Foxconn and Nvidia are construction ‘AI factories’ to boost up self-driving vehicles | TechCrunch

Symbol Credit: FoxconnNvidia and Foxconn are running in combination to construct so-called “AI factories,” a brand new elegance of knowledge facilities that promise to supply supercomputing powers to boost up the improvement of self-driving vehicles, self sustaining machines and commercial robots.
Nvidia founder and CEO Jensen Huang and Foxconn chairman and CEO Younger Liu introduced the collaboration at Hon Hai Tech Day in Taiwan on Tuesday. The AI manufacturing facility is founded off an Nvidia GPU computing infrastructure that will probably be constructed to procedure, refine and grow to be huge quantities of knowledge into treasured AI fashions and data.
“We’re construction this whole end-to-end method the place at the one hand, you’re construction this complicated EV automotive…with an AI mind within that permits it to have interaction with drivers and engage with passengers, in addition to autonomously pressure, complemented by means of an AI manufacturing facility that develops a instrument for this automotive,” stated Huang onstage on the tournament. “This automotive will undergo lifestyles revel in and acquire extra information. The knowledge will pass to the AI manufacturing facility, the place the AI manufacturing facility will reinforce the instrument and replace all the AI fleet.”
The AI manufacturing facility tie-up builds off a partnership between Nvidia and Foxconn introduced in January to increase self sustaining automobile platforms. That settlement concerned Foxconn turning into a number one provider of digital keep watch over gadgets (ECUs) for automakers, which will probably be constructed with Nvidia’s Force Orin system-on-a-chip (SoC), a supercomputing AI platform that helps self sustaining riding purposes. On Tuesday, Foxconn additionally dedicated to production ECUs with Force Thor, Nvidia’s next-gen SoC, after manufacturing begins in 2025.
As a part of that partnership, Foxconn — which has been frequently unveiling off-the-shelf EV platforms for automakers to buy — stated the cars it makes as a freelance producer will probably be constructed with Nvidia’s Force Hyperion 9 platform, which contains no longer simplest Force Thor, but in addition a set of sensors like cameras, radar, lidar and ultrasonic which are important for self-driving features.
Foxconn is already shrunk to construct EVs for Fisker, even because it will get sued by means of its erstwhile spouse Lordstown Motors. The automaker will want scale so as to make its AI factories viable, particularly if it’s going to compete with Tesla.
As a result of those AI factories are necessarily competitors to Tesla’s Dojo supercomputer, which the Elon Musk-owned automaker began manufacturing on over the summer season. Dojo will teach Tesla’s neural nets, that are used to energy, teach and reinforce “complete self-driving” (FSD), the automaker’s complicated driving force help method. Musk hopes FSD will in reality be totally self riding someday, which is the place the robust compute of Dojo is available in.
Tesla lately makes use of a big Nvidia GPU-based supercomputer, however the brand new Dojo will probably be custom-built the use of chips designed by means of Tesla. The Foxconn-Nvidia AI factories will probably be in keeping with Nvidia’s GH200 Grace Hopper Superchip and AI Undertaking instrument, in line with the corporate.
The factories can have different programs past self-driving vehicles.
On degree at Hon Hai’s tech tournament, Liu shared Foxconn’s function to transform itself “from a producing carrier corporate to a platform answers corporate” by means of scaling the AI factories throughout quite a lot of industries. Foxconn is concentrated on 3 platforms to begin: Good EVs, sensible towns and sensible production.
“This can be a manufacturing facility that takes information enter and produces intelligence as an output,” stated Huang, as Liu nodded his assent. “Someday, each business, each corporate can have an AI manufacturing facility.”

Author: OpenAI