Space data pours into AI

Space data hits AI! Musk took out his SpaceX fortune and the 2 trillion-parameter Grok model was instantly built. In order to build stronger artificial intelligence, Musk once again demonstrated his ability to integrate resources across borders. He recently publicly announced that the top engineering data accumulated by SpaceX since its establishment in 2002 will be officially used to train the next generation Grok large model. Eliminating sensitive information to ensure compliance. The scale of data participating in this training is unprecedentedly large, covering many core areas such as manufacturing processes, material science, and Starlink hardware design. However, due to U.S. export control regulations, defense-sensitive technologies such as rocket engines and guidance control have been strictly excluded. It is understood that the number of parameters of the upcoming new model has reached an astonishing two trillion, doubling the size of the just-launched Grok 4.5. This massive data project is expected to complete its final training this week and demonstrate inferencing capabilities beyond what has ever been achieved before. Multi-party collaboration to build a data empire This cross-border linkage is fully in line with Musk's consistent data collaboration strategy. Prior to this, Tesla provided real driving and manufacturing experiences, and social platforms contributed a large amount of daily conversation material. Now with the addition of aerospace engineering data, the Grok large model will have extremely unique capabilities and advantages. From cars to social networking to rockets, Musk is step by step transforming the core assets of his empire into a moat in the AI ​​field. via AI News (author: AI Base)