The world’s most-capitalised companies in artificial intelligence are embedding their products directly into Telegram’s billion-user ecosystem — deploying agents, building bots, and integrating their model APIs into the platform’s messaging layer. Not one of them appears to have secured their corresponding Fragment @name on the TON blockchain. The contradiction is not subtle.

A Billion Users and a Distribution Channel IP Teams Are Ignoring

Telegram crossed 1 billion monthly active users in March 2025. By mid-2026, the platform reports 500 million daily active users and processes over 15 billion messages per day. Telegram’s revenue reached $1.4 billion in 2024 — its first profitable year — positioning the app as a genuine operating platform for commercial AI services rather than a consumer messaging novelty.

Within this ecosystem, TON Mini Apps now reach more than 500 million monthly users inside Telegram. A 2026 survey of Telegram Mini App developers found that 97% use AI-powered tools in production. Analysts now describe the combined AI and mini-app integration trajectory as making Telegram “the largest open conversational AI platform” in operation. This is not a prediction. It is the current deployment landscape.

The Deployment Is Real, Accelerating, and Led by the Biggest Names

In December 2025, Meta acquired Manus — a Singapore-based AI agent firm with $100 million in Annual Recurring Revenue — for more than $2 billion. The strategic rationale was explicit: agentic AI integration across WhatsApp, Messenger, Facebook, and Instagram. On February 16, 2026, the Manus integration launched — not on WhatsApp, not on Facebook Messenger, but on Telegram, making it the first messaging platform to receive agentic AI deployment under Meta’s ownership.

Anthropic followed a parallel path. In early 2026 the company shipped Claude Code Channels — an integration that enables Claude Code sessions to interface directly through Telegram via Anthropic’s Model Context Protocol. This is not a community workaround. It is official Anthropic infrastructure routing developer interactions through Telegram. Third-party bots built on OpenAI, Anthropic, Google Gemini, and Groq APIs are delivering tens of millions of daily model interactions to Telegram’s user base. The platform is the distribution layer. The AI labs know it.

@openai Is Not OpenAI

Against this backdrop, fragment.com/username/openai is an active live listing. The username @openai exists on the TON blockchain as a tradable NFT. OpenAI — which reached a $300 billion valuation in 2025 and is the company most synonymous with conversational AI globally — does not appear to hold it. The listing is held by a market participant who is not OpenAI.

The same pattern extends across the sector. @anthropic, @mistral, @gemini, and @perplexity are each brand identifiers for companies collectively valued in the hundreds of billions of dollars. Each company runs models that power Telegram interactions at scale. The Fragment @name status of each appears, to the knowledge of this publication, to be outside the control of the companies whose brands they carry.

The risk this creates is not theoretical. On a platform with 1 billion registered users and 15 billion daily messages, a Fragment @name grants its holder the Telegram username @[brand]. An entity controlling @openai on Telegram can operate a channel, group, or bot presenting as OpenAI to that entire user base — including users who have already interacted with official OpenAI-powered bots through third-party aggregators. The company whose brand it is has no on-chain reclaim mechanism without negotiating with, and compensating, whoever holds the NFT.

The Price of Inaction Is Compounding

Fragment @name prices are establishing a clear precedent for brand-value correlation. @boss sold for $500,000 in April 2026. @danbao set a market record at $2.2 million in February 2026. @news sold for $1.78 million. The pricing logic is consistent: names with high recognition, commercial weight, and broad applicability command six- to seven-figure premiums in the secondary market.

AI lab brand names sit squarely in that category. @openai, @anthropic, and @perplexity rank among the most-searched brand terms of the past three years across every major market. Their Fragment @name acquisition cost — had it been undertaken in 2022 or 2023 — would have been a rounding error on a Series B term sheet. The market trajectory since then is instructive: @boss was acquired in November 2022 for approximately $64,000 worth of Toncoin. By April 2026 it cleared $500,000. That is not a unique data point. It is the directional signal for every brand-correlated @name still outside corporate control.

The Structural Blind Spot

The structural reason IP teams are missing Fragment is familiar: Fragment @names are NFTs on the TON blockchain, not domain names subject to ICANN policy, not trademark registrations subject to WIPO procedures, and not social media handles subject to platform-level impersonation takedowns. There is no UDRP-equivalent. There is no registry-level dispute mechanism a trademark holder can invoke. Acquisition on Fragment requires either being first to bid or paying the market price to whoever holds the NFT.

IP departments that monitor ICANN gTLD registrations, track WIPO filings across 192 jurisdictions, and run brand-monitoring sweeps across social platforms can still miss Fragment entirely. The asset class sits outside every existing workflow in the standard IP protection stack.

  • No UDRP equivalent — trademark-based disputes cannot force a transfer
  • No takedown mechanism — Telegram’s platform policies do not extend to on-chain NFT ownership
  • No registry recourse — Fragment operates as a decentralised auction with no central arbitration body
  • Rising floor prices — secondary market premiums are compounding as brand-aware buyers accumulate inventory

The Strategic Implication

The industry that makes its living on language, naming, and conversational brand recognition is ceding its own namespace on the world’s fastest-growing conversational AI platform. Meta staked $2 billion on Telegram’s future as an AI distribution channel in December 2025. Anthropic built official developer tooling for Telegram in early 2026. The platform commitment is categorical. The namespace strategy is absent.

IP teams advising AI companies have a narrowing window. As brand-correlated Fragment @names migrate from casual holders to informed market participants with an understanding of AI brand value, acquisition cost and negotiating leverage both shift unfavourably. The question for IP counsel is not whether the Fragment market is significant enough to warrant attention. The question is whether the $64,000-to-$500,000 trajectory of @boss is a cautionary tale they want to read again, this time with their client’s brand in the headline.