Nutanix launches its own software stack for companies looking to leverage AI. It’s platform independent, ensuring you don’t have to sign up with a hyperscaler.
Nutanix starts GPT-in-a-Box. That’s a nice name for a solution that doesn’t actually come with the box. After all, GPT-in-a-Box is a software solution. Nutanix describes it as a full-stack, software-defined AI platform with services to quickly get AI initiatives off the ground.
Nutanix provides a platform that ties complex hybrid multicloud environments together. Nutanix customers can theoretically run workloads anywhere, on-premises or across cloud providers, while still working from a single, overarching management plane. GPT-in-a-Box is now bringing this concept to AI development.
Support
To achieve this, the new solution combines support for GPU servers, file and object storage, and open source software such as PyTorck and KubeFlow. Nutanix AHV hypervisor is Nvidia AI Enterprise 4.0 certified and compatible with Nvidia GPUs. AMD GPU support will follow later. The GPT models Llama2, Falcon andmosaikML are supported right from the start.
Nutanix aims to reconnect with organizations operating in a complex hybrid IT environment. Do you have some data in the cloud and some in another cloud, and would you rather run AI workloads in your own data center? Then the Nutanix platform makes that possible. In addition, Nutanix states that you have more control over exactly what happens to your data. The new solution is available immediately.