CES 2026 – Autotech and Mobility Highlights
January 2026
CES 2026 has closed its doors in Las Vegas. Below I review the mobility and autotech highlights collected during my tenth participation in this major tech event with over 35 km of walking. You will find here similar reports on previous CES editions since 2017.
About 150,000 professionals visited the 4,100 exhibitors, including around 1,500 startups, and attended keynotes and conferences across a variety of tech domains. For several years now, mobility and automotive technologies have been major points of focus at CES, filling up all of West Hall, making up 25% of CES exhibition areas. Korea had an oversized presence for the second year in a row at Eureka Park, the space dedicated to early-stage startups, occupying close to a third of the surface area.
Major Trends: AI at the Forefront
Let’s start with a broad overview of CES 2026. The main theme was without a doubt AI. Compared with 2024 and 2025, messages have evolved from chatbots and assistants, genAI use cases, and a wide range of wild AI-based concepts presented by startups, to more mature applications. We are moving into the industrialization phase with use cases offering clear operational benefits, particularly with physical AI. Applications presented include engineering, simulation and validation, manufacturing, robotics with several humanoid demos, and of course autonomous mobility (more of this later). In addition, agentic AI was present on multiple exhibits with concrete applications and tool chains, for instance offering extended automotive assistant features.
The massive push in everything-AI is impacting other domains including energy, and indirectly automotive. Indeed, cooling experts, such as Valeo, are pursuing AI’s high power compute market. Excess battery capacity is being partly redirected to stationary energy storage.
On the mobility side, the main themes entailed autonomous driving (and autonomy at large), advanced chips and general compute, embedded AI, maturing Software-Defined Vehicle (SDV) and emerging AI-Defined Vehicle, and cockpit technologies. Electric vehicles and batteries took a step back vs. previous events as part of a wider challenge in the US.
The difficulties the auto industry is currently experiencing was reflected at CES in that few major players exhibited, as was already the case in 2024 and 2025. On the OEM side, BMW, Hyundai, Geely, Great Wall Motors and newcomer Sony Honda Mobility were the only ones showing vehicles in the main areas. Mercedes-Benz had a side event. A Xiaomi YU7 was on display at a software provider. But Toyota, VW Group, GM, Stellantis, Ford, and more were absent. It was a similar story for major Tier 1 suppliers. The list of exhibitors was limited to Bosch, Magna, Valeo, Aumovio (Continental spin-off), Aptiv, Schaeffler, Mando, and LG Electronics. A few others had a small presence in private spaces.
Few OEM Announcements
BMW presented the iX3 which is launching now. It is the first vehicle on the Neue Klasse platform, fitted with an SDV architecture (4 domain HPCs, 4 zonal controllers) and panoramic iDrive, which was presented as a concept at CES 2025. The latter features a customizable, display- and mirror-based set of screens at the bottom of the windshield in addition to a trapezoidal 17.9-inch center display (see above).
Mercedes-Benz announced the introduction of navigation on autopilot or NOA-capable L3 for urban use, which was launched in China in 2025. Its infotainment system will now integrate AI models from both Microsoft and Google. In addition, the latest CLA (launch in 2026), introduces a new Level 2+ ADAS developed with Nvidia (also makes the dedicated controller) and OTA update capability addressing all vehicle functions. The CLA can be optioned with a pillar-to-pillar set of three displays, which are quite intrusive compared to BMW’s solution (see below). I anticipate the market will follow the BMW way rather Mercedes’.
Hyundai’s very large exhibit was focused on human-centered AI robotics. It included "Atlas”, a humanoid bot designed by Boston Dynamics, a subsidiary. The OEM plans to produce 30k units per year by 2028 and deploy them in its car manufacturing plant by then, in a move similar to that of Tesla or Xpeng. Hyundai also presented an EV charging robotic arm as well as MobED, a small, modular, autonomous 4-wheel platform.
Geely, which sold 3.0 M vehicles in 2025 including 1.7 M BEVs and PHEVs, displayed the Zeekr 009 and 9X, as well as several subsystems, including a 1 MW (1340 hp) tri-motor powertrain with integrated power electronics and gear reducer (see below). More importantly, leaders presented a new AI strategy addressing all vehicles domains, i.e., chassis, cockpit, body, and powertrain, to enable human-like decision-making. The company sees mobility as the optimal platform to host multimodal AI (language, image/video, and agentic) and is inching towards the AI-Defined Vehicle. The Zeekr 9X is Level 3-ready, with 5 Lidars and 1,400 TOPS of compute (two Nvidia Thor chips) while the OEM is developing G-ASF, its own assisted driving solution. Geely also announced a partnership with Cerence, leveraging the latter’s xUI agentic AI platform to provide in-car experiences that are optimized for regional linguistic and cultural habits.
Great Wall Motors (1.3M vehicles sold in 2025, incl. 500k outside China) was the other established Chinese OEM to exhibit. Three vehicles were on display alongside — surprisingly — a V8 gas engine with its auto transmission, a fuel cell, and 2.0 liter, flat-8 motorcycle aiming to challenge Honda’s Gold Wing. Like Geely, GWM is showing off its capability to the world.
Sony Honda Mobility displayed a pre-production version of the Afeela 1, its first vehicle which will launch in the U.S. in late 2026. First presented at CES 2023, the sedan has aged. One can hope for a wow effect from the user experience given the Sony affiliation, not just for gaming. The vehicle is likely Level 3 ready with its Lidar and two cameras integrated in the roofline. SHM also introduced its second model in concept form. The SUV-coupe is expected to launch in 2028. Why present it again that early? See below the sedan on the left and SUV on the right.
There was also a surprise prototype presented by a Chinese electronics company. The Kosmera Star Matrix is a sleek 4-door featuring 2,000 hp from four motors, featuring a Bugatti-light front end and Porsche Tayco profile. I am not sure whether this car has a future but it certainly looks great!
Autonomous Driving Building Momentum
AD tech has become integral to CES. Waymo, which leads the robotaxi market at least in the U.S., again had a strong presence. On display were not only the current Jaguar iPace but also the soon-to-be introduced Zeekr RT (now Ojai) and the future Hyundai Ionic 5. The Alphabet subsidiary operates 2,500-3,000 robotaxis across five U.S. cities commercially and is currently testing without safety operators in five more where commercial operations are to start in 2026. It is also testing in London and Tokyo. I have taken multiple rides in San Francisco with very good experiences.
Zoox had a more prominent exhibit than previous years. Its purpose-design robotaxi was on display, featuring 2x2 seats facing each other, bi-directionality up to 120 km/h, no steering wheels or pedals, and 2x2 sliding doors (see above). I have had a chance to ride in this vehicle in the San Francisco Bay Area five times so far although the service is not yet open to the public. It offers a unique experience which I got accustomed to rather easily.
Lucid and Nuro presented a version of the recently introduced Gravity SUV, fitted with Nuro’s AD tech. The Lidars, radars, cameras and compute specified by Nuro will be installed at Lucid’s factory (see below). No major redesign of the vehicle was required as it is Level 3 ready with redundant safety functions, i.e., steering, braking, and power. This project is part of three-way partnership with Uber which will own and operate the robotaxi on its ride-hailing platform starting in late 2026 in the SF Bay Area. Uber is committed to buying at least 20,000 such vehicles over the next six years.
In addition, Tensor (formerly AutoX) presented its Robocar, an autonomous vehicle destined for private ownership although the company has an agreement with U.S. ride-hailing operator Lyft to deploy the vehicle on its platform. In my November 2025 article, you will find more information on this vehicle and on plans by other companies to introduce privately-owned AVs.
Other Auto Tech
In a dedicated space, Nvidia offered an impressive display of the tech — its own and that of a few partners — which fuels its amazing growth. In the AD space, the 4.5 trillion-dollar company introduced “Alpamayo”, an open-source platform designed to develop reasoning-based autonomous driving solutions. It is a foundational “tool kit” featuring models, a dataset, and simulation tools which companies can use to develop their AD tech. Nvidia insisted on the benefit of its solution to address long-tail scenario.
Bosch presented a cockpit solution powered by an AI assistant developed with Nvidia and Microsoft. It provides “comprehensive scene understanding of the vehicle interior, precise navigation, and extensive entertainment options” according to Bosch. The company also presented SiC power semiconductors, a family of radar SoCs, and an ultrasonic sensor IC chipset.
Valeo introduced an evolution of its Scala 3 Lidar with higher point cloud accuracy and a more compact design, electrifications subsystems (compact 5‑way refrigerant valve, battery pack encloser tech, fan system with a lateral blower for EVs’ heat exchangers), updated SDV tech and more. As stated earlier, the company has developed thermal management solutions specifically for the growing stationary high-performance compute market, in particular AI factories.
Last, agentic AI solutions were presented by both SoundHound and Cerence for new in-car experiences. They include an AI parking agent, a dealer assistance agent, and an ownership companion agent which combine the driver’s input, their calendar and messaging app, vehicle parameters, and more. The companies also demonstrated no-code tool chains to easily develop new agents. Agentic AI will come very soon to our cars!
There were also multiple announcements on the SDV front for compute hardware (HPCs), software, and development tools. They came essentially from Elektrobit, Qualcomm, Renesas, QNX, Vector and are aimed at accelerating and simplifying the deployment of SDV.
Happy new year to all!
Marc Amblard
Managing Director, Orsay Consulting