"The ChatGPT moment for physical AI is here when machines begin to understand, reason, and act in the real world." In a statement regarding the GPU manufacturer's most recent autonomous driving initiative, Nvidia (NVDA) CEO Jensen Huang threw down the robotic gauntlet at CES in Las Vegas.
For self driving cars and robotaxis, Nvidia's Alpamayo is a chain of thought, reasoning-based vision language action (VLA) paradigm that integrates observation, language, and action planning in decision-making, according to the company.
During his presentation, Huang showed a video of Alpamayo in operation. In it, a test car drove through San Francisco's streets without assistance, executing movements similar to those of a human driver.
The key question at hand is whether Nvidia has developed a system that is both better than Tesla's (TSLA) and comparable to Waymo's (GOOG, GOOGL) best in class robotaxis.
Huang is very optimistic about self-driving cars. According to the CEO, there will be one billion autonomous vehicles on the road in the future.
For more than ten years, Nvidia has been working on self driving technology. Huang forecast a "multitrillion dollar" opportunity for physical AI solutions, such as autonomous driving, last year.
Huang stated at CES that the next Mercedes CLA EV would be the first to use Nvidia's entire self-driving (FSD) stack, which includes Alpamayo in Q1.
By 2027, it intends to have autonomous robotaxis, similar to Waymo's, with partners including Lucid (LCID) and Uber (UBER).
Alpamayo is currently Level 2 advanced, which means that while it can function independently, human oversight is still necessary.
The objective of all the key participants in the autonomous arena is Level 4 autonomous, which means that the vehicles are completely self driving within a certain area.
Although Waymo has succeeded in certain regions, Tesla and Nvidia's DRIVE Hyperion systems are currently at Level 2. Nvidia plans to soon use Alpamayo to achieve Level 4 capability.
