Comment on page
Shared Responsibility Model
How CoreWeave and its clients platform use responsibilities
A Shared Responsibility Model is a tool to help communicate the distinction between the responsibilities held by CoreWeave and the responsibilities held by our clients when using the CoreWeave platform. The diagram below illustrates the basic aspects of CoreWeave Cloud for which the client is ultimately responsible, as well as those for which CoreWeave itself is ultimately responsible. You can read more about each listed aspect under the diagram.
Security and compliance on CoreWeave operates as a shared responsibility between CoreWeave and our customers. This shared model relieves the operational burden of customers while CoreWeave operates, manages, and controls its components.
The Shared Responsibility Model below delineates the responsibilities that platform customers ("Client") maintain, versus those that CoreWeave itself maintains.
- Client: A CoreWeave Cloud customer
- End user: A user connecting to the Client's services; the Client's customers
Any aspects of using CoreWeave Cloud that are within the Client's absolute control are the responsibility of the Client.
The following are elements that fall under the user's responsibility.
- Client software and applications and client-side data: the software and applications users choose to run within their containers, as well as all client-side data, is the sole responsibility of the user
- Any guest Operating System used on Virtual Servers: users are responsible for guest Operating Systems and network configurations, including firewall configurations
The following are all Client responsibilities described in depth.
CoreWeave’s responsibilities begin at the physical level across the nationwide data centers housing CoreWeave hardware.
It is our responsibility to ensure that our physical data centers are always secure, and that power is always on and flowing to where it is needed.
Storage, compute power, and networking infrastructure are all the responsibilities of CoreWeave. Our high-performance NVIDIA GPUs and CPUs all fall under the care of our dedicated teams. Networking infrastructure is also the responsibility of CoreWeave’s infrastructure teams.
CoreWeave runs a multi-tenancy Kubernetes cluster on bare metal in order to ensure that our clients receive the lowest latency, the fastest compute times, and the most reliable service in the industry. Our responsibility model diverges between the core operations of Kubernetes and the client software running within it. CoreWeave assumes full responsibility for the container infrastructure - including all components that generate Virtual Servers - within the cluster.
The following are all CoreWeave responsibilities described in depth.