According to a source, Nvidia CEO Jensen Huang had a secret meeting with Mercedes-Benz and Hyundai executives at CES 2026. This encounter happened shortly after the massive chip manufacturer unveiled Alpamayo, its first autonomous vehicle technology.
Secret CES 2026 Meeting Sparks Industry Speculation
The timing of the meeting with Hyundai Executive Chair Euisun Chung has sparked speculation regarding Hyundai’s potential deployment of Nvidia’s new technology in its own vehicles.
Hyundai has been investing substantially in self-driving technology for years, but it still lags behind competitors like Tesla, which already has more sophisticated capabilities, according to a report by Korea JoongAng Daily. By the end of 2027, the Korean automaker hopes to equip its standard vehicles with Level 2+ self-driving capabilities.
Hyundai’s Autonomous Driving Ambitions
🤖 Nvidia–Hyundai Artificial Intelligence Collaboration
- Focus: Artificial intelligence for self-driving vehicles
- Technology: Nvidia Alpamayo autonomous vehicle platform
- Investment: 50,000 Blackwell GPUs planned
- Goal: Level 2+ autonomous driving by 2027
- Use Case: Robotics and vehicle automation
- Impact: Faster AI deployment in Hyundai vehicles
According to the article, Hyundai has already shown interest in collaborating more closely with Nvidia. with order to aid with the development of robotics and self-driving vehicles, the carmaker earlier declared intentions to purchase 50,000 Blackwell graphics processing units (GPUs) from Nvidia.
GPU Investments and Artificial Intelligence Strategy
Mercedes-Benz, on the other hand, has advanced more quickly in this field and is the first automaker to use Nvidia’s Alpamayo AI. Mercedes-Benz’s next CLA model, which is scheduled to enter US roads in the first quarter of this year, followed by Europe in the second and Asia in the third, will be the first Nvidia AI-powered self-driving car.
Mercedes-Benz Leads With Nvidia Alpamayo AI
🚗 Nvidia Alpamayo Artificial Intelligence Platform
- Innovation: Thinking and reasoning self-driving AI
- Model Type: Vision-language-action (VLA)
- Capability: Complex navigation and decision-making
- Tools: Simulation for rare and hazardous scenarios
- Access: Open-source AI models and datasets
- Deployment: First used in Mercedes-Benz CLA
The world’s first “thinking and reasoning” self-driving vehicle AI is Nvidia’s Alpamayo family of open-source AI models. It will present vision-language-action, or VLA, models that enable self-driving systems to comprehend their surroundings, navigate challenging driving scenarios, and execute precise driving maneuvers. Large-scale reasoning models, simulation tools for testing uncommon or hazardous situations, and publicly available datasets for AI system training and information verification will also be part of the platform.
How Alpamayo Artificial Intelligence Works
During a recent Q&A session with the media and analysts in Las Vegas, Huang said, “We open source to everyone, so if a client would want to utilize our model that we train, they are free to do so. We simply want to allow the world’s autonomous industry.” “Everything that moves need to be independent,” he said.
Open-Source Vision for Autonomous AI
In response to a question about possible memory shortages, Huang said that he is unconcerned since Nvidia is the “first and sole customer of HBM4,” a high-bandwidth memory technology. He added, “We are not anticipating anyone else to be utilizing HBM4 for some time.” Thus, we have the advantage of being HBM4’s main and exclusive customer, Huang said. “Our demand is so strong, every plant, every HBM provider is gearing up and we are all doing fantastic,” Huang said. Nvidia’s top two HBM suppliers are Samsung Electronics and SK Hynix.
HBM4 Memory and Nvidia’s Supply Advantage
During an onstage discussion at CES 2026, Siemens CEO Roland Busch and Nvidia CEO Jensen Huang cited major Korean shipbuilder HD Hyundai as a crucial example. By integrating Nvidia’s Omniverse platform with Siemens’ digital twin technology, HD Hyundai was able to digitize its shipyards.
