Tensor has unveiled its personal Robocar at CES, positioning it as an AI-first autonomous vehicle built from the ground up for Level 4 driving rather than a conventional EV retrofitted with autonomy, Interesting Engineering reports.

The vehicle is powered by an in-vehicle supercomputer built around eight NVIDIA DRIVE AGX Thor system-on-chips based on the Blackwell GPU architecture.

The setup delivers more than 8,000 TOPS of GPU computing power, allowing the Robocar to process massive volumes of sensor data in real time.

Tensor frames the platform as a rolling data center, with autonomy as its core function rather than an add-on feature.

Tensor said the Robocar was designed specifically for private ownership while meeting commercial-grade autonomy requirements.

The Robocar architecture is designed entirely around AI control.

Safety is built around full drive-by-wire redundancy.

The deliveries are targeted for late 2026, with the first Lyft-ready markets planned for 2027, subject to regulatory approvals.