The project, called Cocoon, aims to create a decentralized network for artificial intelligence that keeps user data private while connecting people who own powerful computers with developers who need computing power.

Telegram founder Pavel Durov announced a major new project on October 29, 2025, at the Blockchain Life 2025 conference in Dubai.

Cocoon stands for Confidential Compute Open Network. It works by allowing people who own graphics processing units (GPUs) to share their computer’s processing power with the network. In return, they earn Toncoin (TON), which is the cryptocurrency used on The Open Network blockchain. On the other side, developers and companies can use this computing power to run AI programs without relying on big tech companies like Amazon or Google.

How Cocoon Works

The system operates like a marketplace for computing power. GPU owners connect their hardware to the network and get paid in TON tokens when their computers process AI tasks. Developers pay for this service using the same TON tokens, creating a direct exchange between the two groups.

What makes Cocoon different from traditional AI services is its focus on privacy. All data processed through the network stays encrypted, which means even the GPU owners providing the computing power cannot see what information they are processing. This approach addresses growing concerns about how centralized AI companies handle user data.

Source: @ton_blockchain

Durov explained his motivation for creating Cocoon during his presentation. “The world has been moving towards a weird direction. For the last 20 years, we’ve been gradually losing our digital freedoms,” he said, emphasizing why decentralized AI matters for protecting individual privacy.

Telegram’s Role and Launch Timeline

Telegram will be the first major customer of Cocoon. The messaging app plans to integrate the network into its mini-apps and bot ecosystem starting in November 2025. With over one billion users worldwide, Telegram’s adoption could quickly bring decentralized AI to hundreds of millions of people.

The company will use Cocoon to power AI features like message summarization and draft writing. Instead of sending user data to centralized AI providers, these functions will run on the decentralized network, keeping conversations private.

Applications for both GPU providers and developers are already open. GPU owners need to provide details about their hardware, including the number of GPUs they have, the model type, memory capacity, and how often the hardware will be available. Developers must specify which AI models they plan to run, such as DeepSeek or Qwen, along with their expected usage volume.

Major Investment from AlphaTON Capital

Shortly after Durov’s announcement, AlphaTON Capital, a Nasdaq-listed company, revealed plans to invest heavily in GPU infrastructure to support Cocoon. The company will deploy high-performance GPUs across data centers worldwide to ensure the network has enough computing capacity from day one.

Brittany Kaiser, CEO of AlphaTON Capital, called the launch “a pivotal moment in the convergence of blockchain technology, data protection and artificial intelligence.” The investment shows serious industry support for the project even before its official launch.

Enzo Villani, Executive Chairman of AlphaTON Capital, added that “privacy and security are paramount in the AI era” and that Cocoon’s architecture addresses a critical need in the market.

Why Decentralized AI Matters

Centralized AI systems pose several risks that Cocoon aims to solve. When companies like OpenAI or Google process AI queries, they can see everything users input, including private information, personal data, and search patterns. This creates privacy concerns and makes user data vulnerable to breaches or hacks.

Additionally, centralized providers could change their algorithms without users knowing, potentially manipulating information or censoring content. Blockchain technology can help verify that AI-generated data remains accurate and unchanged by creating permanent records on a decentralized ledger.

Cocoon positions itself as a challenger to the dominance of Big Tech infrastructure services like Amazon Web Services and Microsoft Azure. By creating a market-driven infrastructure that promotes transparent pricing, the network could offer more competitive rates than traditional cloud providers.

Connection to Kazakhstan AI Lab

The Cocoon announcement follows recent developments in Telegram’s AI initiatives. On October 2, 2025, President Kassym-Jomart Tokayev of Kazakhstan met with Durov at the Digital Bridge 2025 forum to discuss cooperation in AI and cybersecurity. The two discussed plans for Telegram to establish an AI laboratory in Kazakhstan, working with the country’s supercomputing infrastructure.

This collaboration shows Durov’s commitment to building AI technology outside the control of Western tech giants. By partnering with different countries and creating decentralized networks, Telegram is pursuing an alternative path for AI development.

Market Response and TON Ecosystem

Following the announcement, Toncoin experienced mixed trading activity. The token’s price was $2.20 with a market capitalization of $5.6 billion, down 1.5% in 24 hours. However, trading volume increased by 3.4%, suggesting heightened investor interest in the project’s potential.

The TON ecosystem has grown significantly over the past year, with Telegram integrating blockchain features like payments, mini-apps, and NFTs into its platform. Cocoon represents the next step in this evolution, combining AI capabilities with blockchain infrastructure.

Max Crown, CEO of the TON Foundation, described Cocoon as marking “a shift toward an open, user-driven compute economy that returns control of AI infrastructure to users—rather than corporations that exploit it.”

Other decentralized AI networks already exist, including Akash Network and Render Network, which let users rent distributed computing resources. However, Cocoon’s integration with Telegram’s massive user base gives it a potential advantage in reaching mainstream adoption.

The Road Ahead

The November 2025 launch will test whether Cocoon can deliver on its promises. Success depends on attracting enough GPU providers to meet demand, ensuring the network can compete with centralized providers on speed and reliability, and convincing developers to switch from established cloud services.

Durov has a track record of prioritizing user privacy, having built Telegram as an encrypted messaging platform resistant to government surveillance. His comments during the announcement suggest he sees Cocoon as part of a broader mission to preserve digital freedoms in an increasingly centralized tech landscape.

Durov also revealed during a recent interview that he funds his lifestyle through Bitcoin investments, demonstrating his long-standing belief in decentralized systems and cryptocurrency. This philosophy now extends to AI infrastructure through Cocoon.

The project faces competition not just from other decentralized networks but also from tech giants with far more resources. However, the combination of Telegram’s user base, TON’s blockchain capabilities, and growing concerns about AI privacy could give Cocoon the momentum it needs to succeed.

A Bet on Digital Freedom

Cocoon represents more than just a new AI network. It’s a statement about who should control the technology that increasingly shapes our daily lives. By building infrastructure that keeps data private and gives users direct access to computing power, Telegram is challenging the current model where a few large companies dominate AI services. Whether this vision succeeds depends on execution, adoption, and whether users ultimately value privacy enough to choose decentralized alternatives over convenient centralized options.