Jensen Huang thinks it will take 20 years to achieve chip independence despite booming fabrication construction and subsidies. (Image credit: Nvidia)
Nvidia has been in the news as of late. Not to quash any leaked rumors surrounding the upcoming RTX 5000 series of GPUs, but rather the export restrictions imposed on the chip maker by the US government. Following the initial wave of sanctions in 2022 regarding sending advanced AI chips to China, which the US government states could find their way into military weapons, the Secretary of Commerce has issued Nvidia a warning against exporting revisions designed to skirt those sanctions.
Gina Raimondo issued a cautionary statement that any new product representing a redesign of a previously prohibited GPU would promptly become subject to additional restrictions. She did not identify what chips those might be, but all of Nvidia’s RTX series GPUs feature Tensor cores, which enable acceleration of AI workloads. “If you redesign a chip around a particular cut line that enables them to do AI, I’m going to control it the next day,” stated Gina at the Reagan National Defense Forum in Simi Valley, California.
The US imposed restrictions back in August of 2022 on Nvidia’s A100/H100 chips and their exportation to China. In November of that year, Nvidia introduced the A800/H800 for the Chinese market, which was subject to restrictions in October of this year. Since then, Nvidia has launched their HGX H20, L20 and L2 chips and is currently prepping their RTX 4090D for shipment, drawing the scrutiny of the federal government. Nvidia is currently shipping those banned chips to other Western countries, which are exempt from the ban, as a method for navigating around those sanctions.
Now, Nvidia CEO Jenson Huang has released a statement saying that it might take 20 years for the US to become self-reliant on chip manufacturing, weening itself from manufacturers in Taiwan and China. The US has introduced initiatives and incentives over the last few years to build plants in the US and become a dominant player in the semiconductor industry. Speaking at the New York Times’s DealBook conference in New York, Jenson stated, “We are somewhere between a decade and two decades away from supply chain independence. It is not a really practical thing for a decade or two.”
So far, those incentives seem to be working, as Intel, TSMC, and Samsung Foundry are building leading-edge fabs in the US, and Micron is set to bring its advanced 3D NAND and DRAM production to America as well. Huang noted that Nvidia’s chips rely on components sourced from different parts of the world and doesn’t see them being produced in the US any time soon.
Have a story tip? Message me at: http://twitter.com/Cabe_Atwell