NVIDIA Corp is busily looking for ways to circumvent newly enacted rules barring export of computer chips and chip technology to the People’s Republic of China.
Nvidia Corp has begun offering an alternative to a high-end chip hit with US export restrictions to customers in China, after the new rules threatened to cost the American company hundreds of millions of dollars in lost revenue.
Nvidia said the new graphics-processing chip, branded the A800, meets US restrictions on chips that can be exported to China under new rules rolled out last month. The chip went into production in the third quarter, the company said.
On the other hand,
According to a memo Nvidia sent to its channel distributors last Thursday, the A800 has the same computational performance but a narrower interconnect bandwidth, the capacity of a chip to send and receive data from other chips, crucial for training large-scale AI models or building supercomputers.
It’s not the data rates, though, that matter; it’s the computational techniques and the technology used to implement those techniques that are important.
NVIDIA claims,
The A800 meets the US government’s clear test for reduced export control and cannot be programmed to exceed it.
This is disingenuous. The chip can be reverse-engineered to learn how the computational techniques are achieved. Indeed, simply programming the chip—accepting, arguendo, NVIDIA’s claim about programmability—would be a useless enterprise when the goal is to gain the technology itself.
It’s true enough that it takes some little time to relocate manufacturing/assembly sites and to move supply chains. However, why should NVIDIA or any American company, especially our technology-based companies, do business with any PRC company beyond a—adjusted apace—period of transition away from that nation?
Why would an American company be so willing to transfer, or risk transferring, American technology to an enemy nation by doing business with that nation or any business domiciled in it?