In a bid to boost EV and autonomous vehicles production, NVIDIA and Foxconn have joined forces to build ‘AI factories’. No, they won’t be churning out robots dressed in silicon suits, but they will be revolutionising the manufacturing landscape in a way that’s nothing short of extraordinary, bringing in electric cars, and training autonomous vehicles. 

“A new type of manufacturing has emerged – the production of Intelligence. And the data centres that produce it are AI factories,” said Jensen Huang in a statement. He further adds that, “Foxconn, the world’s largest manufacturer, has the expertise and scale to build AI factories globally.”

This partnership was unveiled during Foxconn’s annual tech day in Taipei, where Huang made a surprise appearance. Foxconn’s CEO Young Liu drove to the stage in their new electric, and Huang said, “This car is an ideal vehicle for a young couple like Young and me.” 

The term ‘AI factory‘ is a metaphorical expression of a new era of manufacturing, one that harnesses unstructured data from an array of sources, including automobiles, robots, and IoT sensors, and transforms it into valuable insights and products. 

The presence of Huang on the stage highlights that NVIDIA, known for its prowess in GPUs and accelerated computing, is the wizard behind the curtain. These AI factories are essentially massive data centres equipped with high-performance GPUs, ready to turn raw data into actionable intelligence.

Foxconn, the Taiwanese electronics giant responsible for producing gadgets for tech giants like Apple, is the one building these AI factories.

New foray for both the partners

These AI factories will be powered by NVIDIA HGX, GH200 SuperChips, and OVX reference designs, along with Mellanox-derived networking equipment. Foxconn plans to employ these systems across a range of products and services built upon NVIDIA’s enterprise software platforms, such as DGX.

One such endeavour is the Smart EV platform, using NVIDIA’s Drive Thor autonomous driving computer system, set to hit the market in 2025. Foxconn has already been using the Drive Orin platform to create electronic control units for automobiles.

Interestingly, Foxconn does not want to produce its own EVs, but wants to partner with other car makers, and mass produce vehicles on their behalf. Looks like Foxconn is looking to find its Apple in the EV industry. But as Tu Le, analyst and managing director from Sino Auto Insights said, “Foxconn is trying to enter a market they have absolutely no experience in.” He calls it a very daunting task for them. 

But Foxconn does not stop here. Beyond electric vehicles, Foxconn is also eyeing “smart manufacturing” applications that leverage NVIDIA’s Isaac autonomous robotics platform. Additionally, they have their sights set on creating a “smart city” platform, integrating NVIDIA’s Metropolis video analytics service. However, the realisation of “smart cities” has been somewhat elusive, with progress lagging behind the hype.

What’s in it for NVIDIA?

NVIDIA’s influence goes beyond providing hardware. They are making strides in the software domain as well. The OVX systems, designed to power the Omniverse platform, are key to visualising machine learning and sensor data. They allow for the creation of digital twins of actual factory floors, enabling simulations to test operational changes before implementing them in the real world.

It’s like a sophisticated game of SimCity but for industry.

The company has been working diligently to expand the horizons of accelerated computing beyond the conventional domains of hyperscale and cloud data centres. But the concept of AI factories for NVIDIA, is just another buzzword for building AI.

Last year, they were talking about the “industrial metaverse“, in collaboration with Siemens, to bring photorealistic digital twins to mainstream industry using their Omniverse platform. However, it seems the allure of the metaverse has dimmed, and they’ve pivoted toward AI factories.

As Huang said in a CNBC interview that the next big invention for NVIDIA would be to make “AI meet the physical world”, and it seems like with Foxconn, that dream is taking place sooner than expected. The company recently also expanded its Jetson platform for edge AI and robotics, bringing generative AI into the real world, and out of the omniverse. 

It seems like this partnership between NVIDIA and Foxconn will not only process data but also help train autonomous machines, making production smarter and more efficient. Moreover, as Huang said, if successful, Foxconn would be able to make robots, and eventually foray into self-driving cars.