ADI Adopts NVIDIA Jetson Thor to Advance Humanoid Robotics

Aaron Blutstein 9 September 2025  

Humanoid robots are moving closer to real-world deployment, with progress increasingly dependent on physical intelligence and real-time reasoning. Analog Devices, Inc. (ADI) has announced it is adopting NVIDIA’s newly released Jetson Thor platform to accelerate the development of humanoids and autonomous mobile robots (AMRs).

According to ADI, combining its edge sensing, precision motion control, power integrity and deterministic connectivity with Jetson Thor’s compute capabilities, Holoscan Sensor Bridge and Isaac Sim will create “a path to scale reasoning-enabled robots from simulation to deployment.”

Jetson Thor’s Capabilities

ADI highlighted Jetson Thor as a major advance for robotics. The platform integrates a NVIDIA Blackwell GPU, transformer engine, Multi-Instance GPU (MIG), a 14-core Arm Neoverse V3AE CPU, and up to 128GB of LPDDR5X memory. It delivers 2070 FP4 TFLOPS server-class AI compute in a mobile power envelope, with high-throughput I/O including 4×25GbE for dense multimodal sensing in real time.

This capability makes Jetson Thor the first platform able to run robotics foundation models at scale, from vision-language to vision-language-action models. ADI says this enables robots to progress beyond perception into reasoning and physically intelligent behaviour, aligning with its research and development focus on sensing, perception, control and connectivity.

Paul Golding, VP of Edge AI, ADI commented: “For the first time, robots can understand complex tasks. ADI delivers the precision physical substrate which, combined with NVIDIA Jetson Thor’s reasoning, responds to real world physics in real time. Together, we’re taking humanoids from simulation to shift ready deployment.”

Foundation Models and Physical Intelligence

Robotics foundation models are seen as key to advancing reasoning and physical intelligence. They enable perception-rich humanoids capable of dexterous, human-speed manipulation while integrating multimodal inputs to plan, adapt and act in real time.

On its third-quarter 2025 earnings call, ADI noted that its opportunities in robotics expand alongside this shift, citing demand for precise control at every joint, tactile and sensory feedback at each contact point, and multiple perception nodes across humanoid systems.

Closing the Sim2Real Gap

ADI is embedding robotics foundation models into its development stack to help close the “Sim2Real” gap, aiming for hardware that behaves in NVIDIA Isaac Sim as it would in the real world. The company said: “Our goal: build the most physically accurate robotics content in NVIDIA Isaac Sim, enabling teams to iterate at simulation speed and then scale seamlessly to real systems with ADI hardware and NVIDIA Jetson Thor.”

Physical intelligence, ADI stated, requires high-fidelity edge sensing, energy-efficient and functionally safe power, deterministic connectivity to central compute, and a digital twin that links simulation with physical deployment.

ADI’s Contribution

ADI outlined several areas where it is contributing to humanoid robotics:

• High-fidelity edge sensing for contact-rich manipulation:multimodal tactile sensing, ToF depth, high-accuracy IMUs, joint encoders, and multi-axis force/torque sensors.
• Precision motion and safe power control:drivers and control for current, position and torque, plus advanced multi-turn magnetic sensors.
• Deterministic connectivity to central compute:time-synchronised data paths integrated with Holoscan.
• Simulation and digital-twin fidelity:high-quality sensor models and parameterised device behaviour for NVIDIA Isaac Sim/Omniverse.

Integration with Jetson Thor

ADI also detailed how its robotics stack integrates with Jetson Thor:

• Holoscan Sensor Bridge streams synchronised sensor and actuator data into Jetson Thor’s GPU/CPU with bounded latency.
• 4×25GbE interconnect supports high-throughput fusion across system nodes.
• Thor’s 2070 FP4 TFLOPS compute supports foundation models such as NVIDIA Isaac GR00T, alongside reasoning.
• MIG-based workload partitioning enables functional decomposition across locomotion, grasp planning, perception and vision-language-action policies.

Golding added: “With NVIDIA Jetson Thor as the brain and ADI’s high-fidelity sensing, signal-chain fidelity and deterministic connectivity as the nervous system, we take robots from NVIDIA Isaac Sim to the factory floor with physical accuracy – faster.”

Future Applications

ADI sees demand for humanoids across logistics, agriculture and surgical robotics, with frontier use cases in dexterous manipulation for data centres and automotive manufacturing. The company is also collaborating with NVIDIA on digital twins and policy training in Isaac Sim, and extending its approach to AMRs, including integration with cuVSLAM via IMUs, depth sensors and wheel encoders.

“We’re just getting started. NVIDIA Jetson Thor opens a new chapter in our partnership with NVIDIA,” ADI said, pointing to potential applications across multiple industries.

Availability

• NVIDIA Jetson AGX Thor Developer Kit and NVIDIA Jetson T5000:see NVIDIA for current availability, NVIDIA JetPack 7 support, and ordering.
• ADI evaluation hardware and software:available through ADI’s robotics team, including early access to simulation models and tactile sensing.

Photo by Tara Winstead from Pexels (C) 2021.