Foxconn, a contract manufacturer known for assembling the iPhone, is making a major move into AI hardware. The company has announced that it’s building a new manufacturing facility in Mexico, which will produce Nvidia high-performance GB200 superchips. These chips are a vital component of Nvidia’s upcoming Blackwell platform, designed to meet the growing demand for powerful AI servers and high-performance computing.
The expansion into AI hardware marks a notable shift for Foxconn, traditionally focused on consumer electronics. With the new facility, set to be the largest of its kind, the company will help address the increasing computational needs of industries using AI, from cloud computing to large-scale data processing.
The GB200: A New Class of Computing Power
Nvidia’s GB200 superchip isn’t your typical GPU. It pairs two Blackwell B200 GPUs with a Grace CPU, combining them into one unified system with nearly a terabyte of memory. This design, which features Nvidia’s NVLINK interconnect technology, is optimized for AI workloads, such as training models and running inference tasks on massive datasets.
What makes the GB200 stand out is its ability to scale. The chip can be deployed in configurations that bring together hundreds of GPUs into a single system, making it a powerhouse for data centers and AI researchers. By building these superchips at scale, Foxconn is positioning itself to supply the growing number of companies building out AI infrastructure.
Foxconn’s Growing Footprint in Mexico
Foxconn’s choice of Mexico for this new facility makes sense considering its existing operations in the country. The company has been investing heavily in Mexico, particularly in Chihuahua, where it has already spent over $500 million on expanding its manufacturing presence. By adding the GB200 production line, Foxconn is doubling down on its strategy to diversify its product portfolio and tap into new markets.
Chairman Young Liu emphasized that the capacity of the Mexico plant would be “enormous,” although no specific details about the timeline or production volume were given. The new facility will bolster Foxconn’s role in the AI hardware supply chain at a time when companies around the globe are racing to secure the resources needed to support their AI-driven businesses.
AI Servers and Supercomputing: What’s Next for Foxconn and Nvidia
Beyond its production of GB200 chips, Foxconn is also working closely with Nvidia on a major project in Taiwan. The two companies are building Taiwan’s largest supercomputer, which will feature over 4,600 Tensor Core GPUs based on the same Blackwell architecture. The system, which is expected to deliver more than 90 exaflops of computing power, will be one of the most powerful in the world when it goes online in 2025.
Putting a focus on AI servers and supercomputing shows how both companies are betting on the future of artificial intelligence. As more industries integrate AI into their operations, the demand for servers capable of handling huge amounts of data is only going to increase. Foxconn’s role as a manufacturer of these systems puts it in a strong position to meet that demand.
Diversifying Beyond Consumer Electronics
The move into AI hardware isn’t the only way Foxconn is looking to expand its business. It has also been making waves in the electric vehicle (EV) sector. Through its subsidiary Foxtron, Foxconn is developing vehicles for contract manufacturing, aiming to take advantage of its deep expertise in electronics assembly to break into the auto industry.
Foxconn’s strategy to diversify its product lines is driven by a desire to move beyond its reliance on building consumer electronics for companies like Apple. By focusing on high-growth sectors like AI and EVs, the company hopes to secure its future in a rapidly changing tech landscape.
Last Updated on October 14, 2024 12:02 pm CEST