J.P. Morgan sees Nvidia pivot to Rubin trays
According to an analysis by J.P. Morgan, Nvidia may be shifting its strategy to capture a larger portion of the artificial intelligence server market. The financial firm reported that Nvidia could move beyond selling individual graphics processing units to providing nearly complete AI systems. This initiative would reportedly begin with its upcoming Vera Rubin platform, where the company might supply pre-assembled L10 compute trays containing the core server components.
This approach would centralize the most complex engineering, including the integration of the Vera central processing unit, Rubin GPUs, memory, and liquid-cooling plates. Partners would then focus on integrating these ready-made compute trays into server racks and managing system-level validation. Such a move is seen as a way for Nvidia to ensure consistent quality and accelerate deployment for hyperscale clients, while also securing more revenue.
This potential shift, however, would redefine the role of its hardware partners, reducing their design responsibilities and potentially limiting product differentiation. While not officially confirmed, this strategy could significantly reshape the AI hardware supply chain.
