Jacobs has won a three-year software agreement to deploy a digital twin at one of NVIDIA’s US AI research and development facilities. The system is intended to help operators model workloads, power and cooling, bringing practical infrastructure management into the AI story.
NVIDIA Watch analysis
What happened
Jacobs says its Data Center Digital Twin will combine engineering models, operational technology and live facility data in a real-time environment. Built on NVIDIA Omniverse libraries, it is intended to support operational planning and facility performance.
The company lists use cases including power-load balancing, energy forecasting, liquid-coolant leak monitoring, predictive maintenance and operator training. Teams will be able to model changing AI workloads and examine their implications for power availability and cooling. Jacobs announced the agreement on 1 October. Its announcement does not name the facility or disclose the contract value.
Jacobs also describes AI agents that could learn from facility data, predict conditions and recommend or optimise operations with increasing autonomy as a future direction for its digital twin work. That is not the same as saying those agents are already operating at NVIDIA’s facility.
Why it matters
AI capacity depends on more than accelerators. Power demand, cooling and maintenance all have to keep pace as workloads change, and the cost of getting those decisions wrong can be substantial. A live digital model could give operators a place to test scenarios before changing the physical facility.
The announcement offers a concrete example of NVIDIA’s Omniverse being used in AI infrastructure planning, not just visualisation. The eventual value will depend on how well the model reflects the facility and whether it helps operators make better decisions in practice.
Our read
This is a notable infrastructure contract with a useful set of intended applications, rather than evidence that AI agents are already running an AI factory. The near-term story is simulation and operational visibility; the more autonomous system is a direction Jacobs says it is pursuing. Keep those two timelines separate, and the announcement becomes more interesting, not less.
What to watch
- Whether Jacobs or NVIDIA identifies the facility or discloses measurable results from the deployment.
- How the digital twin handles live operational data and changing workloads.
- Whether the proposed AI-agent capabilities move from future direction to deployed, operator-supervised features.
Discussion spark: For managing AI data centres, should operators trust a digital twin to recommend changes to power and cooling, or should it remain a simulation tool until its predictions are proven in live facilities?
Sources and evidence
- Jacobs to deploy data center digital twin for NVIDIA’s R&D facility – Jacobs (1 October 2026, 01:18 UTC)
Independent WittyWires tracker for public updates about NVIDIA. Not affiliated with or endorsed by NVIDIA; this is not an official account.