AI, Robots and Robotaxis Dominate CES 2026
Key Points
- Nvidia launched Alpamayo, open ‘reasoning’ AI models for self-driving cars
- Lucid, Nuro and Uber unveiled a production‑intent robotaxi for San Francisco
- Hyundai detailed plans to deploy Boston Dynamics’ Atlas robots in factories
- Qualcomm introduced Dragonwing IQ10 chips targeting humanoid robots
Nvidia pushes ‘reasoning’ AI for autonomous vehicles
At CES 2026 in Las Vegas, Nvidia founder and CEO Jensen Huang unveiled Alpamayo, a family of open AI models, simulation tools and datasets designed to bring chain‑of‑thought “reasoning” to autonomous vehicles. Nvidia describes Alpamayo as a vision language action model that can think through rare, complex driving scenarios, improve explainability and support level‑4 autonomy. Alpamayo 1, a 10‑billion‑parameter model, is being released on Hugging Face with open weights and inference scripts, while the AlpaSim simulation framework and large “Physical AI” driving datasets are available to help developers test and refine systems. Rather than running directly in vehicles, Alpamayo is positioned as a large teacher model that automakers and researchers can distill into their own stacks. Nvidia says mobility players including Lucid, Uber, JLR and Berkeley DeepDrive are looking to use Alpamayo to accelerate reasoning‑based autonomy. Analysts cited in BBC coverage said the launch reinforces Nvidia’s shift from pure compute supplier to a broader platform provider for physical AI ecosystems.
Mercedes‑Benz CLA showcases Nvidia’s AI‑defined driving
Nvidia is already moving Alpamayo‑based technology into production vehicles. Huang said the first passenger car featuring the company’s full autonomous stack will be the Mercedes‑Benz CLA, powered by Nvidia’s DRIVE platform and Alpamayo models. A video demonstration showed the AI‑equipped CLA driving through San Francisco with a passenger seated behind the wheel and hands in their lap, while the system narrated and explained its intended actions. Huang said the car “drives so naturally because it learned directly from human demonstrators” and that in every scenario it explains what it is about to do. According to Huang, the CLA will launch with AI‑defined driving in the US in the coming months, followed by Europe and Asia. Nvidia has also begun producing a driverless version of the CLA with Mercedes‑Benz. The company’s broader vision, Huang told audiences, is that eventually every car and truck will be autonomous, and Nvidia is planning a robotaxi service with an unnamed partner by around 2027 using level‑4‑capable vehicles.
Uber, Lucid and Nuro advance robotaxi plans with Nvidia tech
Ride‑hailing group Uber and partners Lucid and Nuro used CES to debut a production‑intent robotaxi based on Lucid’s all‑electric Gravity SUV. The vehicle, which uses Nuro’s level‑4 autonomous driving system and Nvidia’s DRIVE AGX Thor computing platform, is being positioned for a commercial launch on Uber’s platform in the San Francisco Bay Area later this year, pending regulatory clearance. The robotaxi features a roof‑mounted sensor “halo” integrating cameras, lidar and radar for 360‑degree perception, and an Uber‑designed cabin with interactive screens that let riders adjust climate, seating and music, contact support or request the vehicle to pull over. The partners said on‑road testing with safety operators began in December, alongside closed‑course testing and simulation, with production expected to start at Lucid’s Arizona factory later in 2026 after final validation. Uber, which now focuses on partnering with third‑party AV developers rather than owning the technology, already allows users to hail Waymo robotaxis in some US cities and aims to add the Lucid–Nuro vehicles to its global network.
Hyundai outlines large‑scale humanoid robotics rollout
Hyundai Motor Group used CES 2026 to detail an AI robotics strategy built around Boston Dynamics’ Atlas humanoid robot and a broader “physical AI” roadmap. The company said it plans to roll out human‑like Atlas robots across its global manufacturing network from 2028, including at a plant in Georgia that was the site of a major immigration raid in 2025. Atlas is being developed for general industrial use to work alongside people, autonomously manage machines and gradually take on more tasks. Hyundai said the robots are intended to ease physical strain on workers, handle dangerous jobs and support human‑centric smart factory operations. The group aims to deploy Atlas at its Hyundai Motor Group Metaplant America facility by 2028 for parts sequencing tasks, with more complex assembly roles targeted by 2030. Hyundai, which owns a majority stake in Boston Dynamics, also plans to build an end‑to‑end AI robotics value chain with affiliates such as Hyundai Mobis and Hyundai Glovis, and to expand Robotics‑as‑a‑Service offerings as it scales toward producing tens of thousands of robots annually.
Qualcomm targets physical AI with Dragonwing robotics chips
Chipmaker Qualcomm is also using CES 2026 to expand into physical AI and robotics. The company introduced the Dragonwing IQ10 system‑on‑chip, optimized for humanoid robots and industrial autonomous mobile robots, as part of a broader robotics architecture that integrates hardware, software and AI. According to Qualcomm, Dragonwing IQ10 can power robots equipped with more than 20 cameras and multiple sensors, delivering 350 TOPS of AI performance, more than triple its previous flagship robot chip. The platform includes a neural processing unit, GPU, 18‑core CPU, safety island and error‑correction features. Qualcomm is building an ecosystem around Dragonwing with partners such as Figure and Kuka Robotics, and has also announced related Dragonwing Q‑7790 and Q‑8750 processors for edge AI devices. Separately, Innodisk launched its “AI on Dragonwing” computing series using Qualcomm Dragonwing SoCs, including the EXMP‑Q911 COM‑HPC Mini module, which targets up to 100 TOPS of AI performance for industrial edge deployments and is supported by open‑source developer tools and cloud management software.
Key Takeaways
- CES 2026 is emerging as a key inflection point where AI models, chips and vehicles for autonomous driving are being tied together into open, reusable platforms.
- Nvidia is moving beyond data‑center GPUs into full physical‑AI stacks, spanning open AV models like Alpamayo, in‑car platforms and planned robotaxi services.
- Automakers and tech suppliers, including Hyundai, Qualcomm and Uber’s partners, are positioning robotics and robotaxis as long‑term pillars of manufacturing and mobility.
Get premium market insights delivered directly to your inbox.