works with familiar APIs
The gateway supports OpenAI-style /chat/completions requests and Anthropic-style /v1/messages requests. Teams can connect through direct HTTP or compatible clients, while JouleCloud handles routing behind the endpoint.
Automatic health checks and retries
JouleCloud checks each GPU before routing work. If a selected GPU becomes unavailable, it retries on another ready GPU while the caller receives one response.
A transparent private preview
JouleCloud is in private preview. The model catalog shows what is available now, the status page reports live checks, and the pricing page lists current metered rates. The Security page explains the protections and compliance available today.
Talk with us
Email hello@jouledns.com or see the contact page. The founder writes at The First 5 Percent.