2026 may mark the beginning of AI's large-scale entry into the physical world—from walking robots to autonomous vehicles, AI is accumulating ecosystem hardware.
According to information, a Deutsche Bank research report released on January 13 indicates that the bank's analyst team attended the CES exhibition in Las Vegas last week and sensed a significant surge in market enthusiasm and relevance. The bank pointed out that vehicle autonomous driving (Robotaxi + consumer-level L4) and the most eye-catching humanoid robots took center stage at the exhibition.
Deutsche Bank concluded in its report: "Overall, we predict 2026 will be a year where autonomous vehicles increasingly transition from testing/validation to scaling, while humanoid robots will move from laboratory experiments to small-scale deployment."
The report emphasized that the humanoid robot field is cultivating an entirely new supply chain, with suppliers attempting to transition into this area in hopes of achieving large-scale volume production in the future. Meanwhile, the deployment momentum for Robotaxi in the autonomous driving sector is strong, and chip giants like NVIDIA are reshaping the competitive landscape by launching new platforms.
Deutsche Bank listed 10 key observations in its report: 1. Humanoid robot supply chain taking shape: Actuators become the "muscle" entry point Deutsche Bank believes that although it is still early stages, suppliers are already pivoting towards the humanoid robot supply chain, a path similar to electric drive systems: offering both integrated solutions and underlying components.
Schaeffler is attempting to become a primary "muscle" provider for humanoid robots, supplying linear and rotary actuators. At CES, it showcased an integrated planetary gear actuator for humanoid robots: a compact integration of a two-stage planetary gearbox + motor + encoder + controller. This unit features high thermal stability, a 60–250 Nm torque range, and very low backdrive capability, allowing it to withstand external forces and prevent accidental reversal of drive components, making it suitable for continuous operation. Deutsche Bank mentioned that NEURA has agreed to use Schaeffler actuators in its humanoid robots, and it appears other customers are already using (at least some components) or will use them in the future. Hyundai Mobis also announced it will supply actuators for Boston Dynamics' Atlas, aiming to leverage the automotive industry's scaled supply chain for robot manufacturing.
When a supply chain begins to "automotivize," the first things to be priced are often not concepts, but the penetration and scaled manufacturing capabilities of key components. 2. Onboard chip landscape: NVIDIA remains the preferred choice, but differentiation is emerging Deutsche Bank observed that NVIDIA still dominates in onboard processors for humanoid robots, primarily due to performance and ease of use. Companies using Jetson Orin or Thor include: 1X, Agility, Apptronik, Boston Dynamics, Figure AI, Mentee, (currently) NEURA, UBTECH, Unitree, among others. In contrast:
Tesla and Xpeng use self-developed inference chips. At CES, Qualcomm launched its next-generation solution (Dragonwing IQ10 Series) for a robotic "full-stack architecture," but Deutsche Bank stated it's unclear if it will gain large-scale customer adoption; meanwhile, VinMotion's Motion 2 humanoid robot uses the IQ9 Series, with the IQ10 initially targeting industrial AMRs and more advanced full-size humanoid robots.
3. "Physical AI" moving from scripting to Agentic: VLA becomes the main theme One of the most notable paradigm shifts on-site was the move from "pre-programmed/scripted actions" to Vision-Language-Action (VLA), enabling robots to "reason" to complete tasks.
Boston Dynamics replaced traditional MPC (Model Predictive Control) with Google DeepMind Gemini Robotics' VLA model, allowing Atlas to understand previously unseen environments (such as unstructured, chaotic factory scenes). Its action execution is supplemented by TRI's Large Behavior Model (LBM), similar to Figure's Helix dual-system model: System 1 handles high-frequency, rapid responses, while System 2 handles low-frequency, high-level reasoning and language; Deutsche Bank also noted that Figure appears to be developing both sets of models in-house.
4. Training debate escalates: Real-world data and simulation "closed loops" are the real focus Deutsche Bank judges that the industry debate has shifted from "which is better, simulation or real-world" to "how to achieve an efficient closed loop."
NEURA adopts a more "physics-first" approach, building the large physical training center NEURA Gym, believing simulation is an "approximation" that can be inaccurate for complex contact tasks (e.g., "threading a needle"); it collects high-fidelity data through hundreds of robots performing real tasks like sorting and assembly, inputs this into "Neuraverse" to generate "synthetic twins" of real failures for training in simulation, and finally pushes the repaired solutions back to the real robots. Another company mentioned the inability to simulate the "feel" of objects, requiring human demonstration first: through teleoperation, a human wearing a VR suit controls a humanoid robot to perform actions like "picking up a grape"; after a few "perfect examples," it uses NVIDIA GR00T-Mimic to generate "100,000+" action variations in simulation and uses reinforcement learning to smooth the actions. In contrast, Mobileye emphasized that its Mentee will primarily be trained using simulation.
5. "General" gives way to "roles": Commercial proof takes priority Deutsche Bank believes that in the short term, "general-purpose humanoid robots" will be deployed more into specific scenarios to prove commercial viability before discussing entry into homes.
Keenon Robotics (China): Already holds 40% global market share in service robots, with cumulative exports of approximately ~100,000 units overseas; product prices range from under 10,000 to about 100,000 RMB, focusing on strong task customization. CES 2026 highlighted its flagship humanoid robot XMAN-R1, capable of making popcorn, pouring drinks, and engaging in anthropomorphic gesture interactions; its "Brain" is the Keenon Operator Model 2.0, a VLA model tailored for the service industry, capable of understanding commands like "find the guest at table 4 and give them candy." Keenon also mentioned building a collaborative ecosystem at the Shanghai Shangri-La Trade Hotel: MAN-R1 handles human-robot interaction as the "front desk," W3 delivers items to rooms, S100 moves heavy luggage, and C40/C55 handle cleaning. Furthermore, in high labor-cost markets like Japan, its robots have a service life reaching 8 years, significantly higher than the industry's typical 3–5 years. Deep Robotics focuses on industrial inspection: Measured by coverage distance (up to 63km), it can perform 24/7 autonomous patrol monitoring in hazardous areas like substations, power plants, and oil & gas facilities; used in emergency scenarios for disaster relief, firefighting, and toxic gas detection, and employs swappable batteries to reduce charging friction.
6. The cost reduction formula is straightforward: Scale = prerequisite for cost decline On the humanoid robot side, Deutsche Bank attributes the main drivers of cost reduction to: increased volume improving expense amortization + improved supplier bargaining power.
One company claimed costs have already dropped from "$200,000 to $100,000" and plans to reduce them to "$50,000" in the "coming years," provided sales reach several thousand units. Boston Dynamics and Hyundai Motor announced a target of achieving an annual production capacity of 30,000 units by 2028; furthermore, its entire 2026 production output has been pre-allocated to Hyundai's automotive factories. The company also noted that actuators account for approximately ~60% of the Bill of Materials (BoM), and this portion will be manufactured by Hyundai Mobis within the Hyundai system to accelerate scaling. Mobileye, following its acquisition of Mentee, disclosed: at an annual production volume of 50,000 units, the manufacturing cost for a relatively simplified design (without a tendon-drive system) would be about "$20,000/unit"; if annual production reaches "100,000 units," the cost could halve to "$10,000/unit," targeting a ramp-up by 2028, with production handled by Aumovio.
7. Robotaxi momentum building: 2026 looks more like a "commercial acceleration year" Deutsche Bank believes that with Tesla's launch of Robotaxi in 2025, the commercial momentum for multiple players will be stronger in 2026, as signaled by the substantial presence of Waymo and Zoox at CES:
Waymo: Has provided 10 million+ paid rides since inception; latest disclosures show it reached 450,000 paid rides per week in December 2025, expanding to Houston, Miami, and international markets like Tokyo and London. Amazon's Zoox: Progressed from public testing in Las Vegas to showcasing a "market-ready product," featuring a "carriage-style" Robotaxi designed for dense cities, completely lacking a traditional driver's cabin. Mobileye and Volkswagen: Will launch an L4 Robotaxi service in Los Angeles this year using specially prepared ID. Buzz electric vans. Additionally, an autonomous vehicle initiative based on the Lucid Gravity, jointly advanced by partners Nuro, Lucid, and Uber, is planned to launch in the San Francisco Bay Area by the end of 2026, with further expansion to more cities.
8. NVIDIA Alpamayo: Bundling the "brain + skull" for automakers, but validation is ongoing NVIDIA announced the launch of Alpamayo (the "brain") for autonomous driving, paired with Thor (the "skull"), attempting to lower the barrier for automakers to deploy high-level capabilities: companies like Lucid and Mercedes-Benz can "plug in" NVIDIA's solution without needing to invest billions from scratch to build AI infrastructure. Deutsche Bank remains cautious: this indeed sparks discussion about Tesla's moat, but it's premature to worry now; NVIDIA still needs traditional OEMs to deliver on their promises, and whether its models can cover real-world edge cases remains to be seen. The bank points out that its training data volume is only a fraction of the data collected by Tesla. Even if Alpamayo proves effective, Deutsche Bank still believes Tesla holds a structural cost advantage due to vertical integration (vehicle, chips, AI infrastructure, network, etc.); if autonomous driving/Robotaxi becomes commoditized, cost will become the biggest differentiator. 9. Aptiv: End-to-end AI ADAS + connectivity & software platform, pitching a "cross-industry" story Aptiv's showcase centered on its next-generation End-to-End (E2E) AI-driven ADAS platform: utilizing the newly released Gen 8 radar and PULSE sensors to achieve L2++ hands-free driving with "human-like logic" in complex urban environments. On the software side, it launched LINC, a cloud-native middleware platform built with Wind River, enabling truly software-defined vehicles through 5G and C-V2X; it also demonstrated with Verizon how vehicles sharing real-time data can "see pedestrians/cyclists around the corner." Aptiv also emphasized expanding its sensors into aerospace and collaborative robots—Deutsche Bank views this as the narrative the "new Aptiv" needs to prove to争取 a re-rating of its valuation multiples. 10. Visteon: 700 TOPS domain controller, plug-in upgrades, focusing on "execution" Visteon launched the SmartCore HPC domain controller at CES, boasting 700 TOPS of computing power, capable of integrating up to 14 cameras and multiple high-speed data links into a single "central brain." It also expanded its collaboration with Mahindra, introducing the SmartCore Pro (triple screen + 360-degree view) for the upcoming XUV7X0. To address obstacles with "existing platforms," Visteon also launched an AI-ADAS Compute Module plug-in solution powered by NVIDIA DRIVE AGX Orin, allowing automakers to add AI assistants or safety features without completely redesigning their architecture; Deutsche Bank mentioned this product is already deployed on China's Zeekr models. Furthermore, Visteon released an "Entry Cockpit" for screens under 7 inches, bringing smartphone projection and digital navigation to two-wheelers and entry-level vehicles. Deutsche Bank assessed that its "surgical" vertical integration aids cost competitiveness and drives further expansion among automakers where it has historically had lower penetration (especially Asian OEMs). In Deutsche Bank's view, the message from CES 2026 is straightforward: autonomous driving and humanoid robots are shifting from "can it be done?" to "can it be scaled, and can the cost be driven down?" When Boston Dynamics highlights that actuators account for ~60% of costs and pre-allocates its 2026 production, the industry has started pricing using manufacturing language; meanwhile, Waymo's 10m+ paid rides and 450,000/week pace push Robotaxi from concept towards harder operational data. For investors, the next phase is not about tracking flashier demos, but rather supply chain commitments, production ramp-ups, and unit cost curves.
Comments