Foxconn and Nvidia join forces to build "artificial intelligence factory" to accelerate the development of autonomous driving technology

·Foxconn announced that it will use Nvidia chips and software to build new data centers for the development of autonomous vehicles, autonomous robots and industrial robots. Nvidia CEO Jensen Huang said that the two companies will jointly build these "artificial intelligence factories."

·The "artificial intelligence factory" called by Huang Renxun is expected to become a direct competitor of Tesla's Dojo supercomputer. Dojo is used to train Tesla's neural networks, which are used to support, train and improve its "Full Self-Driving".

Nvidia CEO Jensen Huang and Foxconn Chairman Liu Yongwei shared the stage at Foxconn’s annual technology showcase.

On October 18, Nvidia CEO Jen-Hsun Huang, wearing his iconic black leather jacket, appeared in Taipei again, this time with Foxconn Chairman Liu Yongwei at the Foxconn (Hon Hai Precision Industry Co., Ltd.) annual technology exhibition. tower.

Foxconn, the world’s largest electronics manufacturing outsourcer, announced that it will use Nvidia chips and software to build new data centers for the development of autonomous vehicles, autonomous robots and industrial robots. Huang Renxun said that the two companies will jointly build these "artificial intelligence factories" (AI factories).

"A new type of manufacturing has emerged - intelligent production. And the data centers that produce intelligence are artificial intelligence factories." Huang Renxun added that Foxconn has the expertise and scale to build these factories around the world.

Jensen Huang showed a hand-drawn sketch explaining how an “artificial intelligence factory” would continuously receive and process data from self-driving electric vehicles to make them smarter. He said that the entire end-to-end system is currently being built, on the one hand, building an advanced electric vehicle with an artificial intelligence brain inside, allowing it to drive autonomously and interact with the driver and passengers. On the other hand, it is supplemented by building an artificial intelligence factory to develop software for this car.

Jensen Huang showed a hand-drawn sketch explaining how an “artificial intelligence factory” would continuously receive and process data from self-driving electric vehicles to make them smarter.

"This car will go through its life and collect more data. This data will be sent to the artificial intelligence factory. The artificial intelligence factory will improve the software and update the entire artificial intelligence fleet." Huang Renxun said, "In the future, every company , every industry will have an artificial intelligence factory.”

The “artificial intelligence factory” that Jensen Huang calls is expected to become a direct competitor to Tesla’s Dojo supercomputer. Dojo is used to train Tesla’s neural networks, which are used to support, train and improve Full Self-Driving (FSD, Tesla’s driver assistance system). Tesla CEO Elon Musk hopes that FSD can eventually achieve true fully autonomous driving, which requires Dojo's powerful computing power. Tesla currently uses large GPU supercomputers based on Nvidia, but the new Dojo will reportedly be custom built using Tesla-designed chips.

Nvidia said in a statement that the artificial intelligence factory it will build will use Nvidia chips and software, including its cutting-edge GH200 super chip, which is banned from sale in China. It is worth noting that Nvidia said on October 17, local time, that the latest U.S. export restrictions will also prevent the company from producing two weaker high-end artificial intelligence chips (A800 and H800) for the Chinese market as well as a top-level game Chip sales.

Foxconn, the largest maker of Apple iPhones, hopes to replicate its success in assembling personal computers and smartphones to produce electric vehicles for other companies. This cooperation between Foxconn and NVIDIA is based on the partnership announced in January this year: the two parties will cooperate to develop an autonomous vehicle platform, and Foxconn will produce automotive electronic control units (ECUs) based on NVIDIA's DRIVE Orin chips for sale to the global market. .

As part of the collaboration, Foxconn has gradually rolled out off-the-shelf electric vehicle platforms for automakers to purchase. The company said the cars produced as a contract manufacturer will use Nvidia's Drive Hyperion 9 platform, which includes not only Drive Thor but also a suite of sensors such as cameras, radar, lidar and ultrasound that are essential for self-driving capabilities. necessary.

Liu Yongwei, standing next to Huang Renxun, said that Foxconn "is trying to transform from a manufacturing services company to a platform solutions company." Foxconn’s primary goal is to launch three platforms: smart electric vehicles, smart cities and smart manufacturing.

On the same day, Foxconn launched a new electric truck called Model N, which is the sixth prototype in its electric vehicle promotion plan. The company's initial goal is to capture 5% of the global electric vehicle market by 2025 and generate $33 billion in revenue from manufacturing electric vehicles and components. The long-term goal is to produce nearly half of the world's electric vehicles, which is an ambitious goal. exuberant.

Jun Seki, head of Foxconn’s electric vehicle business, said the company is in talks with 14 potential customers and called India and Japan countries with promising prospects for the development of electric vehicles.

The above is the detailed content of Foxconn and Nvidia join forces to build "artificial intelligence factory" to accelerate the development of autonomous driving technology. For more information, please follow other related articles on the PHP Chinese website!