NVIDIA Unveils DRIVE PX Autonomous Driving Platform: Jensen Huang's Bold Foray into Automotive AI

On January 5, 2015, at the International Consumer Electronics Show (CES 2015) in Las Vegas, NVIDIA CEO Jensen Huang delivered an historic keynote that electrified the global automotive industry: the official unveiling of NVIDIA DRIVE PX, an automotive computing platform engineered exclusively for deep-learning-based autonomous vehicles. This move signaled NVIDIA’s decisive transition into reshaping the future of transportation through mobile supercomputing.

DRIVE PX: A 2.3 TFLOPS Supercomputer for Autonomous Vehicles

Legacy automotive electronics relied on fragmented, discrete Electronic Control Units (ECUs) managing localized sensor logic. NVIDIA DRIVE PX unified this paradigm with centralized compute:

1. Dual Tegra X1 Superchips in Parallel: Each Tegra X1 featured 256 Maxwell GPU cores paired with an 8-core 64-bit CPU cluster. Working in tandem, the dual processors delivered an astonishing 2.3 TFLOPS of deep learning throughput inside a vehicle chassis.
2. Ingestion of Up to 12 High-Resolution Camera Streams: With a video processing pipeline capable of handling up to 1.3 gigapixels per second, DRIVE PX ingested 360-degree sensor arrays simultaneously.
3. Deep Neural Network (DNN) Scene Semantic Understanding: Rather than relying on fragile hand-coded heuristics, DRIVE PX deployed deep convolutional networks to detect, track, and classify surrounding vehicles, pedestrians, cyclists, road markings, and road hazards under challenging night, rain, and glare conditions.

Autonomous Valet Parking and Audi Partnership

NVIDIA demonstrated an operational Auto-Valet Parking application: drivers exited their vehicle at a garage entrance and tapped a smartphone app; the vehicle navigated multi-level garages independently, parked in an empty bay, and subsequently returned autonomously upon summons.

Luxury automaker Audi collaborated onstage, showcasing advanced piloted driving prototypes driven by NVIDIA DRIVE silicon, demonstrating that AI was rapidly transitioning from research laboratories to public roadways.

Historical Perspective: Establishing the Global Standard in Automotive AI

The launch of DRIVE PX crystallized the global consensus that deep neural networks are fundamental to solving autonomous driving. This platform initiated the lineage running through DRIVE Xavier, DRIVE Orin, and the modern DRIVE Thor processor—cementing NVIDIA as the foundational compute silicon across global automakers and Robotaxi operators.