Elon Musk to Train Grok Using SpaceX Engineering Data

Elon Musk announced that his artificial intelligence venture, xAI, will incorporate SpaceX engineering data into its upcoming model. The goal is to improve the problem solving and coding capabilities of the next Grok version. This training set will support a 2 trillion parameter model, which is a significant jump from the 1.5 trillion parameters found in the current architecture.

Musk noted that the data will exclude any information restricted by United States arms export laws. This compliance measure ensures that sensitive defense and space technology remains protected while still allowing the AI to learn from complex rocket and satellite engineering documents. This unique data set gives Grok a resource that competing firms like OpenAI and Anthropic currently lack.

This update follows a period of mixed reviews for the platform. While some developers claim that Grok now performs well on complex coding tasks, others have pointed to inconsistent benchmark results. The move to add specialized industry data aims to bridge the performance gap between Grok and top tier competitors.

Simultaneously, xAI has released a new tool called Grok for Excel. This allows users to build financial models and analyze data directly within spreadsheets using the Grok 4.5 engine. Investors are watching to see if these improvements will translate into higher adoption rates and if the new 2 trillion parameter model will meet the high expectations set by Musk.