Cocoon enables secure decentralized AI compute built entirely on TON. Telegram becomes the first client to integrate Cocoon’s confidential compute services. Hardware providers earn TON tokens by contributing encrypted GPU resources.
Pavel Durov has officially introduced the Cocoon Network, a decentralized compute system built on the TON blockchain, during the Blockchain Life 2025 conference in Dubai. According to Durov, Cocoon, short for Confidential Compute Open Network, aims to create a privacy-focused infrastructure for AI inference and compute tasks while rewarding contributors with TON tokens.
Hardware providers such as GPU owners can contribute computing power and earn rewards in TON, while developers pay in the same token to access processing capacity for their AI models. The network is designed with a strong focus on confidential computing, ensuring all data remains encrypted during processing so that service providers cannot view user inputs or requests.
Cocoon’s official launch is scheduled for November 2025, and applications are already open for hardware providers and software developers seeking to participate. Telegram will serve as the first major client of the network, using Cocoon’s secure compute services to enhance its AI-driven mini-apps and bot features.
Telegram Integration and TON Expansion
This development links two rapidly growing sectors: blockchain infrastructure and decentralized AI compute. By handling encrypted data securely, Cocoon seeks to address increasing concerns over data privacy and centralized AI control.
The initiative could also strengthen TON’s ecosystem by broadening its use beyond messaging and payments. Every transaction on the Cocoon network will use TON, potentially increasing both token demand and real-world utility.
Additionally, Telegram’s integration offers Cocoon an immediate distribution advantage, granting access to nearly one billion users worldwide. This combination of blockchain security, decentralized infrastructure, and mass-market access positions Cocoon as a potential key player in the decentralized compute space.
While other blockchain-AI projects such as Render Network and Ankr have explored similar models, Cocoon’s direct link to Telegram and TON distinguishes it through scale and built-in user adoption. Industry observers will be watching closely to see how the network performs once it launches and whether it meets its goals of secure, private, and distributed computation.