Amazon Web Services will integrate Nvidia's NVLink Fusion technology into its Trainium4 chip, enhancing the performance of AI servers.

Amazon Web Services (AWS) has announced the integration of Nvidia's NVLink Fusion technology into future generations of its artificial intelligence chip, Trainium4, to enhance its offerings for AI customers. This technology will enable faster connections between processors, essential for training advanced models. The announcement was made at the annual AWS conference in Las Vegas, where Nvidia is already collaborating with other companies such as Intel and Qualcomm to promote NVLink.

AWS claims that NVLink Fusion will facilitate the construction of larger AI servers capable of communicating with low latency, which is essential for managing clusters of machines. Additionally, Amazon has launched servers based on the Trainium3 chip, which offers four times the processing power compared to the previous generation, while also reducing energy consumption by 40%.

The company also presented new AI models under the Nova brand, including Nova 2, which promises faster speeds and more accurate responses, as well as a multimodal model capable of generating various types of content. Although AWS has not managed to compete directly with leaders like ChatGPT, it reported a 20% increase in sales in the previous quarter, due to demand for cloud and AI solutions. Nova Forge was also introduced, a service that allows companies to build and train their own AI models based on internal data. Amazon's shares rose by 0.9% following these announcements.