CrowdStrike & NVIDIA develop real-time AI agents for cyber defence

CrowdStrike and NVIDIA have announced an expanded partnership to develop autonomous, always-on AI agents aimed at providing real-time cybersecurity across cloud, data centre, and edge environments.

This collaboration introduces agents built with CrowdStrike's Charlotte AI AgentWorks, leveraging NVIDIA's Nemotron open models, NeMo Data Designer synthetic data, Nemo Agent Toolkit, and NIM microservices. The joint effort seeks to advance the deployment and operation of AI-driven defence mechanisms specifically designed for critical infrastructure protection.

Real-time AI defence

According to the companies, the new initiative will expand their ongoing efforts to build, power, and secure what they refer to as the agentic ecosystem. The ultimate goal is to deliver autonomous, real-time AI agents with continuous learning capabilities that adapt to ongoing security challenges in real world environments.

AI is transforming cybersecurity, and defenders need speed and edge intelligence to outpace the adversary. Addressing AI-driven cyber threats requires AI to protect systems from the speed and volume of attacks, and we're working with NVIDIA to deliver autonomous, AI agents that learn continuously to defend the critical infrastructure powering the global economy.

That was the view expressed by George Kurtz, Chief Executive Officer and Founder of CrowdStrike.

Cybersecurity in the era of AI demands intelligence that thinks at the speed of machines. Together with CrowdStrike, we're building real-time, AI-driven security agents that defend cloud, data centre, and edge infrastructure - protecting the systems that power our economy and national security.

Jensen Huang, Founder and Chief Executive Officer of NVIDIA, highlighted the importance of real-time, AI-supported security measures at multiple layers of digital infrastructure.

Extending protection to the edge

The companies noted that by bringing agents built with Charlotte AI AgentWorks and NVIDIA technology to the edge, organisations can place continuously learning AI models closer to their data sources. This is intended to enable more immediate threat detection and response, particularly within data centres and controlled environments.

The approach makes use of training NVIDIA Nemotron open models with data curated by CrowdStrike experts, facilitated via the NVIDIA NeMo Data Designer. This process aims to allow customers to fine-tune and optimise the models for deployment in their own environments using CrowdStrike's Agentic Security Platform.

According to CrowdStrike, the integration allows enterprises to scale their security operations, benefiting from improved detection accuracy and the ability to respond to threats locally. This may prove useful for organisations seeking to maintain control over sensitive data and comply with regional regulations regarding data sovereignty.

Unified security pipeline

CrowdStrike's Agentic Security Platform, which includes Falcon LogScale, Onum, and Pangea, will be integrated with NVIDIA's accelerated computing and CUDA-X libraries. The companies state that this provides a unified telemetry pipeline, delivering high-fidelity real-time security data to defenders.

This architecture will enable enriched telemetry to be fed directly into locally hosted AI models and agents, all optimised using the NVIDIA NeMo Agent Toolkit. Operating at the edge, these systems are designed to learn safely and act within defined enterprise guardrails while reasoning accurately about evolving threats.

Government and compliance focus

CrowdStrike has also indicated it will support the latest NVIDIA AI Factory for Government reference design, which offers guidance for deploying AI agents in sectors subject to high security standards, such as federal and high-assurance organisations. The support is expected to allow enterprises to manage multiple AI workloads on-premises and within hybrid cloud settings, while meeting stringent requirements for regulated industries.

Both companies assert that their expanded partnership is intended to unify continuous learning, real-time intelligence, and machine-speed defence, as the technologies are aimed at protecting digital infrastructure underlying both economic activity and national security.