Nvidia has announced a new GPU tracking technology that can help identify the country where its AI chips are operating. The development was reported by Reuters, citing sources familiar with the matter.

The technology has been revealed in private demonstrations over recent months. However, Nvidia has not officially released the feature yet. It is designed as an optional software update that customers can install.

The system uses confidential computing capabilities built into Nvidia’s graphics processing units. As a result, it can monitor performance and estimate location data. It does this by measuring time delays when communicating with Nvidia-controlled servers.

Moreover, the software was created to help customers track overall computing performance. This practice is already common among companies operating large data center processor fleets. According to an Nvidia official, the location estimation is similar in accuracy to other internet-based location services.

In addition, Nvidia stated that it is “in the process of implementing a new software service” to help data center operators monitor the health and inventory of their AI GPU fleets. The company said the customer-installed software agent uses GPU telemetry to track fleet health, integrity, and inventory.

The feature will first be made available on Nvidia’s latest Blackwell chips. These chips include stronger security features for a process known as attestation. Furthermore, they offer more protection than the older Hopper and Ampere generations.

Meanwhile, Nvidia is examining options to extend this GPU tracking technology to prior chip generations. This includes both Hopper and Ampere semiconductors.

Overall, the announcement highlights Nvidia’s efforts to strengthen export compliance and limit the smuggling of AI chips into countries where exports are restricted.

Related post

View all