ACTIVATE AI brings Kubernetes, cloud, and on-premises GPU resources and model access under one control plane, with quotas and cost tracking across Kubernetes, virtual machines, and cloud environments.
Technical features
- Kubernetes support: connect existing clusters, on premises or in the cloud, and use them from the same environment that runs remote desktops, HPC jobs, and virtualized workloads.
- AI gateway: every model request from chat, the CLI, PW Code, or an external tool passes through the gateway, which checks access and meters tokens against an allocation. Models include frontier APIs under the customer's keys, Azure OpenAI and Amazon Bedrock, and self-hosted vLLM or Ollama.
- GPU providers: reservations at CoreWeave, Lambda, Voltage Park, and the major clouds connect under one login. Customers bring their own accounts, or Parallel Works sources and runs a managed cluster.
Kubernetes with ACTIVATE
ACTIVATE creates a namespace for each group and the RoleBinding that gives the group access. It sets compute, memory, GPU, and storage quotas per namespace, configures NVIDIA MIG profiles on nodes, and tracks consumption for chargeback and budgets.