Skip to main content
Standing up a GPU fleet at scale is not a hardware problem. It is a path: securing power, finding a location that can carry the load, financing the build, procuring the hardware, installing it, and managing the fleet once it is live. Each of those sits with a different party, and a compute-heavy enterprise coordinating them separately carries the risk on all of them. Acasia takes that whole path on as a single team, and stays on to operate the result. Every deployment comes with Acasia Cloud, the web platform where your team sees what the fleet is doing and gets alerted when it needs attention.

​
What Acasia covers

​
Two ways to buy GPU capacity

Purchases made through the sales team have far more room for customization, and in exchange require a larger and longer commitment. If you are planning a smaller deployment, or you want to get started quickly with an inference endpoint, rent from the Acasia Marketplace instead. Capacity there is self-serve and available immediately after checkout.

​
What you see in Acasia Cloud

Acasia Cloud is one platform, but what it reports depends on how you hold the hardware. The control plane is software installed on the GPU hardware itself. It is required on every Marketplace rental and installed automatically at checkout. See The control plane.

​
Listing your own hardware

Self-serve listing of GPU hardware you own on the Acasia Marketplace is coming soon. Contact acasia.com/contact to be notified when listings open.

​
Where to go next

Bare Metal

Reserve dedicated Acasia GPU hardware through sales.

Acasia Cloud

Telemetry, alerting, and cluster management for your fleet.

Marketplace

Rent available GPU capacity on your own schedule.