NVIDIA's BlueField-3 data processing units now run FortiGate VM directly on-chip, enabling organizations to enforce firewall, segmentation, and zero-trust policies at line rate without degrading GPU workload performance.1 The integration offloads security infrastructure from compute resources, addressing what NVIDIA's Kevin Deierling describes as the need for "an entirely new class of secure, accelerated infrastructure" to support AI factories.2
The deployment reflects broader capital allocation shifts as enterprises build infrastructure for AI workloads. Hyperscale operators are converting legacy facilities into GPU-optimized data centers while scaling cloud platforms for accelerated computing. Enterprise AI adoption by companies like Adobe and UiPath, plus emerging agent-based systems, are driving infrastructure investment beyond traditional data center capacity.
Hyperscale Data projects 2026 as a critical expansion year, with growth driven by organic AI infrastructure buildout and digital platform operations.3 CEO Milton Ault III noted the company expects full-year contribution from Ballista following its Chapter 11 restructuring, stating that multi-year investments in "infrastructure, software platforms and digital ecosystems" are positioned to generate scaling revenue while improving operating efficiency.4
The NVIDIA-Fortinet integration specifically targets isolation requirements for multi-tenant GPU environments. By running security functions on dedicated DPU silicon rather than shared CPU or GPU resources, operators can maintain performance guarantees for compute-intensive AI training and inference workloads. This architectural approach enables cloud providers to offer secure GPU instances without the performance penalties typically associated with virtualized security appliances.
Infrastructure investment patterns show enterprises prioritizing integrated security rather than retrofitting protection onto existing GPU deployments. The shift toward purpose-built AI infrastructure with embedded security reflects operational learnings from early AI factory deployments, where security overhead constrained GPU utilization rates and increased total cost of ownership.
Sources:
1 Yahoo Finance, "Crypto Currents: SEC, CFTC sign MOU for Joint Harmonization Initiative" (March 14, 2026)
2 Yahoo Finance, "Fortinet Delivers Isolated Infrastructure Acceleration for the AI Factory with NVIDIA" (December 16, 2025)
3 Milton Ault III, via Yahoo Finance
4 Kevin Deierling, via Yahoo Finance
5 Kevin Deierling, via Yahoo Finance
