In this Article
Applied Intuition And Nvidia Accelerate L2+ ADAS Deployment
Applied Intuition has announced a strategic collaboration with NVIDIA to streamline the development of Level 2+ driver assistance systems for global automotive manufacturers. Recognized as a recommended software provider at NVIDIA GTC, Applied Intuition offers a production-ready ADAS stack optimized for NVIDIA DRIVE AGX Orin and the forthcoming DRIVE AGX Thor platforms. The partnership targets reduced time-to-market and integration risk for OEMs across internal combustion, hybrid, and electric powertrains by aligning autonomy software with high-performance AI compute infrastructure.
The collaboration integrates Applied Intuition’s data and simulation flywheel with NVIDIA Cosmos World Foundation Models to enhance data augmentation and synthetic generation. This technical synergy allows for the curation of petabyte-scale datasets and closed-loop simulation across diverse environmental conditions and sensor behaviors. The resulting solution encompasses a full L2+ feature set for highway driving and automated parking, supported by a safety-critical Vehicle OS middleware architecture compatible with NVIDIA DriveOS to facilitate the industry transition toward software-defined vehicles.
DeepRoute.ai Debuts 40B Parameter VLA Architecture At GTC 2026
DeepRoute.ai introduced its 40-billion-parameter Vision-Language-Action (VLA) Foundation Model at NVIDIA GTC 2026, marking a shift toward unified autonomous driving architectures. The model integrates perception, reasoning, and action into a single framework, serving simultaneously as driver, analyst, and critic. This consolidated approach enables the system to execute driving maneuvers while performing real-time causal reasoning and safety evaluations, addressing the cognitive limitations of traditional decoupled ADAS stacks.
The VLA architecture optimizes the data closed-loop by reducing iteration cycles from five days to approximately 12 hours. By automating data mining, root-cause diagnosis, and behavior scoring, DeepRoute.ai eliminates manual annotation bottlenecks that typically hinder scaling. Following a reported 40% market share among third-party suppliers in late 2025 and 250,000 systems currently in production, the company aims for one million vehicle deployments by year-end 2026 leveraging this self-evolving AI flywheel.
Tier IV And Isuzu Deploy Level 4 Autonomous Buses On Nvidia Thor
TIER IV and Isuzu Motors have announced the deployment of Level 4 autonomous driving systems for Isuzu ERGA electric and diesel bus models, utilizing the NVIDIA DRIVE Hyperion computing platform. Revealed at NVIDIA GTC 2026, the integration features the NVIDIA DRIVE AGX Thor system-on-chip (SoC) to provide 2,000 TFLOPS of performance and ASIL-D safety redundancy. The solution combines TIER IV’s Autoware-based open-source software stack with Isuzu’s vehicle platforms to address mass transit scalability and regional driver shortages.
The partnership establishes a fail-operational architecture for high-capacity public transport by leveraging standardized sensor suites and high-performance AI compute. By utilizing the DRIVE Hyperion reference architecture, TIER IV and Isuzu aim to transition autonomous transit from development to commercial production. This deployment signifies a critical expansion of the Autoware ecosystem into the heavy-duty commercial vehicle segment, prioritizing operational rigor and functional safety for essential public services.
AEye Joins Nvidia Halos AI Systems Inspection Lab
AEye, Inc. has announced its participation in the NVIDIA Halos AI Systems Inspection Lab, a move aimed at integrating functional safety and cybersecurity into a unified AI safety framework. As the first ANAB-accredited AI inspection lab, NVIDIA Halos provides end-to-end safety guardrails across the AI development lifecycle, from design to deployment. AEye will collaborate within this ecosystem to validate interoperability and system-level integration of its perception technology with NVIDIA DRIVE platforms, ensuring alignment with global regulatory standards for autonomous systems.
The partnership reinforces AEye’s commitment to safety-certified physical AI, specifically supporting its Apollo 1-kilometer lidar which is already validated for NVIDIA DRIVE AGX Orin and DRIVE AGX Thor. By joining the Halos program, AEye aims to provide automakers with high-performance perception solutions that meet rigorous functional safety and validation requirements. This collaboration facilitates a streamlined path for OEMs to deploy advanced driver assistance and autonomous driving features with increased confidence in system integrity and compliance.
Lyft Integrates Nvidia AI Infrastructure For Hybrid Autonomous Network
Lyft announced a multi-faceted collaboration with NVIDIA at GTC 2026 to enhance its enterprise AI capabilities and develop a foundation for Level 4 autonomous fleet architectures. The rideshare platform will leverage NVIDIA DRIVE Hyperion as a reference architecture for future autonomous systems, integrating high-performance compute and multi-sensor processing. Lyft aims to create a hybrid ecosystem where partner-deployed and fleet-owned autonomous vehicles operate alongside human drivers, supported by the recently acquired Freenow network in Europe.
The integration utilizes the NVIDIA AI Enterprise suite, including Nemotron models and cuOpt, to optimize real-time rider-driver matching and predictive modeling. Furthermore, Lyft is transitioning to an agentic AI mapping platform powered by NVIDIA Cosmos Reason series for physical AI and multimodal vision-language reasoning. This infrastructure shift is designed to accelerate data enrichment and map-making precision, reducing compute costs while improving navigation and routing safety across Lyft’s global multimodal network.
PlusAI And Nvidia Expand Collaboration For Driverless Trucking By 2027
PlusAI announced at NVIDIA GTC 2026 an expanded collaboration with NVIDIA to commercialize its SuperDrive virtual driver for fully driverless factory-built trucks. The partnership integrates the NVIDIA Alpamayo foundation model for AI reasoning, the NVIDIA DRIVE Hyperion platform for hardware-level execution, and the NVIDIA Halos safety framework. PlusAI is among the first to adapt Alpamayo for heavy-duty trucking, utilizing a teacher-student architecture that compresses a 10-billion-parameter Vision-Language-Action (VLA) model into a 0.5-billion-parameter edge model to ensure low-latency reasoning on automotive-grade hardware.
Working with International Motors, PlusAI is developing production-ready Level 4 autonomous trucks powered by the NVIDIA DRIVE AGX Thor SoC and DriveOS. These vehicles feature a factory-integrated sensor suite providing 360-degree awareness and the compute redundancy necessary for long-haul freight. To support global commercial scaling, PlusAI has joined the NVIDIA Halos AI Systems Inspection Lab to undergo independent assessment against ISO 26262, ISO/SAE 21434, and ISO PAS 8800 standards, facilitating the certification required for major international markets.
Gatik Joins Nvidia Halos AI Systems Inspection Lab To Scale Driverless Freight
Gatik has announced its participation in the NVIDIA Halos AI Systems Inspection Lab, the first ANAB-accredited program for validating AI-driven physical systems. Operating fully driverless trucks across Texas, Arkansas, Arizona, and Ontario, Gatik serves as a real-world proving ground for the Halos framework as it scales toward hundreds of autonomous vehicles by year-end 2026. The collaboration focuses on establishing rigorous inspection methods that unify functional safety, cybersecurity, and AI integrity for high-frequency commercial logistics.
The Gatik Driver™ third-generation autonomous system incorporates automotive-grade hardware and reasoning architectures designed for middle-mile delivery. In parallel with its work with NVIDIA, Gatik is collaborating with Isuzu Motors to develop a mass-production, autonomous-ready vehicle platform. By integrating with the NVIDIA Halos safety ecosystem, Gatik aims to transition safety validation into a continuous operational discipline, ensuring compliance with evolving global standards while maintaining its $600 million contracted revenue pipeline for major retailers.
WeRide Showcases Robotaxi GXR At GTC 2026 For Southeast Asian Expansion
WeRide debuted its Robotaxi GXR at NVIDIA GTC 2026, announcing plans to deploy the vehicle across Southeast Asia through a strategic partnership with Grab. The GXR utilizes the NVIDIA DRIVE Hyperion platform and DRIVE AGX Thor SoC, integrated into WeRide’s HPC 3.0 computing unit. This hardware-software synergy reportedly reduces autonomous driving suite costs by 50% and total cost of ownership by 84%. WeRide and Grab will launch public AV trial operations in Singapore’s Punggol district on April 1, 2026, marking a critical step in their regional commercial roadmap.
Manufactured by Geely’s Farizon New Energy Commercial Vehicle Group, the purpose-built GXR is designed for high-density urban environments. WeRide has secured an agreement for 2,000 mass-produced units by 2026, with current hardware costs at approximately US$40,000 per vehicle. As the company scales toward a target of 2,600 active robotaxis this year and tens of thousands by 2030, the standardized NVIDIA architecture facilitates cross-market validation and rapid deployment across WeRide’s footprint of 40 cities in 11 countries.