PHISON HCI

One control plane for compute, storage, and GPU

Phison HCI is the infrastructure management layer for the broader Phison AI Data Platform. Planned capabilities bring applications, compute, GPUs, VMs, storage, access, monitoring, and cost information into a common management environment across servers.     

This page describes the expanded platform capabilities coming soon. For the first AIDP offering, explore GenBI on one server.

Reliability and performance as the platform grows

Resource management, supported failover, and AI cache reuse address different parts of operating private AI. The broader platform will bring these capabilities together for supported multi-server deployments.  
Recovery and maintenance 

Health monitoring helps keep service running. Recovery time depends on the workload and deployment design.

More work from the same GPUs

Reuse eligible KV cache through Pascari aiDAPTIV™ to reduce repeated processing and free GPU capacity for additional work.

Grow across servers

Add servers as your workloads grow. The platform manages compute, storage, and cache across them.

What Phison HCI will handle

Every AI workload needs infrastructure behind it. Phison HCI will coordinate applications, compute, GPUs, VMs, storage, access, and operations across supported servers, with a common management view.
Applications
Compute and GPUs
VMs and storage
Workspace and access
Operations and monitoring

Less to manage, more to run

Coming soon. Six planned capability areas will support the broader platform. Supported features and configurations will be defined by release.
Simplified AI application deployment

Prebuilt Helm charts and deployment templates for supported models and applications, reducing manual setup and the Kubernetes expertise required for routine deployment.

A common view of CPU, GPU, and storage resources, with centralized monitoring and operational visibility.

Planned support for compatible GPU and storage nodes, with resource pooling across nodes where the configuration supports it.

Integration with aiDAPTIV Cache Memory and middleware to reuse eligible model context. Less repeated processing can support more simultaneous AI activity on the available GPUs.

Resource and permission controls for teams and workloads using shared infrastructure.

Monitoring for performance, utilization, and system health, with cost tracking for supported resources and services.

Ready to talk about Phison HCI?

Discuss the broader HCI capabilities coming soon. For a current project, start with Phison AI Data Platform: GenBI.