Stack8s provides a unified control plane that allows you to run your workloads across 20+ cloud providers, on-prem infrastructure, and edge environments - all as a single, seamless platform.
Instead of managing separate environments, accounts, and vendors, Stack8s pools compute into one global cluster, fully managed through a single interface.
Step 1
Create your First Sovereign Environment
Create your sovereign control plane by selecting the jurisdiction that aligns with your compliance and data residency requirements, such as the UK, EU, or KSA. stack8s then automatically provisions a highly available, fully managed control plane in that region, acting as your single point of control. From there, you can seamlessly attach compute from multiple cloud providers or on-prem infrastructure into one unified environment, where Stack8s handles provisioning, orchestration, networking, and billing. Workloads are deployed through a single interface while compute is dynamically pooled and scheduled across providers, allowing you to operate as one cluster with full policy control, sovereignty, and no infrastructure complexity.
Why this control plane matters
- High Availability (HA) by design - fully managed, resilient control plane with no single point of failure
- Sovereign by default - lives and operates entirely within the jurisdiction you define, ensuring full compliance and data boundary control
- Single source of truth - one control layer for policy, security, orchestration, and workload placement across all environments
- Powered by AiKCord Fabric - a secure mesh connecting 15+ cloud providers and private data centres into one unified system
- Pool your compute - attach CPU, GPU, and bare metal nodes from any provider, even behind firewalls, and operate them as worker nodes in a single environment - See Step 2
- True hybrid without complexity - run on-prem, sovereign cloud, and public cloud as one cluster with unified scheduling and control
Step 2
Add, Extend, and Pool Your Compute
Once your control plane is live, you begin attaching worker nodes within your chosen sovereign region, creating a unified pool of compute across providers and environments. For example, if your control plane is based in the UK, you can combine.. for example:
10 GPUs from Civo + 5 bare metal nodes from GCP + 2 GPU nodes from your own private data centre into a single environment. stack8s abstracts these into one logical cluster, even if they are behind firewalls or within private networks. Expand this cluster horizontally across any cloud and we make sure your data and compute never leaves the jurisdiction.
This model follows a data-first approach where compute is brought to where your data already resides, removing the need to move sensitive datasets and avoiding costly egress. Secure, private connectivity is automatically established so workloads and data can flow seamlessly across the pooled infrastructure, while staying within your defined sovereign jurisdiction boundary.
To integrate existing on-prem HPC infrastructure, stack8s uses the AiK Edge plugin, which enables your private data centre resources to join the environment as worker nodes. This setup can be arranged directly with the stack8s team based on your data centre location and connectivity requirements. Where the Stack8s fabric is available, connectivity of up to 400 Gbps across clouds and data centres is supported through a secure, encrypted mesh.
Step 3
Start Building from the Marketplace
Once your compute pool is ready, you gain access to the Stack8s Marketplace, where you can deploy fully managed services directly onto your environment with just a few clicks.
This is just a subset of what’s available, with new services continuously being added.
What you can deploy
Databases
- PostgreSQL, MySQL, MongoDB, CockroachDB
Vector Databases (AI/LLMs)
- Weaviate, Qdrant, Chroma
Data & Analytics
- ClickHouse, Elasticsearch, InfluxDB, TimescaleDB
Storage & Object Systems
- MinIO
Caching & Real-Time
- Redis
Backend Platforms
- Supabase, PocketBase
AI & Machine Learning
- Kubeflow
- Pre-configured AI pipelines
AI Ready - Out of the Box
- vLLM-powered inference for high-throughput model serving
- Hugging Face models ready to deploy instantly
- Ray Serve integration for scalable distributed inference and training
- No need to configure NVIDIA GPU operators, drivers, or runtimes - everything is fully managed and pre-configured
What this unlocks
- Deploy complex stacks in minutes, not weeks
- Run fully managed services on your sovereign multi-cloud infrastructure
- No DevOps overhead or manual configuration
- Seamless integration with your pooled CPU, GPU, and bare metal resources
The Outcome
You now have:
- A sovereign control plane
- A pooled multi-cloud compute layer
- A marketplace of ready-to-run services
All operating as one unified platform, no need to pay serlerately for your stack you own end-to-end.