Serve many tenants from one sovereign AI estate

Governments and national champions share GPU estates serving ministries, enterprises, and researchers. Sovereignty for each tenant depends on where requests are processed, what data leaves the jurisdiction, and what can be proven later.

Sovereignty has to hold in the request path


A sovereign AI estate is judged on more than where its servers sit. Tenants want to know where their requests are processed, what leaves the country, and how the operator would prove it. Those answers depend on controls in the traffic path and at the border of the estate, not on a rack diagram. F5 provides those controls on the F5 Application Delivery and Security Platform.

Keep requests and data in-country
In-country steering, in-country S3 data paths, and per-tenant egress policy.

Separate tenants in the network
Ministries, enterprises, and researchers each get their own routing domain.

Protect the estate at its border
DDoS protection, firewalling, and DNS for public AI services and APIs.

Keep evidence for audits
Per-tenant firewall logs show what crossed each border and which policy applied.

F5 helps sovereign AI operators

Egress and data residency

PRODUCT

BIG-IP Next for Kubernetes

Per-tenant routing domains and egress policy with stable source addresses.

PRODUCT

BIG-IP DNS

Global server load balancing that steers users to in-country sites.

PRODUCT

BIG-IP Next for Kubernetes deployed on NVIDIA BlueField DPUs

Tenant enforcement on the DPU, separate from the host operating system.

PRODUCT

BIG-IP LTM

Load balancing and failover across in-country S3 storage endpoints.

Tenant isolation

PRODUCT

BIG-IP Next for Kubernetes

Per-tenant routing domains and egress policy with stable source addresses.

PRODUCT

NGINX Gateway Fabric

Gateway API implementation for application traffic inside tenant clusters.

PRODUCT

BIG-IP Next for Kubernetes deployed on NVIDIA BlueField DPUs

Tenant enforcement on the DPU, separate from the host operating system.

Border protection at scale

PRODUCT

BIG-IP DDoS Hybrid Defender

DDoS protection for public AI services and APIs, in-country.

PRODUCT

VELOS

Chassis platform for high-throughput border services.

PRODUCT

BIG-IP Advanced Firewall Manager

Network firewalling at the border of the AI estate.

PRODUCT

rSeries

Appliances for border services at smaller sites.

Evidence and AI security

PRODUCT

BIG-IP Next for Kubernetes

Per-tenant routing domains and egress policy with stable source addresses.

PRODUCT

F5 AI Red Team

Adversarial testing of models, apps, and agents before release.

PRODUCT

F5 AI Guardrails

Runtime protection for prompts and responses in tenant AI apps.

PRODUCT

Distributed Cloud API Security

Discovery and protection for the APIs that expose models.

Resources

FAQs

A sovereign AI factory is AI infrastructure built, operated, and governed inside one jurisdiction, so that data, models, and processing stay under that jurisdiction's laws and control. Most serve many tenants at once, such as government agencies, regulated enterprises, and researchers, from a shared GPU estate. This makes control over outbound traffic and evidence of what crossed each boundary as important as where the data center sits.

To keep AI data inside a jurisdiction, control the paths it can take. Steer users and requests to in-country sites with global traffic management, process them on in-country GPU clusters, keep the storage path in-country, and give each tenant an egress path that is policy-controlled and logged by destination, so external calls by agents and workloads can be allowed or denied. Then keep the logs, because regulators ask for evidence rather than intent.

No infrastructure product makes an organization compliant on its own. Compliance depends on the organization's policies, processes, legal review, and audits. Infrastructure provides controls that support a compliance program, such as tenant isolation, egress policy, and logs that show what crossed each boundary, and those controls make the evidence easier to produce.

Yes, F5 can run in disconnected or air-gapped sovereign estates. BIG-IP software and hardware have long run in isolated networks. For BIG-IP Next for Kubernetes and F5 AI security products, confirm with F5 which telemetry, licensing, and update connections each deployment needs, so the estate's operator can design the disconnected or brokered paths its policy requires.

The operator operates the software in a sovereign estate. F5 products deploy into the operator's own Kubernetes clusters, hardware, and management systems inside the jurisdiction, and the operator controls configuration, policy, and logs. F5 provides support under the terms the operator and its regulator agree to.