TrendAI and Nvidia innovate with simulation-driven security for AI datacentres

TrendAI and Nvidia have partnered to incorporate simulation tools into the security design of AI-focused datacentres, enabling organisations to assess and validate protections before physical deployment, thus enhancing safety and efficiency.

TrendAI and Nvidia have combined simulation and security tools to let organisations vet protections for AI-focused datacentres before they are built, aiming to move threat controls into the design phase of so‑called “AI fact...

The integration connects TrendAI’s security capabilities with Nvidia’s DSX Air, a cloud-hosted environment that creates digital twins of data centre infrastructure so teams can model network layouts, power and cooling, and operational behaviour in a virtual setting. According to the announcement, security teams will be able to run end‑to‑end assessments within those simulations to measure how controls affect performance and resilience during a planned build, reducing the need for costly physical labs during early proof‑of‑concept work.

Rachel Jin, Chief Platform and Business Officer at TrendAI, said: “True innovation requires the best of both worlds: AI plus cybersecurity. Securing AI at scale isn’t something you can bolt on later. It requires a purpose-built foundation. By empowering customers to check the impact of security on digital twin simulations, we’re pioneering a new Secure AI Factory approach.”

Nvidia framed the work as part of a broader effort to validate large AI datacentre designs before rollout. Amit Katz, VP of Networking at Nvidia, said: “NVIDIA is focused on simplifying and accelerating the design and validation of next generation AI factories. Working with partners like TrendAI provides organisations with the visibility to detect threats across the entire stack, from cloud to endpoint, so they can focus on scaling AI without compromising security.”

The collaboration comprises two principal components. The first is an agentless host visibility capability, deployed on Nvidia BlueField data processing units and integrated with the Nvidia DOCA Argus framework, which TrendAI says captures file activity, network interfaces and process information while analysing traffic with its threat intelligence. The second element is simulated network defence using TrendAI TippingPoint, enabling virtual patching and network detection and prevention trials inside the digital twin so customers can see whether mitigations would introduce operational side effects prior to hardware installation.

TrendAI cited IBM research showing more than one in 10 organisations suffered data breaches involving AI models or applications in the previous year and said firms without AI or automation faced materially higher average breach costs, using the research to underline gaps such as access‑control failures and supply‑chain exposure from compromised apps, APIs and plug‑ins. According to TrendAI, red‑team exercises modelled in the simulated environment can reproduce recognised adversary behaviours, using frameworks such as MITRE, to help evaluate configuration and posture before systems go live.

Industry observers say the move reflects a larger shift in datacentre engineering and security practice toward greater reliance on simulation and automation as infrastructure grows more complex. Digital twin workflows give security and compliance teams a controlled way to document and demonstrate control effectiveness, a capability that matters when deployments must meet tight regulatory or audit standards in sectors such as finance, healthcare and government.

Nvidia’s DSX Air sits alongside the vendor’s Omniverse DSX Blueprint, a reference architecture for gigawatt‑scale AI facilities that Tom’s Hardware reported maps every aspect of an AI factory from compute and networking to power and cooling. According to that coverage, the reference design highlights the enormous energy footprints such sites can generate, likening single facilities’ power requirements to those of a nuclear reactor. The blueprint and DSX Air together underscore Nvidia’s strategy of using digital twin simulation to plan and optimise both the physical and security dimensions of next‑generation AI deployments.

While vendors present the integration as a route to reduce deployment risk and speed validation, independent testing and operational experience will be needed to confirm how well simulated controls translate into sustained protection in live environments. The companies say customers are expected to employ the tools during early planning and validation to provide evidence of control effectiveness ahead of procurement and build‑out.

As enterprises stitch together increasingly intricate AI stacks that incorporate third‑party models, plug‑ins and APIs, the ability to assess security impact before committing to hardware and wide release is likely to become a more prominent part of procurement and compliance workflows. TrendAI and Nvidia say their combined offering is intended to give organisations an earlier, more measured way to discover weaknesses and validate mitigations as AI systems migrate into business‑critical roles.

Source: Noah Wire Services