Telegram Launches Cocoon Confidential Compute Network for Secure AI Processing

Telegram Launches Cocoon Confidential Compute Network for Secure AI Processing

Telegram’s new Cocoon platform processes AI requests with full privacy, rewards GPU providers with TON tokens, and will introduce additional AI features in the coming weeks.

TON

Fact Check
The evidence strongly supports the truthfulness of the statement. Two independent sources with medium-to-high authority and high relevance directly corroborate the key claims. One source confirms the launch of a 'Confidential Compute Open Network' as part of Telegram's expansion into AI. A second source explicitly names this platform the 'Confidential Compute Open Network (COCOON)' and confirms it is a decentralized AI platform associated with Telegram's founder. The combined information from these sources directly validates every component of the statement: the entity (Telegram), the name of the network ('Cocoon Confidential Compute Network'), and its purpose (secure AI processing). Other high-authority sources are either irrelevant to this specific launch or discuss unrelated security topics, offering no contradictory evidence. The consistency and directness of the supporting evidence result in a high confidence assessment that the statement is likely true.
    Reference123
Summary

Telegram founder Pavel Durov announced the launch of Cocoon, a decentralized confidential computing AI network built on The Open Network (TON). The system processes users’ initial AI requests privately, while GPU contributors are rewarded with TON tokens. Telegram plans to roll out new AI capabilities in the coming weeks, reinforcing its strategy to combine privacy-preserving AI with blockchain-based incentives.

Terms & Concepts
  • TON tokens: The native cryptocurrency of The Open Network (TON), used for transactions and incentives within the ecosystem.
  • GPU: Graphics Processing Unit, a hardware component optimized for parallel computations, widely used in AI model training and inference.