Think about the controversy with the american weapons sold outside the USA, which could potentially be remotely deactivated… What if Nvidia were to use the same method to continue exporting its AI chips on a massive scale to China? So far, the embargoes imposed on both China and Russia haven’t really worked. If a state with considerable financial power wants to obtain a banned product, no doubt it will be able to do so without too much difficulty. But this is becoming a crucial issue for Nvidia, as China generated $17 billion in revenues, or 13% of the company’s total sales last year.
Nvidia could (can?) disable an AI chip remotely
As a result, legislation is currently being drafted in the US that would force Nvidia to geolocate marketed chips, according to Reuters. As well as tracking chips after they have been put on the market, a key aspect of the legislation stipulates that if these chips are used for malicious purposes without proper authorisation, they should be prevented from functioning. In this way, the chips would be rendered inoperable remotely. The technologies used to achieve this aren’t new, as some have already been incorporated into other products. For example, Google is already tracking the location of its AI chips in its vast network of data centres for security purposes.
How and when such a provision might be applied remains the subject of another document. Wide-scale application, using secure connections to the company’s servers with appropriate position determination and permissions in line with current licensing, is seen as a complex procedure, particularly for products already on the market. Nevertheless, independent experts have stated that the technology to track chips after they have been sold is readily available and even largely already built into Nvidia chips.
