NVIDIA has selected engineering firm Jacobs to deploy a digital twin at one of its largest US AI research and development facilities. The three-year project aims to bring live facility data and simulation together, with Jacobs describing a path towards more autonomous operations as AI data centres grow harder to manage.
NVIDIA Watch analysis
What happened
Data Centre Magazine reports that the deployment will use NVIDIA Omniverse libraries to combine engineering models, operational technology and live facility data. Jacobs says the system is intended to help operators model changing workloads and manage power, cooling and other infrastructure needs.
The listed applications include forecasting energy demand, balancing power across AI compute clusters, detecting liquid-coolant leaks, scheduling predictive maintenance and practising responses in simulated scenarios. Jacobs also describes AI agents that could use facility data to flag potential problems, recommend actions and, increasingly, act with less constant human input. Those are the company’s stated aims for the system, not evidence that every function is already running autonomously.
Why it matters
AI infrastructure depends on more than chips and models. Power, cooling and reliable operations are part of the same equation, especially when workloads shift quickly. A digital twin can give operators a virtual place to test responses before making changes in a live facility.
The more consequential ambition is to move from seeing and simulating a data centre to letting software help manage it. That could give operators earlier warnings and quicker responses, but it also makes the quality of the underlying data and the boundaries on automated actions important.
Our read
This is a concrete example of AI infrastructure being managed with a digital twin, rather than another prediction that data centres will somehow run themselves. Jacobs lays out practical targets, from coolant-leak detection to workload modelling. The test is whether those tools make operations more reliable and efficient in practice, and how much authority the system gets as it grows more autonomous.
What to watch
- Which capabilities are deployed first at NVIDIA’s facility.
- Whether Jacobs or NVIDIA share measured results on energy use, uptime or maintenance.
- How operators review and control actions recommended or taken by the AI agents.
Discussion spark: Should AI agents be allowed to act on data-centre operations when they can respond faster than people, or should every consequential change require an operator’s approval?
Sources and evidence
- How Jacobs' Digital Twin Deployment for NVIDIA Advances AI – Data Centre Magazine (5 October 2026, 11:16 UTC)
Independent WittyWires tracker for public updates about NVIDIA. Not affiliated with or endorsed by NVIDIA; this is not an official account.