I. TL;DR — September Dense New Policy Period Triple-Track Leap

Between September 8-12, 2026 (5 days), China’s data industry shifted from the “14th Five-Year Plan” wrap-up to the “15th Five-Year Plan” opening with three coordinated tracks: ① Scale Maturity Track — The 2026 China International Big Data Industry Expo (Aug 28-30, Guiyang) released the “China Data Industry Development Report (2026)”: data industry scale 6.78 trillion CNY in 2025 (2023: 5.06T → 2024: 5.86T → 2025: 6.78T, YoY +15.7%), data companies 482K (+17.9%); ② Regional Coordination Landing Track — Sep 1 Jiangsu “AI+ Industrial Software Breakthrough Development Action Plan (2026-2028)” 5 actions landing; Sep 4 Shenzhen “Promoting AI and Application Development Action Plan (2026-2028)” 5 tasks landing; ③ Data Opening Deepening Track — Aug 28 National Data Bureau announced 14 central SOEs opening 6B+ high-value data records (covering power / oil pipeline / logistics / auto / agricultural machinery 10+ industries) + Aug 28 second batch 32-city data annotation pilot (cumulative 39-city “7+32” nationwide network) + Sep 8 Liu Liehong signed article “Grasping the Digital-Intelligent Development Trend, Deeply Advancing Digital China Construction” authoritative tone-setting.

Core Conclusion: China’s data industry has completed three-stage transition: “Infrastructure Construction Period (2020-2022) → Element Allocation Reform Period (2023-2025) → Scale Maturity Period (2026-)”; Moving from “Central Tone-Setting” to “Local Landing + Data Opening + Industrial Scale” three-dimensional coordination; Data element × AI bidirectional empowerment enters “high-quality dataset + agent patent + token service + annotation industry” four-piece stacking period.

II. Scale Maturity Track: 6.78T CNY Industry + 482K Companies + 17.9% Growth

2.1 Data Industry Scale Three-Year Three-Step Leap

National Data Development Research Institute Director Hu Jianbo released the “China Data Industry Development Report (2026)” at the Aug 28 Guiyang 2026 Big Data Expo opening:

YearData Industry Scale (T CNY)YoY GrowthData Companies
20235.06--
20245.86+15.8%-
20256.78+15.7%482K (+17.9%)

Key Data:

  • Data industry scale 6.78T CNY in 2025, +33.9% growth vs 2023
  • Data companies 482K, eastern data industry scale reaches 5.23T CNY (77.1% of national)
  • AI-related data products & services exceed 70% of total data industry output (Big Data Expo official data)
  • Agent patent authorizations 3,400+, growth rate 2x previous year (2025 vs 2024)

2.2 High-Quality Datasets 126K + 1,815 PB

National Data Bureau Sep 8 Liu Liehong signed article disclosed: as of August 2026, the nation has cumulatively built over 126K high-quality datasets, total data volume exceeding 1,815 PB (petabytes).

Comparison Timeline:

  • 2026-03: Pre-expo data approximately 960 PB (implicit)
  • 2026-08: 126K / 1,815 PB, half-year growth +89%
  • 2026-08 Big Data Expo: National data infrastructure connected 40K+ entities / listed 60K+ products (+10x / +4x vs end of 2025)

2.3 Token Service: From “Concept” to “Industry”

Sep 11 People’s Daily Overseas Edition “Token Surging, Intelligence Starting New Journey” (Sep 11, 2026, Page 05) systematically reviews token economy 14-day leap from “concept proposed at Guiyang Big Data Expo (Aug 28)” to “industrial landing”:

  • China Unicom Yuanjing MaaS one-stop token operation service platform has aggregated 230+ mainstream AI large models, covering three major areas: large language, voice, vision
  • Platform equipped with “select / modify / use” one-stop toolchain, Vietnamese small-language simultaneous interpretation instruction-following accuracy improved 30%, real-time translation accuracy exceeding 90%
  • Huawei Ascend 950 super-node computing cabinet debuted, 1024-card high-speed interconnection, supporting trillion-parameter large model training tasks
  • National Data Development Research Institute “China Data Industry Development Report (2026)” first inclusion of “token service” as data industry new business form special item

III. Regional Coordination Landing Track: Jiangsu + Shenzhen + Yangtze River Delta + Greater Bay Area September Four-Hit

3.1 Jiangsu “AI+ Industrial Software Breakthrough Development Action Plan” 5 Actions (Sep 1)

Jiangsu Provincial Department of Industry and Information Technology Sep 1 issued the “Jiangsu Province ‘AI+ Industrial Software’ Breakthrough Development Action Plan (2026-2028)”, 5 actions:

  1. Industrial Software Development Capability Foundation: Focus on “AI+” industrial software underlying core technology research, promote industrial software and AI capability integration
  2. Industrial Software Enterprise Intelligentization Enhancement: Develop “Model-as-a-Service MaaS / Agent-as-a-Service AaaS / Result-as-a-Service RaaS” three new business forms, cultivate pay-by-result, pay-by-usage new models
  3. Industrial Software Empowering Manufacturing: Promote high-quality industrial dataset construction, promote data and model collaborative optimization
  4. Industrial Software Innovation Carrier Cultivation: Accelerate industrial software research institute construction, promote region-specific construction of innovation verification, compatibility adaptation, application promotion platforms
  5. Industrial Software Ecosystem Optimization: Encourage issuing “computing power vouchers / model vouchers / token vouchers” to lower enterprise AI application thresholds

Key Significance: Jiangsu is the first provincial-level “AI + Industrial Software” special, having benchmark significance for the national 1.5T CNY industrial software scale (2025), “Three-Voucher Linkage” (computing power voucher + model voucher + token voucher) becoming the first provincial-level AI application subsidy combination.

3.2 Shenzhen “Promoting AI and Application Development Action Plan” 5 Tasks (Sep 4)

Shenzhen Municipal Bureau of Industry and Information Technology, Development and Reform Commission, Science and Technology Innovation Bureau Sep 4 jointly issued the “Shenzhen Promoting AI and Application Development Action Plan (2026-2028)”, 5 tasks:

  1. Strengthen Computing Power Foundation: Deepen computing-electricity coordinated reform, promote computing facility green power direct connection, explore offshore wind grid-connected + wind-solar-storage grid-connected computing-electricity coordination
  2. Support Model Algorithm Innovation: Around intelligent emergence mechanisms, recursive self-improvement, new computing paradigms, brain-inspired computing theory, support AGI development paths
  3. Increase High-Quality Data Supply: Build embodied intelligence data resource and circulation platform, form high-quality training datasets and evaluation benchmark supply capability for typical scenarios
  4. Promote Platform Empowerment and Scenario Opening: Rely on Digital Shenzhen Joint Innovation Center + Guangdong-Hong Kong-Macao Greater Bay Area Application Scenario Innovation Center, “one-place innovation, multi-place reuse” model radiating across province
  5. Strengthen Talent Cultivation: Implement AI talent special policy, vigorously attract talents to develop in Shenzhen

3.3 Sep 8 Liu Liehong Signed Article Authoritative Tone-Setting

National Data Bureau Director Liu Liehong Sep 8 published signed article “Grasping the Digital-Intelligent Development Trend, Deeply Advancing Digital China Construction” in People’s Daily Overseas Edition, systematically elaborating “15th Five-Year Plan” Digital China construction 8 priorities for the first time:

Foundation Consolidation (3 aspects):

  • Digital Infrastructure Layout Construction and Intensive Efficient Utilization: National integrated computing power network + 5G / Gigabit optical network + Satellite internet + Data annotation / circulation / delivery / application bases
  • Data Element Market Allocation Reform: National unified data property registration system + National integrated data market + “Data Element ×” action + Data industry cluster
  • Digital Field Key Technology Research and Talent Cultivation: Integrated advancement of key core technology + Data element related discipline specialty

International Cooperation (3 principles): Two-way integration / Universal inclusion / Joint consultation and construction

3.4 Changchun / Chengdu / Hefei and other 39-City Data Annotation “7+32” Network (Aug 28-Sep 12)

Aug 28 Guiyang Big Data Expo, National Data Bureau Deputy Director Xia Bing announced: based on 7 comprehensive national data annotation bases (Chengdu, Hefei, etc.), add 32 second batch data annotation pilot cities, cumulative 39-city “7+32” nationwide network. National Data Set Management Service System as of Aug 25 registered authenticated entities 725, released datasets 1,721.

Typical Case (Changchun Sep 1 inclusion announcement):

  • Changchun: Existing data annotation companies 406, “Changchun City Cultivating Data Annotation Industry High-Quality Development Action Plan (2025-2027)”, Changguang Satellite Remote Sensing Trusted Data Space selected as national first batch enterprise trusted data space innovation development pilot
  • Second batch key fields: Industrial manufacturing / Medical health / Transportation / Embodied intelligence 4 major industries

IV. Data Opening Deepening Track: 14 Central SOEs 6B+ Records + Token Business Model + Agent Patents

4.1 14 Central SOEs Open 6B+ High-Value Data Records (Aug 28-Sep 1)

During Aug 28-Sep 1 Big Data Expo period, 14 central SOEs initiated the first domestic “Data Opening Month” activity, opening high-value data resources to society free of charge:

  • Expected to open data total volume exceeding 6 billion records, covering power / oil & gas pipeline / logistics / auto / agricultural machinery 10+ industries
  • Aggregating over 500 high-quality data resources
  • China Southern Power Grid first launch of data “trial use” mechanism, opening 100+ items of grid operation / new energy consumption / equipment operation data, allowing power generation enterprises, research institutes zero-threshold trial and data value verification, “use well then talk cooperation”

Key Significance: This is the first domestic central SOE group centralized data resource opening, “use first then talk” (data trial mechanism) is a landmark breakthrough in China’s data element market allocation reform — paradigm shift from “data resource holder” to “data element circulator”.

4.2 Token Business Model Exploration (Aug 29-Sep 12)

National Data Bureau Director Liu Liehong Aug 29 Big Data Expo period announced: will explore token business models adapting to industrial reality —

Research and explore “token value-added subscription + pay-by-effectiveness” two business models (National Data Bureau Aug 29 Big Data Expo disclosure).

Supporting Landing:

  • National Data Bureau joint multi-department implementing “Data Field High-Skilled Talent Cluster Cultivation Plan”, improve data element related discipline specialty bachelor-master-PhD integrated talent cultivation mechanism
  • China Unicom Yuanjing MaaS platform 230+ mainstream models + Vietnamese small-language simultaneous interpretation 30% improvement + real-time translation 90% accuracy (Big Data Expo Aug 28-30 debut)

4.3 Agent Patents + Token Industry + Data Annotation + High-Quality Dataset “Four-Piece Set”

As of August 2026, data industry forms “four-piece set” stacking:

DimensionScaleYoY GrowthSource
Data companies482K+17.9%China Data Industry Development Report 2026
High-quality datasets126K / 1,815 PB-National Data Bureau 2026/8
Agent patent authorizations3,400++200%China Data Industry Development Report 2026
Data annotation pilot cities39 cities (7+32)+457%National Data Bureau 2026/8

Agent patents 3,400+ with growth 2x first disclosed, marking agent innovation entering “patent explosion period”, paradigm leap from “model training data” to “agent interaction data”.

V. September Dense New Policy Period vs Historical P8 Digitalization Angles

DateAngleMain Events
6/101.73T + 88% AI adoption + 4-stage pathEarly digital economy overview
7/17APEC + Data Bureau + WAIC Agent PC + Central SOE entryInternational + policy resonance
8/21Data property registration + 10-place international cooperation + ISO/IEC 25005Institutional landing
8/29Guiyang Big Data Expo + Token economy + 126K / 1,815 PB + 372 companiesGuiyang home court
9/5Dual green-digital + National standard + Shijiazhuang Big Data Expo + AI SME + 15FYP 109 projectsLayered landing
9/126.78T + 482K + Jiangsu / Shenzhen + 14 Central SOEs 6B+ + 9/8 Liu Liehong tone-setting + 9/11 People’s Daily Token SurgingTriple-track parallel

This Angle Innovation: Moving from “top-level tone-setting / single-point breakthrough” (6/10 / 7/17 / 8/21 / 8/29 / 9/5) to “scale maturity + regional coordination landing + data opening deepening” three-track parallel, first focusing on data industry hard data (6.78T / 482K / +17.9%), regional coordination (Jiangsu 5 actions / Shenzhen 5 tasks), central SOE data opening (14 / 6B+).

VI. 5-Step Enterprise Landing Path

Step 1 Data Asset Assessment: Inventory enterprise internal data resources, identify openable / tradable / annotatable / trainable 4-type assets.

Step 2 High-Quality Dataset Construction: Rely on national data annotation “7+32” pilot network, proactively connect with Chengdu, Hefei, Changchun etc. 39-city annotation bases, focus on industry breakthroughs (industrial manufacturing / medical health / transportation / embodied intelligence).

Step 3 Token Business Model Pilot: Reference “China Unicom Yuanjing MaaS 230+ models” template, explore “pay-by-effectiveness + value-added subscription” dual track, move from “selling data” to “selling tokens / selling model invocation”.

Step 4 Central SOE Data Opening Connection: Track 14 central SOEs “Data Opening Month” dynamics, prioritize connecting with China Southern Power Grid power / PetroChina oil pipeline / FAW auto / COSCO Shipping logistics 4 major fields.

Step 5 Agent + Token Service: Seize agent patent +200% explosion window, layout agent patent application + token service + annotation industry + high-quality dataset “four-piece set” coordination.

VII. 6 Defense Checklists

  1. Data Compliance: Based on 2026 Big Data Expo National Data Bureau new rules, all data opening requires “trusted data space” compliance review
  2. Annotation Quality: Reference “human-machine collaboration + expert annotation” new model, avoid low-end annotation polluting AI training
  3. Central SOE Data Trial Use: Under China Southern Power Grid “use first then talk” mechanism, enterprise connection needs clear “data trial + value verification + secondary cooperation” three-stage process
  4. Token Business Model Compliance: Value-added subscription / pay-by-effectiveness requires supporting data element registration + revenue distribution mechanism, avoid “gray pricing”
  5. Regional Coordination Compliance: Jiangsu / Shenzhen and other local plans mostly include “computing power voucher / model voucher / token voucher” subsidies, enterprises need to pay attention to local government application time windows (Jiangsu plan valid 2026-2028)
  6. Agent Patent Layout: Agent patents 3,400+ with growth 2x, enterprises should pre-layout core interaction process patents, avoid being locked by “four-piece set” patent barriers

VIII. Key Terminology

  • Data Industry (数据产业): Industry collection covering data collection, storage, processing, circulation, application entire chain, 2025 scale 6.78T CNY, 482K companies.
  • Data Element (数据要素): Data as new production factor, 2020 Central Reform Commission first clarified “data as production factor” status.
  • High-Quality Dataset (高质量数据集): High-value data collection meeting AI training requirements, processed through collection, cleaning, annotation, compliance. As of August 2026, nation built 126K / 1,815 PB.
  • Data Annotation (数据标注): “Labeling” raw data, fundamental process for AI training. National data annotation “7+32” pilot cumulatively 39 cities.
  • Token (词元): Callable intelligent service unit, both data element value measurement unit and new business form for AI model service. August 2026 National Data Bureau first clarified as “data element value release new path”.
  • Trusted Data Space (可信数据空间): Safe and controllable data circulation environment based on data infrastructure, 2026 Big Data Expo data market relies on national data infrastructure trusted foundation.
  • Data Opening Month (数据开放月): First domestic central SOE group centralized data resource opening activity, 14 central SOEs 6B+ data records free trial.
  • Computing-Electricity Coordination (算电协同): Computing facility and energy system coordinated layout, Shenzhen Sep 4 plan clearly promotes computing facility green power direct connection.

IX. FAQ (High-Frequency Direct Answers)

Q1: What is the biggest benefit for enterprises in the September dense new policy period? A: Central SOE “Data Opening Month” 6B+ data records + Jiangsu / Shenzhen “computing power voucher / model voucher / token voucher” three-voucher linkage. The former lowers data acquisition threshold (China Southern Power Grid 100+ data items zero-threshold trial), the latter lowers AI application cost.

Q2: How credible is the 6.78T CNY data industry scale? A: Released by National Data Development Research Institute at 2026 Guiyang Big Data Expo opening, included in “China Data Industry Development Report (2026)” — Aug 28 Big Data Expo National Data Bureau official website disclosure, with Sep 1 CCTV News, People’s Daily Sep 11 Overseas Edition, Sep 1 National Data Bureau website simultaneous disclosure, 4-source cross-verification credible.

Q3: How can enterprises participate in the “7+32” data annotation pilot? A: Currently “7+32” is city-level layout (Chengdu, Hefei, Changchun etc. 39 cities), enterprises can: ① Join local data annotation industry alliance (Changchun already has 406 alliance companies); ② Apply for “trusted data space” pilot (Shenzhen “Changguang Satellite Remote Sensing Trusted Data Space” is first batch); ③ Apply for local government “computing power voucher / model voucher / token voucher” subsidies.

Q4: What level is the agent patent 3,400+ growth 2x? A: 2025 vs 2024, China’s agent patent authorizations jumped from 1,100+ to 3,400+, growth 200%+ — far exceeding same-period overall Chinese invention patent growth (+15% or so), first time AI field has “agent” as patent classification high-speed growth period.

Q5: Is Jiangsu AI Industrial Software plan applicable to other province enterprises? A: Not directly applicable, but can reference “5 actions + three-voucher linkage” template. Shenzhen Sep 4 plan clearly established “one-place innovation, multi-place reuse” model, other province enterprises can connect with Shenzhen Digital Shenzhen Joint Innovation Center + Guangdong-Hong Kong-Macao Greater Bay Area Application Scenario Innovation Center, participate in cross-region pilots.

Q6: What is the relationship between data element and digital economy? A: Data element is the “core production material” of digital economy, digital economy is the “macro industrial form” of data element. 2026 data industry 6.78T (digital economy core industry subset) + digital economy overall scale 60T+ (including digital infrastructure, digital industrialization, industrial digitization, digital governance) — Data element is the most active, fastest-growing subset in digital economy.

X. References (Four-Level Classification)

National Policy and Official Documents (Level 1)

  1. National Data Bureau “2026 Digital Economy Development Work Key Points” (8 priority tasks) — https://zjt.nmg.gov.cn/zwdt/qgyw/202605/t20260525_2906909.html
  2. National Data Bureau “2026 Digital Society Development Work Key Points” (23 tasks) — https://www.deo.org.cn/article/hyfz_hydt/detail3125.html
  3. Jiangsu Provincial Department of Industry and Information Technology “AI+ Industrial Software Breakthrough Development Action Plan (2026-2028)” (Sep 1) — https://www.cei.cn/defaultsite/s/article/2026/09/01/4b4ff607-9d914745-01a0-5badfa4b-4bdf_2026.html
  4. Shenzhen Municipal Bureau of Industry and Information Technology etc. three departments “Promoting AI and Application Development Action Plan (2026-2028)” (Sep 4) — https://gxj.sz.gov.cn/gkmlpt/content/12/12967/post_12967724.html
  5. National Data Bureau Deputy Director Xia Bing Aug 28 Big Data Expo speech: 32-city 2nd batch data annotation pilot — https://www.nda.gov.cn/sjj/swdt/mtsy/0829/20260829095640662031992_pc.html
  6. National Data Bureau Director Liu Liehong Sep 8 signed article “Grasping the Digital-Intelligent Development Trend” — https://xasjj.xa.gov.cn/xwzx/xwdt/2097629965500092418.html

Official Data and Industry Reports (Level 2)

  1. “China Data Industry Development Report (2026)” National Data Development Research Institute Aug 28 Big Data Expo release — https://paper.bzrb.net/bzrb/20260901/html/content_20260901007003.htm
  2. National Data Bureau Aug 30 Big Data Expo disclosure: 14 central SOEs 6B+ data + 6T industry scale — https://www.nda.gov.cn/sjj/swdt/mtsy/0830/20260830195001667056747_pc.html
  3. China Securities Journal: Data annotation “7+32” network + National Data Set Management Service System 725 entities — https://jnz.cs.com.cn/app/html/jnz/131858.html

Authoritative Media (Level 3)

  1. CCTV News Sep 1 “China Launches New Measures to Accelerate Data Industry Development” — https://www.deo.org.cn/article/hyfz_hydt/detail3594.html
  2. People’s Daily Overseas Edition Sep 11 Page 05 “Token Surging, Intelligence Starting New Journey” — http://m2.people.cn/news/default.html?s=Ml8yXzQxNjkyOTczXzE5NDgzMV8xNzg5MDg2MzU1
  3. Xinhua News Agency Guiyang Aug 30: Data industry 6.78T in 2025 (Gao Kang / Xiang Dingjie) — https://paper.bzrb.net/bzrb/20260901/html/content_20260901007003.htm
  4. Workers’ Daily Aug 30: Token business model + 32-city annotation pilot — https://www.workercn.cn/c/2026-08-30/8881420.shtml
  5. China Economic Times Sep 4: Data market new infrastructure — https://jjsb.cet.com.cn/show_544670.html
  6. China Economic Information Network Sep 1 Jiangsu AI Industrial Software plan — https://www.cei.cn/defaultsite/s/article/2026/09/01/4b4ff607-9d914745-01a0-5badfa4b-4bdf_2026.html
  7. National Data Bureau Aug 30: China Launches New Measures to Accelerate Data Industry Development — https://www.nda.gov.cn/sjj/swdt/mtsy/0830/20260830195001667056747_pc.html

Local Landing and Industry Dynamics (Level 4)

  1. Changchun Municipal Science and Technology Bureau Sep 1: Changchun selected as second batch data annotation pilot city — http://kjj.changchun.gov.cn/sy/gzdt/202609/t20260901_3509154.html
  2. Inner Mongolia Autonomous Region Government Services and Data Management Bureau Sep 1: 14 central SOEs 6B+ data — https://zsj.nmg.gov.cn/xwzx/sznmg/202609/t20260901_2948154.html
  3. Henan Province Sep 1: China Launches New Measures to Accelerate Data Industry Development — https://sjj.henan.gov.cn/2026/08-31/3408947.html
  4. Shenzhen Municipal Government Sep 4: Promoting AI and Application Development Action Plan — https://gxj.sz.gov.cn/gkmlpt/content/12/12967/post_12967724.html
  5. China Data Element x Project Office Sep 1 reprint — https://www.deo.org.cn/article/hyfz_hydt/detail3594.html
  6. Xi’an Data Bureau Sep 9 reprint Liu Liehong signed article — https://xasjj.xa.gov.cn/xwzx/xwdt/2097629965500092418.html

International Reference (Reference)

  1. Western Media: Spanish China-Knowing Forum “China Builds 126K+ High-Quality Datasets” (Reference News Sep 5 reprint) — https://www.163.com/dy/article/L63123HP0514BQ68.html

XI. Conclusion

In September 2026, China’s data industry is experiencing paradigm leap from “14th Five-Year Plan” wrap-up to “15th Five-Year Plan” opening. From “top-level tone-setting” to “regional coordination + data opening + industrial scale” three-track parallel, from “data element × AI” bidirectional empowerment deepening, data industry hard data (6.78T + 482K + 17.9%) with regional landing (Jiangsu 5 actions / Shenzhen 5 tasks), central SOE data opening (14 / 6B+), authoritative tone-setting (Sep 8 Liu Liehong signed article + Sep 11 People’s Daily Token Surging) four-dimensional signal stacking, marking China’s data industry formally entering “scale maturity + regional coordination landing + data opening deepening” three-track parallel new stage.