Jensen Huang's keynote at CES 2026 showcased NVIDIA's continued dominance in AI and accelerated computing, though some analysts noted a focus on refining existing technologies rather than groundbreaking new product categories. The announcements centered around advancements in autonomous vehicle AI and the launch of the Vera Rubin computing platform. Let's dive into what this means for the industry.
Alpamayo: Human-like Reasoning for Autonomous Vehicles
The most intriguing announcement was Alpamayo, a family of open-source reasoning models for autonomous vehicles. Alpamayo 1, the flagship 10-billion parameter model, uses a chain-of-thought approach to mimic human decision-making in complex driving scenarios. According to NVIDIA, Alpamayo breaks down unexpected situations into smaller, manageable problems, explaining its reasoning at each step to ensure safety. A sister model, AlpaSim, enables closed-loop training for rare and challenging driving situations, a crucial step in ensuring AV safety and reliability.
According to Engadget, the 2025 Mercedes Benz CLA will be the first vehicle to integrate NVIDIA's complete AV stack, including Alpamayo. This signifies a major win for NVIDIA and a step towards Huang's vision of a fully autonomous future for all vehicles. It's worth noting that while the model is open-source, the underlying hardware and software ecosystem remain firmly within NVIDIA's control, giving them a strategic advantage.
Vera Rubin: A New Era of AI Compute
NVIDIA officially launched the Vera Rubin platform, named after the pioneering astronomer. The Verge reports that NVIDIA's Dion Harris described it as "six chips that make one AI supercomputer." The platform comprises the Vera CPU, Rubin GPU, NVLink 6th-gen switch, Connect-X9 NIC, BlueField4 DPU, and Spectrum-X 102.4T CPO. This comprehensive integration is designed to dramatically reduce the costs associated with both AI training and inference, crucial for the continued scaling of AI models.
The Rubin GPU boasts five times more AI training compute power than its predecessor, Blackwell, according to TechMeme. Furthermore, Wired notes that the Vera Rubin platform is already in "full production," signaling NVIDIA's confidence in its ability to meet the growing demand for AI compute. The platform will also support 3rd-generation confidential computing, making it suitable for sensitive AI workloads. The move underscores NVIDIA's ambition to provide a complete, integrated AI computing platform, solidifying its position as the dominant player in the field.
While NVIDIA didn't unveil any groundbreaking gaming-related GPUs at the keynote, further announcements may come later during CES. The focus on autonomous vehicles and AI infrastructure reflects the company's strategic shift towards high-margin, enterprise-grade solutions. The Vera Rubin platform, with its promise of reduced training and inference costs, coupled with the advancements in autonomous driving AI with Alpamayo, indicates NVIDIA's commitment to pushing the boundaries of what's possible with artificial intelligence.