China's Moonshot AI reportedly used Nvidia Blackwell chips for training Kimi K3 — company circumvented both U.S. export and Chinese import controls to acquire compute. This development highlights the growing competition in the AI sector, where access to advanced hardware is a key strategic asset. The U.S. government has expressed concerns over such actions, emphasizing the need for stricter oversight of AI-related technologies.

Moonshot AI's acquisition of Nvidia Blackwell chips has raised questions about the effectiveness of current export and import controls. The company reportedly used a combination of direct purchases and cloud-based training to bypass these restrictions. This approach underscores the challenges faced by regulators in controlling the flow of high-performance computing resources across borders.

According to reports, Moonshot AI has reportedly made use of 92 instances of Nvidia Blackwell chips in its efforts to train the Kimi K3 model. This number highlights the scale of the company's investment in advanced computing infrastructure. The use of such a large number of chips suggests a significant commitment to enhancing the capabilities of its AI models.

The circumvention of export and import controls by Moonshot AI could have broader implications for the global AI landscape. It may lead to increased costs for other companies seeking to acquire similar technology, as well as potential vendor lock-in with Nvidia. Additionally, such actions could complicate governance frameworks aimed at regulating the use of AI technologies on a global scale.

The situation remains under active scrutiny, with ongoing discussions about the need for more robust international agreements on AI hardware distribution. As the AI arms race continues, the ability to secure and control advanced computing resources will likely play a central role in determining the competitive advantage of key players in the sector.