AI-Native Aerodynamic Engineering in Practice - From NVIDIA GTC to Production

• Jaguar Land Rover (JLR) showcased at NVIDIA GTC how AI is already operating inside real aerodynamic engineering workflows, powered by Neural Concept and visualized through NVIDIA Omniverse.
• Across the industry, AI is moving from experimentation to production-grade engineering infrastructure inside CAE environments.
• In an on-site conversation at GTC, Chris Johnston (JLR) shared how these AI-driven workflows are being deployed in practice, what it takes to scale AI across engineering, and why robust simulation foundations remain critical.

AI-Native Aerodynamic Engineering in Practice

The primary challenge is no longer whether AI works, but how to scale it across engineering systems and workflows.

At Jaguar Land Rover, AI-driven aerodynamic workflows powered by Neural Concept and visualised through NVIDIA Omniverse enable engineers to explore design variations, access insights faster, and make more informed decisions earlier in vehicle development.

‍

Engineer Perspective: JLR on scaling engineering AI

At NVIDIA GTC, Neural Concept’s US General Manager, Thomas von Tschammer, spoke with Chris Johnston, Senior Technical Specialist at Jaguar Land Rover, about how AI-driven aerodynamic workflows are deployed in practice, and how teams are approaching the challenge of scaling AI across engineering.

Key takeaways: 

• Robust simulation foundations are necessary to ensure reliable gains when embedding AI into engineering workflows,
• AI-native workflows bring together CAE and design teams to work toward optimal vehicle configurations,
• Scaling AI across engineering workflows requires trusted infrastructure and integration layers, beyond models.

‍

JLR's Session at NVIDIA GTC

At GTC, Chris Johnston presented how JLR has deployed AI within aerodynamic engineering, showcasing both the results achieved in practice and the challenges of scaling these approaches across engineering, as teams move from experimentation toward production-scale deployment and organisational transformation:

• AI is production-ready for aerodynamic engineering at JLR, delivering results aligned with CFD and wind tunnel validation, enabled by models trained on industrial-scale simulation data (20,000+ simulations, up to 1B data points per case)
• This unlocks a step-change in design iteration and exploration, increasing the number of evaluations from ~50 to ~1,500 per day
• The challenge is no longer model performance, but scaling AI across engineering systems, with data fragmentation and manual workflows still limiting integration
• This shift is redefining engineering workflows, moving from sequential execution toward continuous, data-driven decision-making across the design–simulation–decision loop.

For a deeper dive, watch JLR's full NVIDIA GTC session: From concept to capability: how JLR is scaling AI surrogates across real CAE workflows

‍

Context: The Evolution of Aerodynamic Workflow

The sections below unpack how this type of workflow is made possible: combining simulation, AI, and visualization into a continuous engineering process.

The Traditional Aerodynamic Workflow

In vehicle development, aerodynamic performance plays a critical role in shaping efficiency, range, stability, and overall vehicle refinement.

Within a proven and validated development framework, engineers refine and optimize vehicle geometry through simulation and validation cycles, introducing design modifications, launching simulations, and assessing results once data becomes available. Aerodynamic insight is typically generated at defined stages of the development cycle to ensure traceability and consistency with program requirements.

As vehicle development programs become increasingly complex and data-intensive, driven by electrification, performance optimization, evolving regulatory requirements, and accelerating time-to-market pressures - including highly demanding domains such as F1 aerodynamics - engineering teams are placing greater emphasis on accessing insight earlier and across a broader range of design possibilities. In this context, the ability to leverage simulation and design data in a more integrated manner is becoming an important dimension of modern automotive engineering.

Integrating AI into the Aerodynamic Workflow

Embedding geometry- and physics-aware AI within aerodynamic workflows introduces a continuous analytical layer capable of generating aerodynamic performance predictions as vehicle geometry evolves. Rather than relying exclusively on discrete simulation checkpoints, engineers can receive rapid, directional feedback on how specific geometric variations influence efficiency, stability, and other key performance targets.

This expanded visibility supports structured exploration of the design space, enabling teams to evaluate trade-offs before committing to full simulation cycles and supporting more physics-informed design exploration.

As insight becomes available more continuously throughout development, iteration cycles become more responsive and engineering decisions more informed. In practice, integrating AI into the aerodynamic workflow enables a fundamental transformation in how design and simulation interact — shifting from sequential evaluation toward a more integrated, data-driven decision process.

Operationally, these AI models are deployed within existing engineering environments and aligned with validated simulation datasets and governance frameworks. This ensures that generated insights remain consistent with production standards, enabling repeatability, traceability, and scalable integration across vehicle programs.

A Production-Focused AI Collaboration at NVIDIA GTC

The gains enabled by AI-native aerodynamic workflows are not theoretical. At NVIDIA GTC, JLR (Jaguar Land Rover) took the stage to illustrate how AI is being embedded into real-world aerodynamic engineering processes, enabled by Neural Concept’s Engineering Intelligence platform, which acts as the orchestration layer across the design–simulation–decision process.

Within this framework, AI operates as an orchestration layer inside established CAE environments – integrating design exploration, performance prediction, and decision support directly into the engineering workflow. Aligned with validated simulation datasets, methodologies, and governance frameworks, the Neural Concept platform integrates directly into the design–simulation–decision loop that underpins modern vehicle development. Rather than functioning as a standalone analytical tool, AI becomes a structured component of the engineering infrastructure — enabling continuous performance insight while preserving production-grade robustness and traceability.

Alongside this AI integration, the session also highlighted the role of NVIDIA Omniverse in supporting cross-functional collaboration: Aerodynamic performance predictions and feature-level sensitivities can be explored within a photorealistic, interactive digital prototype or digital twin environment, allowing aerodynamics, design, and leadership teams to align around design directions and engineering trade-offs. In this way, Omniverse complements the AI-enabled workflow by enhancing how complex aerodynamic insight is communicated and reviewed across disciplines.

From Experimentation to Industrial Deployment

What was showcased at NVIDIA GTC reflects more than a single aerodynamic workflow. The example presented on stage illustrates how AI can already operate within real CAE environments today, supporting engineers as they explore design options and evaluate performance implications earlier in the development cycle. It signals how AI-native engineering capabilities are moving from isolated experimentation toward structured Industrial AI deployment within production environments.

What was showcased at NVIDIA GTC reflects a broader shift already underway across leading engineering organisations. As AI becomes capable of reasoning over geometry, physics, and real-world constraints, its role in engineering evolves — from an analytical support tool to an orchestration layer embedded within the core design–simulation–decision process. This shift enables engineering teams to access insight earlier, iterate more responsively, and make decisions grounded in continuous performance intelligence.

Within the NVIDIA ecosystem, and across leading industrial organisations, this progression marks a broader transition toward Engineering Intelligence: AI not as an add-on, but as an integrated capability shaping how products are conceived, evaluated, and delivered.

‍

‍