運營商Token平台
封面:運營商Token平台
圖片由 AI輔助創作(Z-Image-Turbo)

兩個月近30億Token貸落地:運營商Token平台最值錢的模塊還沒人做出來|鄭哥點評$3B Token Loans in 2 Months — The Most Valuable Module of Telecom Token Platform Hasn't Been Built Yet

【鄭哥點評】【ZHENG GE REVIEW】

我做生意26年,見過無數次「電信運營商轉型」。

2000年代,佢哋靠語音通話食糊;2010年代,佢哋靠流量食糊;2020年代,佢哋靠寬帶食糊。

每一次都有人話:「運營商係管道,好快被顛覆。」

每一次,運營商都找到新嘅計費單位,繼續生存。

今次,佢哋揀咗Token。

第一幕:兩個月30億Token貸——風係點樣吹起嘅

場景1:三大運營商推出Token套餐的熱闘場面
場景1:三大運營商推出Token套餐的熱闘場面
圖片由 AI輔助創作(Z-Image-Turbo)

2026年5月16日至17日,三大運營商相繼推出Token套餐:

中國電信:9.9元/月/1000萬Tokens、29.9元/4000萬、49.9元/8000萬

中國移動:上海移動1元兌40萬Tokens,北京移動5.99元算力次包+24.99元月包

中國聯通:上海聯通免費發放3000萬Tokens測試額度,湖北聯通多檔Token Plan

兩個月內,多個省市級分公司已啟動市場化落地實踐。

運營商已經落場,唔再係概念階段。

第二幕:Token係乜嘢——用我哋做外貿嘅語言嚟解釋

場景2:Token作爲AI時代基本計量單位的圖解
場景2:Token作爲AI時代基本計量單位的圖解
圖片由 AI輔助創作(Z-Image-Turbo)

Token(詞元)係AI時代處理信息嘅基本計量單位。

語音時代:賣分鐘數——我哋打電話,電信公司按分鐘收費。

流量時代:賣GB——我哋上網,按流量收費。

AI時代:賣Token——我哋問AI問題,按Token消耗收費。

本質完全一樣——把稀缺嘅數字能力打包成標準化服務,賣俾用戶。

國家數據局2026年3月正式將Token定位為「智能時代的價值錨點」。

第三幕:運營商嘅護城河——Cloud廠商冇嘅嘢

場景3:運營商網絡優勢與Cloud廠商的差異
場景3:運營商網絡優勢與Cloud廠商的差異
圖片由 AI輔助創作(Z-Image-Turbo)

有人話:「運營商做Token,只係變相賣流量,冇乜特別。」

我話你知:呢個觀點睇漏咗三樣嘢。

第一:網絡層面嘅確定性服務。

當模型部署喺運營商自有數據中心,用戶通過運營商網絡訪問,可以喺網絡層面進行端到端優化——低時延、高穩定性。

呢個係任何純Cloud廠商或AI公司都唔具備嘅條件。

第二:18億個人用戶+7300萬政企客戶嘅銷售觸達網絡。

我哋做貿易嘅知道:最難嘅唔係產品,而係分銷渠道。

運營商唔使重新搭建渠道,佢哋嘅門市、App、客服,已經覆蓋晒中國每一個角落。

第三:數十年的套餐運營經驗。

電信套餐設計、分級定價、預付費模式——呢啲運營商已經玩咗幾十年,经验丰富。

第四幕:最大嘅問題——最值錢嘅模塊仲未做出嚟

場景4:DeepSeek等模型廠商價格優勢示意圖
場景4:DeepSeek等模型廠商價格優勢示意圖
圖片由 AI輔助創作(Z-Image-Turbo)

但係,我今次要潑冷水。

呢篇新聞標題話:「兩個月近30億Token貸落地」——但係問題在於:

呢30億Token,運營商只係賺咗個「辛苦錢」——卖緊係Token同質化產品,唔係差异化價值。

點解咁講?

因為而家運營商賣嘅Token,價格比直接去DeepSeek等模型廠商度高50倍:

  • 中國電信:9.9元/1000萬Tokens,即0.99元/百萬Tokens
  • DeepSeek V4 Flash(緩存命中):0.02元/百萬Tokens

價差50倍,用戶遲早醒水。

運營商嘅真正價值,唔喺於「賣Token」,而喺於:

能不能做出「網絡+算力+應用」一體化嘅封閉生態——令用戶為呢個生態付費,而唔係為Token本身付費。

呢個先至係最值錢嘅模塊,而家仲冇人做到。

第五幕:電信轉型嘅啟示——從「管道」到「生態」

場景5:從「管道」到「生態」的轉型之路
場景5:從「管道」到「生態」的轉型之路
圖片由 AI輔助創作(Z-Image-Turbo)

中國電信董事長柯瑞文嘅說法好清楚:

「Token經營的本質,就係為用戶提供AI服務,搭建算力建網、平台調度、數據積累、模型研發、應用落地五層一體化全棧閉環能力。」

呢個方向係啱嘅,但係執行層面挑戰極大。

問題唔喺技術,而喺:

  • 運營商嘅組織架構,係為咗電信網絡設計,唔係為咗AI應用生態設計
  • 運營商嘅人才結構,係電信工程師為主,唔係AI應用開發者為主
  • 運營商嘅文化,係電信公司文化,唔係互聯網公司文化

要轉型做AI生態,運營商需要嘅唔只係技術,而係整個組織基因嘅重構。

鄭哥三句話

第一句:運營商賣Token,方向正確,但執行係大考驗。 佢哋有網絡優勢、用戶優勢、渠道優勢——但係做AI生態,需要嘅係互聯網思維,而唔係電信思維。從「管道」到「生態」,呢個跳躍唔係加幾條產品線就做到,而係需要整個組織基因重構。

第二句:單賣Token係辛苦錢,生態先至係核心價值。 當用戶可以直接去DeepSeek、Kimi等模型廠商購買更平嘅Token,運營商嘅競爭力必須來自差异化——網絡確定性服務、行業定制解決方案、封閉生態體驗。冇呢啲,運營商就只係貴價中間商。

第三句:我哋做貿易嘅,可以點樣利用呢個趨勢? 運營商嘅渠道能力,係我哋出海嘅工具。如果能夠搭上運營商嘅AI+出海生態,可能比我哋自己搭建渠道更有效率。但係要小心——選擇平台,先睇佢嘅生態封閉程度,越封閉越值錢,越開放越快被淘汰。

兩個月近30億Token貸落地,說明運營商已經正式進入AI變現時代。但係呢30億只係「辛苦錢」——單賣Token係初級階段,「網絡+算力+應用」一體化生態先至係最值錢嘅模塊,而家仲冇人做出嚟。運營商有優勢,但基因重構係最大挑戰。
鄭哥 Hany

關於作者About the Author

鄭哥,90年代大學畢業,26年橫跨電子,家電、服裝、珠寶、教育、AI算力多個行業。做過中學教師、翻譯、外贸,現在搞七善門科技和星辰算力。口頭禪:「點解唔得?」Hany, university grad from the 90s, 26 years across electronics, appliances, fashion, jewelry, education, and AI compute. Former teacher, translator, trader. Now running Qishanmen Tech and Star Compute. Catchphrase: "Why not?"

Cover: 運營商Token平台
封面:運營商Token平台
Image by AI (Z-Image-Turbo)

After 26 years in business, I've seen countless "telecom operator transformations."

In the 2000s, they made money from voice calls; in the 2010s, from data; in the 2020s, from broadband.

Every time, people said: "Operators are just pipes, they'll be disrupted soon."

Every time, operators found a new billing unit and survived.

This time, they chose Token.

Act 1: $3B Token Loans in 2 Months — How the Wind Started Blowing

Scene 1: Three major operators launching Token packages
Scene 1: Three major operators launching Token packages
Image by AI (Z-Image-Turbo)

May 16-17, 2026, the three major operators launched Token packages:

China Telecom: 9.9 yuan/month per 10M Tokens, 29.9 yuan/40M, 49.9 yuan/80M

China Mobile: Shanghai 1 yuan for 400K Tokens, Beijing 5.99 yuan compute add-on + 24.99 yuan monthly

China Unicom: Shanghai giving away 30M Tokens free for testing, Hubei with multiple Token plans

Operators have entered the game, no longer just concept stage.

Act 2: What is Token — Explained in Our Trade Language

Scene 2: Token as the basic unit of AI era measurement
Scene 2: Token as the basic unit of AI era measurement
Image by AI (Z-Image-Turbo)

Token is the basic unit of measurement for AI-era information processing.

Essentially the same — packaging scarce digital capability into standardized service and selling to users.

The National Data Bureau officially designated Token as the "value anchor of the intelligent era" in March 2026.

Act 3: Operators' Moat — What Cloud Vendors Don't Have

Scene 3: Operators network advantages vs cloud vendors
Scene 3: Operators network advantages vs cloud vendors
Image by AI (Z-Image-Turbo)

First: Network-level deterministic service. When models deploy in operators' own data centers, end-to-end optimization at the network level is possible.

Second: 1.8B individual users + 73M enterprise clients. Their stores, apps, and customer service already cover every corner of China.

Third: Decades of package operation experience.

Act 4: The Biggest Problem — The Most Valuable Module Hasn't Been Built

Scene 4: DeepSeek and model vendors price advantage diagram
Scene 4: DeepSeek and model vendors price advantage diagram
Image by AI (Z-Image-Turbo)

These $3B Tokens, operators only earned "hard work money" — selling homogeneous Token products, not differentiated value.

Current operator Token prices are 50x higher than buying directly from DeepSeek:

  • China Telecom: 0.99 yuan/million Tokens
  • DeepSeek V4 Flash (cache hit): 0.02 yuan/million Tokens

Operators' real value is in building an integrated "network + compute + application" closed ecosystem.

Act 5: Telecom Transformation Lessons — From "Pipes" to "Ecosystem"

Scene 5: Transformation from pipes to ecosystem
Scene 5: Transformation from pipes to ecosystem
Image by AI (Z-Image-Turbo)

China Telecom Chairman Ke Ruiwen: "Token operation's essence is providing AI services to users, building five-layer integrated full-stack closed-loop capability."

The problem isn't technology — it's organizational gene restructuring.

Zheng Ge's Three Points

Point 1: Operators selling Token, direction correct, but execution is a major test.

Point 2: Selling Token alone is hard work money, ecosystem is core value.

Point 3: How can traders leverage this trend? Operators' channel capability is our tool for going overseas.

$3B Token loans in 2 months shows operators have officially entered the AI monetization era. But these $3B are just "hard work money" — the most valuable module, nobody built it yet.
鄭哥 Hany

關於作者About the Author

鄭哥,90年代大學畢業,26年橫跨電子,家電、服裝、珠寶、教育、AI算力多個行業。做過中學教師、翻譯、外贸,現在搞七善門科技和星辰算力。口頭禪:「點解唔得?」Hany, university grad from the 90s, 26 years across electronics, appliances, fashion, jewelry, education, and AI compute. Former teacher, translator, trader. Now running Qishanmen Tech and Star Compute.