Hyundai Motor Group is taking a strategic detour in its race toward autonomous vehicles, choosing to deploy NVIDIA DRIVE-based systems in production cars starting in 2028 while delaying its own proprietary self-driving software to late 2029. The decision marks a significant shift in how the world's third-largest automaker approaches vehicle autonomy, prioritizing data collection and refinement over speed to market.

What Changed in Hyundai's Self-Driving Timeline?

Hyundai originally planned to launch its in-house autonomous driving system, called Atria AI, in late 2027. That target has now slipped two years to the second half of 2029. In the meantime, the company will introduce NVIDIA-based Level 2+ systems in the first half of 2028, followed by more capable Level 2++ systems in the second half of that year. Level 2+ systems handle highway driving assistance, while Level 2++ extends to complex urban driving scenarios, though both still require drivers to remain attentive.

Park Min-woo, president of HMG Advanced Vehicle Platform and chief executive officer of Hyundai's software affiliate 42dot, pushed back against the notion that the delay signals trouble. "We could release Atria AI sooner, but we strategically delayed the timeline," he stated. "Rather than rushing mass production to roll out a half-baked Level 2++, we decided to proceed with mass production using NVIDIA's technology while focusing resources on the final goal".

How Is Hyundai Using NVIDIA Technology to Build Better AI?

• Co-Design Partnership: Hyundai is not simply licensing NVIDIA's technology outright but instead co-designing the systems with the chipmaker, maintaining control over the integration and customization process.
• Data Flywheel Strategy: The company plans to collect real-world driving data from NVIDIA-equipped vehicles, use that data to train and refine its own Atria AI system in parallel, and then deploy improved models back to vehicles in a continuous cycle.
• Sensor Configuration: Initial NVIDIA-based vehicles will use eight cameras, ultrasonic sensors, and a single radar, deliberately omitting lidar to keep costs down while maintaining safety through sensor redundancy.

Hyundai's approach centers on what executives call a "Data Flywheel," a self-reinforcing loop in which vehicles collect real-world driving data, that data trains AI models, improved models are pushed back out to vehicles, and the cycle repeats at increasing scale. The company currently operates about 40 dedicated data-collection vehicles around the clock and expects to accumulate more driving data than competitors by 2033.

"The side that first completes the data loop from collection to learning, and from model training to model deployment, will have a competitive edge in automotive software and AI," explained Kwon Jeong-hyeon, head of Hyundai's Autonomous Driving Development Centre.

Kwon Jeong-hyeon, Head of Autonomous Driving Development Centre at Hyundai

Why Does Hyundai's Scale Give It an Advantage?

HMG sells more than seven million vehicles annually across approximately 190 countries, a global footprint that executives argue will accelerate the data-collection advantage once autonomous systems begin shipping in volume. This scale means Hyundai can gather real-world driving data from millions of vehicles simultaneously, feeding that information back into AI training pipelines faster than smaller competitors.

The company is also developing Vision-Language-Action (VLA) technology, which combines visual perception with language-based reasoning to improve decision-making in edge cases and make the system's behavior more explainable. This work is currently in simulation-based validation, with real-vehicle testing scheduled to begin between late 2026 and early 2027.

What Does This Mean for Hyundai's Autonomous Driving Future?

Hyundai's partnership with NVIDIA extends well beyond driver-assistance software, spanning autonomous driving more broadly, AI data centers, and humanoid robots developed by Hyundai-owned Boston Dynamics. During a media briefing, Hyundai demonstrated footage of an Atria AI-equipped Ioniq 6 testbed navigating dense Seoul traffic without driver intervention, including rush-hour conditions, high-density traffic with buses, and rainy conditions, handling scenarios such as parked vehicles, sudden cut-ins, unprotected left turns, and oncoming traffic on narrow streets.

Separately, Hyundai plans to deploy an Atria AI-equipped SDV Pace Car for real-world Level 4 validation testing in Gwangju by the end of 2026, in partnership with South Korea's Ministry of Land, Infrastructure and Transport. Level 4 systems permit hands-off driving under specific conditions, though no timeline has been given for commercialization. This testing will gather large-scale validation data from actual Korean road conditions rather than controlled test tracks, accelerating the refinement of Hyundai's proprietary system.

Park Min-woo, a former NVIDIA executive who joined Hyundai in January, has driven the deepening relationship with the chipmaker. His approach marks a shift from his predecessor, Song Chang-hyeon, who had prioritized internal software development before an abrupt departure in December. The strategic pivot reflects a broader industry trend in which traditional automakers are balancing in-house development with partnerships to stay competitive in a rapidly evolving autonomous vehicle market.