The founder of Telegram announced the launch of “AI computing network Cocoon”: GPU can be used to mine TON, 100% private computing
The founder of Telegram announced the launch of the Cocoon decentralized computing power network: it focuses on allowing edge computing power to solve needs and 100% privacy, allowing GPU holders to rent computing power and earn TON, challenging the monopoly of cloud giants.
(Preliminary summary: Telegram founder Pavel Durov warned: The free Internet is in danger, and the surveillance era eats up digital freedom)
(Background supplement: News flash》Telegram founder Pavel Durov was allowed to leave France, TON jumped 20%)
Contents of this article
Telegram founder Pavel Durov announced his ambitious project yesterday (1st): Cocoon (Confidential Compute Open Network) officially announced that the mainnet is online.
This marks the Telegram ecosystem’s official entry into the field of AI infrastructure and a direct challenge to centralized cloud giants such as Amazon AWS and Microsoft Azure.
Durov proudly stated in the announcement:
It happened. Cocoon, our decentralized confidential computing network, is now online. The first AI requests from users are being processed by Cocoon with 100% confidentiality, and GPU holders have already started earning TON.
What is Cocoon? Not just “another” decentralized network
Cocoon is a decentralized physical infrastructure network built on top of the TON Network. Its core vision is to solve the two major pain points in the current development of AI: Expensive computing costs and Loss of data privacy.
Unlike traditional cloud services that transmit data to centralized servers for processing, Cocoon matches the needs of both ends:
Computing power demand side (AI developers): Cheap and safe computing power is needed to run AI models.
Computing power provider (GPU holder): Owns an idle high-performance graphics card (GPU) and hopes to make a profit by renting out computing power.
Core technology: How to build "trust" in "distrust"?
According to the official description, Cocoon has the following two major features:
100% privacy protection: Through trusted execution environments (TEEs, such as Intel TDX and other technologies), Cocoon ensures that the AI computing process is performed in an encrypted "black box". This means that even GPU miners providing computing power cannot view or intercept the data or AI model content being processed on their hardware.
Decentralized architecture: No single intermediary controls the flow of data, fundamentally eliminating the possibility of abusing user data to train models or conduct censorship like technology giants do.
Fighting technology giants: Breaking the monopoly of Amazon and Microsoft
Pavel Durov bluntly raised the challenge in the announcement: "Centralized computing providers such as Amazon and Microsoft act as expensive intermediaries, driving up prices and reducing privacy."
Cocoon attempts to break this monopoly through market mechanisms. Through the decentralized bidding model, developers are expected to obtain lower inference costs than traditional clouds, and using TON as the settlement layer ensures that global GPU providers can obtain immediate and transparent benefits.
Ecosystem integration: Telegram is the biggest support
Cocoon was born with a rich dad, and behind it is the support of Telegram’s more than 1 billion users.
Telegram Mini Apps Empowerment: Future Telegram robots and mini-programs are expected to directly use Cocoon's computing power to implement "privacy-first" AI functions (such as private chat bots, document analysis, etc.).
GPU Mining: For users with consumer-grade high-end graphics cards (such as NVIDIA RTX 4090) or enterprise-level graphics cards (such as H100, A100), Cocoon provides a new mining method to earn TON tokens by contributing computing power.
As Durov said: "Cocoon wants to bring control and privacy back to where they belong: in the hands of users." But to be honest, how to solve the AI computing power needs through edge computing power, there are still many pain points that need to be solved in terms of implementation. Whether it can realize its ambitious vision remains to be seen.
