Telegram launches Cocoon, a decentralized confidential AI network on TON
Telegram founder Pavel Durov has announced the launch of Cocoon, a decentralized confidential computing network designed to process AI requests with full privacy. Built on the TON blockchain, the platform connects GPU owners with applications that require secure execution of AI models.Private AI requests secured through decentralization
According to Durov, Cocoon already handles the first wave of user AI queries with what he describes as “100% privacy.” The system ensures that requests are not exposed to centralized servers, but instead routed through a network of independent GPU operators participating in the TON ecosystem.The platform is positioned as a response to rising concerns over confidentiality in cloud-based AI processing. By decentralizing computation, Cocoon aims to offer stronger guarantees that sensitive information remains protected while still supporting high-performance model execution.
How Cocoon works: client, proxy and worker
The Cocoon architecture consists of three core components. The client submits an AI request and pays for processing through TON-based transactions. A proxy node then selects an appropriate worker node based on availability, performance and confidentiality requirements. Finally, the worker executes the request using its GPU and returns the result through the proxy.This design allows GPU owners to participate directly in the network and earn TON for providing computational power. For developers and end users, the model offers a decentralized alternative to traditional cloud infrastructure while maintaining predictable workflows and latency.
Incentives for GPU owners and AI developers
Cocoon enables individuals and organizations with spare GPU capacity to monetize their hardware. As AI workloads expand globally, decentralized compute markets such as Cocoon provide new earning models and help distribute processing power away from centralized providers.For AI developers, the ability to run models without exposing data to third-party servers is a significant advantage. Cocoon’s confidential execution environment may appeal to applications involving medical data, finance, identity processing and other sensitive domains.
A strategic expansion of the TON ecosystem
The launch of Cocoon marks a notable step in TON’s evolution toward supporting large-scale decentralized AI workloads. By combining blockchain-based coordination with off-chain GPU compute, the network aims to position itself as a foundation for privacy-first AI applications.Analysts view Cocoon as part of a broader movement toward distributed inference markets, where privacy, cost efficiency and censorship resistance play increasingly important roles. Telegram’s endorsement adds momentum to the idea that decentralized compute layers may become essential infrastructure for next-generation AI systems.
Editorial Team - CoinBotLab
🔵 Bitcoin Mix — Anonymous BTC Mixing Since 2017
🌐 Official Website
🧅 TOR Mirror
✉️ [email protected]
No logs • SegWit/bech32 • Instant payouts • Dynamic fees
TOR access is recommended for maximum anonymity.
🌐 Official Website
🧅 TOR Mirror
✉️ [email protected]
No logs • SegWit/bech32 • Instant payouts • Dynamic fees
TOR access is recommended for maximum anonymity.