Telegram Launches Cocoon: Decentralized AI on TON

Telegram Launches Cocoon: Decentralized AI on TON
This article was prepared using automated systems that process publicly available information. It may contain inaccuracies or omissions and is provided for informational purposes only. Nothing herein constitutes financial, investment, legal, or tax advice.

Introduction

Telegram has launched Cocoon, a revolutionary decentralized AI compute network built on the TON blockchain that pays GPU owners in Toncoin cryptocurrency for processing private AI tasks. Announced by CEO Pavel Durov at Blockchain Life 2025 in Dubai, this initiative directly challenges the dominance of Big Tech cloud providers like AWS and Microsoft Azure while promising unprecedented privacy protection for Telegram’s billion-plus users through confidential computing technology.

Key Points

  • Cocoon creates a marketplace where GPU owners earn Toncoin for providing computing power to private AI inference tasks
  • Telegram will integrate Cocoon across its ecosystem, powering AI features in Mini Apps for its billion-plus user base
  • The network uses confidential computing to keep user data encrypted throughout processing, even from GPU providers

A New Marketplace for Decentralized AI Compute

Cocoon, formally known as the Confidential Compute Open Network, creates a groundbreaking marketplace where individuals can monetize their graphics processing units (GPUs) by contributing computing power to the network. In exchange, GPU owners receive Toncoin cryptocurrency, the native token of the TON blockchain. This creates a new revenue stream for hardware owners while simultaneously building a distributed infrastructure for AI computation. The network is scheduled to launch in November, with applications for both GPU providers and developers opening immediately following the announcement.

On the demand side, developers gain access to low-cost AI infrastructure that processes queries without exposing user data to centralized providers. This dual-sided marketplace operates on supply and demand dynamics, allowing GPU owners and developers to negotiate costs directly through the network. This market-driven approach has the potential to create more competitive pricing than traditional cloud providers typically offer, though it introduces questions about service reliability and consistency that centralized providers have historically addressed through service level agreements.

Privacy-First Architecture Challenges Big Tech Dominance

Cocoon’s confidential computing approach represents a fundamental shift from how major AI systems currently operate. When companies like OpenAI or Google process user queries, they typically have access to everything—user prompts, data, patterns, and metadata. In contrast, Cocoon keeps this information encrypted throughout the entire computation process, even from the GPU owners performing the actual calculations. This privacy-first architecture aligns with both Telegram’s and the TON blockchain’s foundational principles of user control and data protection.

Telegram CEO Pavel Durov positioned the project as a direct response to eroding digital freedoms and the monopoly that Big Tech corporations like Amazon’s AWS and Microsoft Azure have established in cloud computing and AI infrastructure. The timing coincides with growing global concerns about centralized AI systems and their data collection practices. Max Crown, CEO of the TON Foundation, emphasized that Cocoon represents a shift toward an open, user-driven compute economy that returns control of AI infrastructure to users rather than corporations that exploit it.

The project builds on TON blockchain’s multi-chain architecture, which processes millions of transactions per second through its sharded design. This technical foundation allows Cocoon to handle massive AI workloads while maintaining the transparency and market-driven pricing that blockchain technology enables. The structure positions TON as an open, user-controlled alternative to traditional cloud infrastructure providers.

Immediate Scale and Strategic Backing

Telegram’s globally adopted platform provides Cocoon with immediate scale that most decentralized projects struggle to achieve. The messaging app, with its billion-plus user base, will integrate Cocoon across its ecosystem, powering AI features in its Mini Apps and potentially transforming how users interact with artificial intelligence in their daily digital lives. Telegram will serve as the network’s first major customer, ensuring immediate utilization and testing at scale.

The project has attracted significant institutional backing, with AlphaTON Capital—a Nasdaq-listed digital asset infrastructure and TON treasury company—announcing its intention to make a substantial investment in all the hardware required to support the network. The company plans to deploy next-generation, high-memory GPU models across strategic data centers, supporting advanced model architectures including DeepSeek and Qwen. This institutional support provides crucial infrastructure stability during the network’s early growth phase.

Cocoon enters a growing market of decentralized AI networks that includes established players like Akash Network and Render Network, which already enable users to rent distributed computing resources. However, Cocoon’s integration with Telegram’s massive user base represents a significant competitive advantage that could accelerate adoption beyond what previous decentralized compute projects have achieved. As Max Crown noted in the press release, Cocoon represents the convergence of social networking, AI, and decentralized technology at unprecedented scale.

Notifications 0