European chipmakers Infineon, NXP and STMicroelectronics on Monday all announced partnerships with Nvidia to sell hardware for humanoid robots, as they vie for business in a potentially lucrative market.
U.S. chipmaker Nvidia coordinated the announcements on the eve of its annual GPU Technology Conference in California where its efforts to become the “brain”, or central computing platform for robots with its Jetson Thor processors, are likely to be one focus.
The role for Europe’s industrial chipmakers is to provide other parts of the body, including electronics needed to make them work safely and reliably, sensors, motion control, power management, and high-speed internal communications.
OVERLAP WITH TECH HARDWARE USED IN CARS
The chipmakers are all major suppliers of tech hardware used in cars, which analysts say has significant overlap with humanoid and other advanced robots.
George Chowdhury, an analyst at ABI Research, said it was natural for the Europeans to seek partnerships with Nvidia, estimating that Nvidia’s platform is used in more than 80% of humanoid robots. Chowdhury said higher-end humanoids can cost around $200,000, while lower-cost models can be priced at roughly a tenth of that. TrendForce estimates more than 50,000 humanoid robots will be sold this year for the first time.
A spokesperson for Infineon said the company anticipates a market of about $500 in parts per robot. Its pitch on Monday centred on the use of “digital twins” that allow developers to test and fine-tune robot performance in the design phase.
None of the announcements gave any financial details.
STMicroelectronics’ statement focused on sensors, helping robot developers connect cameras and motion sensors to Nvidia-based systems.
NXP emphasised fast, reliable communications inside the robot, helping different parts of the machine to send data quickly to the central processor so movement and sensing remain coordinated.
Gowri Chindalore, head of AI at NXP’s edge computing arm, gave the example of processing part of the information derived from voice commands instantly and then directing “further communication to the brain in a very fast loop”.
(Reporting by Toby Sterling)
