Akamai and NVIDIA have expanded their security collaboration to bring Zero Trust enforcement inside AI factory infrastructure, targeting the growing threat surface created by agentic AI workloads and large language model-driven attacks.

The two companies announced that Akamai Guardicore Segmentation will be integrated with the NVIDIA Vera BlueField-4 STX storage architecture, powered by the NVIDIA DOCA software platform. The integration is designed to apply workload-aware segmentation, agent behaviour monitoring, and threat containment directly at the infrastructure layer — without drawing on the GPU, CPU, or storage cycles that AI workloads require.

Security Without the Speed Penalty

AI factories — the high-performance compute environments that underpin model training, inference, and autonomous agent operations — have historically faced a trade-off between performance and security. Host-based security tools impose overhead that accelerated computing environments cannot absorb.

The Akamai–NVIDIA integration addresses this by offloading security enforcement onto BlueField-4 silicon via NVIDIA DOCA, moving threat detection and policy enforcement into the data path at line speed rather than onto the host. The arrangement means security functions run inside the infrastructure fabric itself, independent of the workloads they protect.

“In environments where every clock cycle matters, traditional host-based security tools behave like a speed bump on a racetrack,” said Ofer Wolf, Senior Vice President, Enterprise Security at Akamai. “By moving workload-aware segmentation onto NVIDIA Vera BlueField-4 STX and DOCA, we are enforcing Zero Trust at the speed of AI workloads themselves, helping organisations contain threats before they spread across high-performance environments.”

How the Integration Works

The combined solution operates across four layers. Akamai Guardicore Segmentation continuously maps communication relationships across data centres, cloud, Kubernetes clusters, and edge systems using an agentless architecture that observes AI workloads — including training pipelines, inference services, and orchestration platforms — without interfering with them.

Workload identity, application context, and runtime behaviour are then used to define explicit communication policies. A preprocessing node may be permitted to access a dataset and a training service but nothing beyond that scope; research environments are explicitly separated from production inference. NVIDIA DOCA applies those policies in BlueField-4 silicon at line speed, enforcing segmentation, telemetry, anomaly detection, and isolation of compromised systems within the infrastructure fabric.

When a workload is compromised, the blast radius is constrained to a small, identified segment of the environment, leaving the rest of the AI factory operational.

Availability and Context

The expanded collaboration builds on an architecture agreement the two companies introduced in February 2026. Akamai Guardicore Segmentation integrated with NVIDIA BlueField and DOCA is expected to be available in the second half of 2026. Integration with the NVIDIA Vera BlueField-4 STX storage and infrastructure partner platforms is targeted for the first half of 2027.

The announcement was made from Singapore, reflecting ongoing enterprise AI infrastructure investment across the region as organisations move from centralised cloud deployments to distributed, agent-driven compute environments.