CrowdStrike (NASDAQ: CRWD) announced an expanded collaboration with Nvidia to extend AI security from software to silicon, according to a company press release issued at Nvidia GTC 2026.

CrowdStrike is integrating its Falcon Guardian AI Detection and Response (AIDR) solution with Nvidia's OpenShell secure runtime software. The company also plans a future integration with Nvidia Sentry for in-silicon policy enforcement on BlueField-4 data processing units (DPUs).

The integration with Nvidia OpenShell, whose reference architecture was announced at the same event, is intended to connect AI agent prompts, model interactions, and runtime activity with CrowdStrike's broader Falcon platform security context. The planned Nvidia Sentry integration would add out-of-band monitoring and hardware-level enforcement to detect and contain unsafe agent behavior beyond the software runtime.

"AI agents have already created a new attack surface, and guardrails alone won't stop an adversary," said Daniel Bernard, chief business officer at CrowdStrike. "Together with Nvidia, we're setting the standard for securing autonomous AI, from detecting threats in software to stopping them at the silicon level."

"Securing agents in production requires protection from the runtime to the underlying infrastructure," said Justin Boitano, vice president of enterprise at Nvidia. "Together with CrowdStrike's AI Detection and Response capabilities, we're working to bring comprehensive security to AI agents at scale."

Autodesk's chief trust officer, Sebastian Goodwin, also commented on the collaboration, noting that Falcon Guardian provides visibility into agent activity and access. "Extending that control deeper into the infrastructure helps customers of every size trust the AI behind their work," he said.

The Nvidia Sentry integration remains a planned feature and has not yet been released. CrowdStrike noted that unreleased features reflect current plans only and are subject to change.