During its earnings call, NVIDIA disclosed that it has provided samples of the Vera Rubin platform—a system tailored for next-generation AI data centers—to a select group of customers. The platform is projected to be officially rolled out in the latter half of 2026 or at the beginning of 2027. The Chief Financial Officer mentioned that the initial batch of samples was dispatched this week, with intentions to initiate mass production and delivery in the latter half of this year. At present, the performance and energy consumption criteria for the Vera Rubin platform have been predominantly established. As the forthcoming architecture for AI data centers, it encompasses a range of components. NVIDIA's partners are required to proactively adjust their software and hardware configurations, with certain suppliers slated to obtain NVL72 VR200 full rack systems, whereas hardware producers like Foxconn will be furnished with chip samples. According to industry speculation, NVIDIA is also set to provide complete computing trays with pre-installed parts. Kress highlighted that the Rubin platform boasts enhanced reliability and ease of maintenance, and it is expected that every cloud service provider will implement it.