From 50 to 1,500 evaluations a day: JLR makes AI production-ready for aerodynamic engineering
- Customer
- JLR
- Industry
- Automotive
- Application
- Vehicle aerodynamics
- Outcome · Design iteration
- From 50 to 1,500 design evaluations per day
Overview
At NVIDIA GTC 2026, JLR showed how Neural Concept’s Engineering Intelligence platform now operates inside its aerodynamic engineering, connecting design, simulation and decision. With results aligned with CFD and wind tunnel validation, JLR raised design evaluations from around 50 to around 1,500 per day.
Challenge
Proving that AI works is no longer the hard part. Scaling it is.
Aerodynamics shapes a vehicle’s efficiency, range, stability and refinement. As programs grow more complex, driven by electrification, evolving regulation and time-to-market pressure, engineering teams need insight earlier and across a broader range of design possibilities.
In the traditional process, that insight arrives in steps: engineers modify the geometry, launch simulations and assess results once the data becomes available. For JLR, the question was no longer model performance, but how to embed AI across real engineering systems with the robustness and traceability that production engineering demands.
Solution
Design, simulation and decision, in one continuous loop.
Neural Concept’s Engineering Intelligence platform acts as the orchestration layer across JLR’s design–simulation–decision process. Deployed inside established CAE environments and aligned with validated simulation datasets and governance frameworks, it brings design exploration, performance prediction and decision support together, with AI models trained on more than 20,000 simulations and up to 1 billion data points per case.
Aerodynamic insight now arrives as the geometry evolves, not only at simulation checkpoints. CAE and design teams work together toward optimal vehicle configurations. Through NVIDIA Omniverse, aerodynamics, design and leadership teams explore predictions and feature-level sensitivities in an interactive digital prototype, and align on design directions. None of this replaces simulation: robust simulation foundations are what make the gains reliable.
At NVIDIA GTC 2026, Chris Johnston of JLR explained how AI-enabled aerodynamic engineering increased evaluations from roughly 50 to roughly 1,500 per day, as JLR moves toward AI engineering across additional domains.
Impact
From 50 to 1,500Design evaluations per day
Presented on stage at NVIDIA GTC, the results show AI working as part of JLR’s aerodynamic engineering, not beside it. Development is shifting from sequential execution toward continuous, data-driven decisions, with the repeatability and traceability needed to scale across vehicle programs.
What’s next
For JLR, the next challenge is scaling AI across engineering systems, which takes trusted infrastructure and integration layers, beyond the models themselves.