Elon Musk‘s startup, xAI, is nearing one of its most important deals – a one-year partnership with Telegram worth $300 million. If the deal goes through, the Grok chatbot will go directly into the app, which is used by more than one billion users worldwide. For xAI, this isn’t just huge exposure – it’s also a potential source of training data at a time when it’s becoming increasingly difficult to acquire.
According to Telegram’s announcement, the deal includes a split of subscription revenues: half will go to the platform, the other half will stay with xAI. The $300 million – paid in cash and shares – is to be a form of down payment, but formal details are still pending. Elon Musk is dampening enthusiasm, saying that the deal has not yet been signed, although the intentions are clear.
There is much more to the background of this collaboration than just the expansion of Grok. Telegram, which has so far successfully balanced openness and privacy, is, according to the announcement, to be a mere medium – xAI is only to have access to data that users openly share in their interactions with the chatbot. However, this does not change the fact that this is an opportunity for Musk’s company to build an edge in a key area: data to train AI.
The depletion of open source resources and increased regulatory pressure around the use of public content (e.g. by the Met) makes access to legitimate, fresh and contextual data one of the most valuable currencies in the AI race. Integration with Telegram could give xAI not only reach, but billions of real-world, conversational interactions – far more valuable than anonymous posts from open repositories.
From Telegram’s point of view, it’s an opportunity to monetise without having to change its business model or violate user privacy. And for xAI, another step towards becoming independent of X Resources and building its own ecosystem of data and users.
If the deal goes through, it could become a model for other AI-communication partnerships. Telegram’s Grok is not just a feature – it is a strategic move that could define how language models are developed and shared in the coming years.
