Telegram chief executive Pavel Durov announced a decentralized computing network called Cocoon at Blockchain Life 2025, positioning the messaging platform as a major player in distributed artificial intelligence infrastructure. The network will enable graphics processing unit owners to sell computing power for TON tokens while offering developers access to AI capabilities without routing user data through centralized systems.
What to Know:
- Cocoon allows GPU owners to contribute computing power and receive TON blockchain tokens as compensation
- Telegram will serve as the network's first major client, integrating AI features into its messaging app and mini-application ecosystem
- The platform launches in November 2025 with applications open to hardware providers and software developers
Decentralized Computing Meets Messaging
Durov described Cocoon, an acronym for Confidential Compute Open Network, as infrastructure that merges blockchain technology with artificial intelligence services. The system addresses privacy concerns that arise when users employ AI tools within messaging applications.
GPU owners can register their hardware to provide processing capacity to the network.
In return, they receive payment in TON, the blockchain token associated with The Open Network. Developers gain access to computing resources for AI applications at rates Durov suggested would undercut traditional cloud providers.
Telegram plans to integrate Cocoon's capabilities across its platform.
The company operates an ecosystem of mini-applications and automated bots that serve its user base. Features like message summarization and automated drafting will run through the decentralized network rather than servers controlled by a single entity.
The network is designed to prevent user data from passing through centralized AI providers. This approach responds to concerns about how technology companies handle private communications when offering AI-enhanced features.
Market Positioning and Technical Framework
Applications for both hardware providers and developers opened following the announcement. The November 2025 launch timeline gives the company roughly one month to finalize the infrastructure before deployment.
Durov characterized the pricing model as market-driven and transparent.
Competition among GPU providers should drive costs down while maintaining service quality. The decentralized structure also aims to resist content filtering or access restrictions that centralized platforms might impose.
The CEO positioned Cocoon within Telegram's stated commitment to user privacy. He cited growing consolidation in technology as motivation for creating alternative infrastructure that operates outside traditional corporate control.
Understanding the Technical Components
TON, or The Open Network, functions as a blockchain platform that processes transactions and executes smart contracts. Users exchange TON tokens for goods and services within applications built on the network. The blockchain's integration with Telegram gives it direct access to hundreds of millions of potential users.
Decentralized networks distribute computing tasks across multiple independent operators rather than routing everything through corporate data centers.
This structure can reduce single points of failure and limit any one entity's control over the system. GPU hardware accelerates the mathematical calculations required for AI processing, making these chips valuable for machine learning applications. Mini-applications run within Telegram's messaging interface, allowing users to access services without leaving the app. These programs range from games to financial tools, creating an ecosystem where Cocoon's AI capabilities could find multiple use cases.
Bottom Line
Telegram's Cocoon network represents the messaging company's expansion into decentralized AI infrastructure, with the platform serving as both developer and primary customer. The system's success will depend on whether it can attract enough GPU providers to meet computing demands while offering developers a viable alternative to established cloud services.