Foxconn building world's largest GB200 chip plant to keep up with Nvidia's demand

Officially announced on March 18th, 2024, during Nvidia's GTX 2024 keynote, the Blackwell GPU microarchitecture is entirely focused on generative AI applications. In the keynote speech, Nvidia CEO Jensen Huang mentioned that the new hardware is already supported and will be used in large-scale applications by giants such as Google, Microsoft, Meta, OpenAI, and Oracle. Now, Reuters reveals that Foxconn is building a factory large enough to meet the demand for these new chips.

According to the aforementioned source (the link can be found in the Source(s) section, as usual), a senior Foxconn executive talked earlier today about the fact that the company "is building the world's largest manufacturing facility making the Nvidia GB200 chip" but failed to give any details about its size or when it will start pumping out the highly-demanded silicon for next-gen AI apps. However, it should be noted that the Nvidia AI chips that will be used by Foxconn in its new facility come from TSMC's foundries.

After mentioning "the largest GB200 production facility on the planet," Foxconn senior vice president for the cloud enterprise solutions business group Benjamin Ting said that it's not the right time to say where yet. When switching the focus of the discussion on the demand for these pieces of hardware, he described it as "awfully huge" and, based on the massive AI investments revealed by the press in many areas of the world, he is not exaggerating at all.

At the end of this month, Amit Bahree will publish Generative AI in Action. This book that will hit Amazon in Kindle and paperback formats can be already pre-ordered starting at $43.99.