NVIDIA Expands DRIVE Hyperion Ecosystem for Level 4 Deployment

DRIVE Hyperion ecosystem expansion

At CES in Las Vegas, NVIDIA announced on January 5, 2026 that its global DRIVE Hyperion ecosystem is expanding with additional tier 1 suppliers, automotive integrators and sensor partners. The announcement positions DRIVE Hyperion as a production-ready, level 4 platform that can serve both autonomous passenger vehicles and long-haul commercial transport. The core message is that autonomy progress depends on a coordinated hardware and software ecosystem, not a single vehicle program.

NVIDIA names partners including Aeva, AUMOVIO, Astemo, Arbe, Bosch, Hesai, Magna, Omnivision, Quanta, Sony and ZF Group. It also says leading companies such as Astemo, AUMOVIO, Bosch, Magna, Quanta and ZF Group are building DRIVE Hyperion-based electronic control units, while AUMOVIO and several sensing vendors are qualifying sensor suites on the reference architecture. The company argues that this qualification path helps reduce integration and testing overhead for automakers.

Platform architecture and performance claims

NVIDIA describes DRIVE Hyperion as a compute-and-sensor reference architecture designed to make vehicles level 4-ready. According to the post, the platform uses two NVIDIA DRIVE AGX Thor systems-on-a-chip based on Blackwell and delivers more than 2,000 FP4 teraflops, or roughly 1,000 INT8 TOPS, for real-time sensor fusion across 360-degree perception. NVIDIA says this compute profile supports transformer-based perception, vision-language-action models and generative AI workloads in driving scenarios.

The release also links ecosystem growth to NVIDIA Halos, its safety and cybersecurity framework spanning data center to vehicle. NVIDIA says Halos supports inspection, validation and certification workflows, and when combined with simulation and AI data factory pipelines, enables continuous testing across large virtual and real-world scenario sets.

Why this matters for AI and mobility

The high-impact part of this update is the explicit attempt to standardize an autonomy stack across many suppliers at once. In autonomous driving, fragmented hardware and incompatible sensor pipelines are major delays. A common reference architecture can shorten validation cycles, improve reproducibility and lower engineering costs for OEMs and software developers.

NVIDIA also announced the Alpamayo family during CES and positioned it alongside Hyperion, signaling a strategy that combines foundation models, simulation and in-vehicle inference under one platform umbrella. If partner qualification and safety certification progress as described, this could accelerate deployment timelines for level 4 programs in both passenger mobility and freight. The next checkpoints are partner production launches, regulator-facing safety evidence and real-world reliability metrics over sustained fleet operation.