Check Point Software Technologies (NASDAQ: CHKP) and NVIDIA have teamed up to launch what they’re calling AI Cloud Protect, a platform designed to protect the emerging “AI factory” infrastructure—those massive, GPU-driven data centers and enterprise clusters powering AI model development, inference, and agentic AI applications. The partnership marries Check Point’s long-standing cybersecurity expertise with NVIDIA’s hardware acceleration, aiming to close one of the biggest gaps in AI adoption: securing the pipeline without slowing it down.
At the heart of this announcement is the challenge enterprises are facing right now. AI adoption has been explosive—more than half of enterprise networks now deploy AI tools—but that speed has created dangerous blind spots. Check Point’s own telemetry suggests one in every 80 GenAI prompts exposes sensitive data, while Gartner reports nearly a third of organizations already suffered AI-related attacks ranging from prompt manipulation to infrastructure compromise. In other words, the AI security problem isn’t hypothetical—it’s here.
AI Cloud Protect tackles this with a “factory-level” mindset. Running on NVIDIA’s BlueField-3 DPUs, it moves the heavy lifting of security away from the CPU and GPU, which are fully consumed by AI training and inference, and offloads it to specialized accelerators. That means full-stack defense—network security to stop model exfiltration or data poisoning, and host-level runtime inspection through NVIDIA’s DOCA Argus direct memory access—without touching the performance envelope. For enterprises already spending millions building GPU clusters, this “zero performance penalty” positioning is crucial.
This is also forward-looking. The solution is validated on NVIDIA RTX PRO Servers today, but will extend to the forthcoming BlueField-4 generation, offering six times the compute power and double the network throughput. That sets up AI Cloud Protect to scale with the next wave of AI data centers. For customers, especially in financial services and sensitive IP-driven industries, the promise is being able to deploy large language models and agentic AI securely without sacrificing the speed that makes them commercially viable.
It’s worth noting this isn’t Check Point’s only foray into AI security. The company is weaving together a portfolio that spans infrastructure (AI Cloud Protect), applications (CloudGuard WAF with Lakera integration to block prompt injection and jailbreaks), and enterprise usage (Infinity GenAI Protect to monitor and govern how employees interact with GenAI tools). Together, that builds the case for an “AI supply chain security” narrative, something regulators and boards are increasingly demanding as AI becomes a critical enterprise dependency.
Strategically, this move positions Check Point as one of the first security vendors explicitly building for “AI factories”—a term NVIDIA itself has been pushing as shorthand for the data-center-as-factory model of AI production. By embedding directly into NVIDIA’s ecosystem and validating on RTX PRO Servers, Check Point is plugging into the hardware supply chain of the AI boom rather than sitting on the sidelines. It also sets them up to compete in a new market space where Palo Alto Networks, CrowdStrike, and others are only beginning to formulate strategies.
For NVIDIA, the partnership is just as important. By highlighting security frameworks like DOCA Argus alongside its BlueField acceleration, NVIDIA is signaling that AI infrastructure must be secure-by-design, not secured later as an afterthought. As AI workloads become mission-critical and regulated, enterprises won’t just buy raw compute—they’ll buy integrated stacks that meet performance and compliance requirements in one package. NVIDIA’s alignment with Check Point is a signal that it wants to lead this narrative, not leave it to security vendors alone.
The big question is adoption. Early pilots with financial services and World Wide Technology suggest traction in verticals where risk tolerance is low and compliance is paramount. But for enterprises at large, the urgency is rising quickly. With statistics showing prompt leakage, model theft, and AI-targeted attacks already widespread, the timing of this launch puts Check Point and NVIDIA in the right place as enterprises move from proofs of concept to scaled AI deployments.
This announcement is more than a new product. It’s a marker that the security conversation has officially moved inside the AI factory floor. Enterprises building LLM clusters and agentic AI systems can no longer assume their traditional defenses will hold. And with AI Cloud Protect, Check Point and NVIDIA are making the argument that if you want to build AI at industrial scale, you need industrial-scale security built in from the silicon up.
Leave a Reply