AIR APACAI Readiness Index · Powered by Exeter Labs

CHINAQ2 2026

China keeps a current record.

66 of 216 ranked companies reach the published APAC top 1,000.

Download the dossierPDF · 20 pages · 4.7 MB

Executive summary

A current, cited record

Readiness 60 of 100, range 53-67. Visibility 40.9 of 100 across 216 scored companies.

A low score here reflects the public record, not a company’s capability. China is among the record leaders: of the five signals the index reads, its record leans hardest on freshness and authority and thinnest on capability visibility. The conditions companies here operate under score 60, and what those companies publish scores 40.9. Of the two, the second is the one companies here control, and the quicker to change: 93% of the 216 companies with enough of a public trail to score publish no proof of capability at all.

What companies in China publish is recent and it gets cited. The 15 that also publish proof of capability average 67 of 100.

Exhibit 1. Tier mix in this market
Tier mix in ChinaEmerging 150, Findable 46, Building 18, Proven 2, of 216 listed companies in China. 2026-08-02T21:45:08.928542 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

66 companies in China reach the published APAC top 1,000.

150 of 216 are scored already, next in line for the published list.

60Readiness score for China. Range 53-67.
40.9Average visibility score across 216 scored companies. Public evidence only.
66Companies here that reach the published APAC top 1,000, of 216 scored.

Why visibility is the asset

Invisible is not the same as absent

Buyers, investors and partners treat it that way. Most decisions about a company are shaped before anyone is contacted, from what they can find. The record is the only version of it most people ever meet. and here most of the market already leaves one, which moves the question from whether a record exists to what it says about capability. In China, 106 of 335 companies screened leave too little behind to assess, which is the addressable group rather than the verdict.

Capital and customers screen before they call

Investors, corporate buyers and procurement teams build a shortlist from public material long before a meeting exists. A company absent from that material is not rejected on the merits. It never reaches the shortlist.

Partners and talent read the same record

Prospective partners, joint-venture counterparties and senior candidates all check what an organisation says about its own AI work. Silence is read as inactivity, which is often wrong and rarely without cost.

AI assistants now answer the question first

When someone asks an AI system what a company is doing with AI, the answer is assembled from public sources. A company that publishes nothing cedes its own description to whatever else the model finds.

Most companies we look at are doing more with AI than their customers, investors and partners can see. The work is real, the evidence of it is thin, and people decide on the evidence.
Indhran SegharFounding Director, AIR APAC
Proof concentration in China
Proof concentration in China15 of 216 ranked companies publish capability evidence, averaging 67 of 100. 93% publish none yet. 2026-08-02T21:45:08.985409 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

Evidence quality mix in China
Evidence quality mix in ChinaShare of ranked companies in this market on each evidence grade A to D. 2026-08-02T21:45:08.995018 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

66Companies here reach the published APAC top 1,000.
62%Of 216 scored rest on grade A or B evidence.
150Scored already, and next in line for the published list.

Where the signal is strong

China’s strongest and weakest public signals

The five signals are not on one scale. Nearly every company in the index is discoverable and nearly none publishes proof of what it can do, so the longest bar below is a property of the index before it is a property of China. Read against its own other four signals, this record leans hardest on freshness and authority and thinnest on capability visibility. The first is where companies here have already shown they can publish; the second is where the same effort has not yet gone, and it is the clearest place to start.

Exhibit 2. Average score by signal in this market
Average score by signal in ChinaDiscoverability: 57.2 of 100. AI & Digital Evidence: 47.2 of 100. Platform & Data Footprint: 34.2 of 100. Freshness & Authority: 51.7 of 100. Capability Visibility: 4.6 of 100. 2026-08-02T21:45:08.940946 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

Signal shape, China’s five signals
Signal shape, China's five dimensionsDiscoverability: 57 of 100. AI & Digital Evidence: 47 of 100. Platform & Data Footprint: 34 of 100. Freshness & Authority: 52 of 100. Capability Visibility: 5 of 100. 2026-08-02T21:45:08.971905 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

The lean is freshness and authority. 15 companies in China publish capability evidence, averaging 67 and well clear of the market. It works where it exists. The other 93% have yet to publish it, against 83% regionally.

Work published only in China’s own language is understated here, not absent.

Four tiers of visibility

Four tiers, and where companies in China sit

A tier is a position in the published regional ranking, not a grade anyone awards. It says how much of a company’s AI work a buyer or an investor can verify, and nothing at all about whether that work is any good.

TierChinaWhat it means
Tier 1Proven2Region rank 1-100. Publishing enough, recently enough, and across enough of the five signals that the public record alone establishes an AI position.
Tier 2Building18Region rank 101-400. A consistent public trail, usually strong on two or three signals and thin on evidence of hiring.
Tier 3Findable46Region rank 401-1,000. Credibly active and possible to verify, but sparse enough that it takes a deliberate search to find.
Tier 4Emerging150Ranked beyond 1,000. Clears the evidence threshold but does not reach the published list. Real companies, scored and unlisted, and the largest group in almost every market.

2 of the 216 companies in China that clear the threshold reach tier 1, and 150 are emerging, 69% of those scored. Emerging is the normal state of the region, and the group standing closest to the tier above.

Moving up a tier is a disclosure decision, not a technology one, because every signal this index reads is something a company already controls.

The visibility ladder in China
The visibility ladder in ChinaScreened 335, scored 229, listed 216, published 66. 2026-08-02T21:45:09.005163 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

Sector strengths in China
Sector strengths in ChinaTech/Fintech: 41 companies, mean 42.9. Professional Services: 41 companies, mean 41.4. Financial Services: 50 companies, mean 41.4. Logistics/Supply Chain: 22 companies, mean 41.2. Manufacturing: 38 companies, mean 40.6. 2026-08-02T21:45:09.017605 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

National champions in China, top ten by visibility
National champions in China, top ten by visibilityChinasoft International 72. DiDi Global 68. Alibaba Cloud 63. Industrial and Commercial Bank of China 63. Geely Automobile Holdings Limited 62. Kingsoft Office 61. Tuya Smart 60. Alibaba Group 59. AsiaInfo Technologies 59. Bank of China 59. 2026-08-02T21:45:09.056283 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

60CERS (0 to 100)triangulated market signal
53-67CERS band
BConfidence grade
213Domestic signal entities
40.9Mean raw RPScompany visibility

Source: AIR APAC AI Readiness Pulse, Q2 2026. CERS is a market-level signal, not a company readiness score. View sources.

Market context

Macro, state, and enterprise benchmarks for China.

BenchmarkValueRole
Country Readiness Prior (CRP v0.1)68.2Macro and state readiness prior
CRP v0.2 (three-block)81.7Full triangulated prior
Oxford Government AI Readiness 202472.0Macro input to CRP
IMF AI Preparedness Index 202363.5Macro input to CRP
Enterprise adoption (Salesforce 2023)43.1%Business AI adoption block
Exhibit 3. China's readiness score and published range
China's readiness score and published rangeChina 60, range 53-67. 2026-08-02T21:45:08.960458 image/svg+xml Matplotlib v3.11.1, https://matplotlib.org/

Source: AIR APAC AI Readiness Index, APAC 5000, Q2 2026. Scores measure public evidence, not verified capability.

China carries the enterprise-adoption block. CERS triangulates the macro prior, public company signal, and enterprise adoption with a coverage-based shrinkage; the precise integer is indicative, so a band is also published.

Public company signal

What the indexed companies in China show.

SignalValue
Mean raw RPS visibility40.9
Sector-stratified signal40.6
Supplier-readiness signal41.0
Domestic-adopter signal43.8

Top organisations

The highest public-signal organisations in China.

RankOrganisationSectorRPSEvidence
15Chinasoft InternationalProfessional Services72.0B
49DiDi GlobalLogistics/Supply Chain68.0B
137Alibaba CloudTech/Fintech63.0A
145Industrial and Commercial Bank of ChinaFinancial Services63.0B
174Geely Automobile Holdings LimitedManufacturing62.0B
239Kingsoft OfficeProfessional Services61.0A
273Tuya SmartTech/Fintech60.0B
280Alibaba GroupTech/Fintech59.0A
281AsiaInfo TechnologiesProfessional Services59.0B
285Bank of ChinaFinancial Services59.0C
289China Construction BankFinancial Services59.0B
290China TelecomTech/Fintech59.0B
292Industrial Bank Co., Ltd.Financial Services59.0B
303Inceptio TechnologyLogistics/Supply Chain59.0B
323Ping An BankFinancial Services59.0A
339Vipshop HoldingsRetail/E-commerce59.0B
343Yitu TechnologyTech/Fintech59.0B
344Yum China HoldingsRetail/E-commerce59.0A
379ZTE CorporationManufacturing58.0B
385Ant GroupFinancial Services57.0A

Ranked by raw public signal (RPS), a visibility measure. Open any organisation for its evidence profile.

Limits and what this is not

What this report does not claim

Printed with the findings rather than in an appendix at the back.

Known blind spots in this market

China carries 2 this cycle: evidence is gathered from English-language search, likely undercounting local-language activity; below the full three-source-block coverage standard. Each is a direction we know the number leans in. The country evidence behind the readiness score is graded B on source coverage of macro=1, enterprise=1.

A high score means published, not proven

A high visibility score means a company publishes a lot that can be found about its AI work. It does not mean the work is good, in production, or making money. A company here running serious AI quietly will score low here, and it should: this index reads the public record and nothing else.

What would change these numbers

Three things, all scheduled and all documented on page 21. Fuller source coverage per market, which is what currently sets the width of every readiness range. Validating the coverage weighting behind those ranges, which is a stopgap in this edition. And merging companies that operate several web domains, so one group cannot appear twice. Nothing here was audited inside a company, and no verified-readiness score is published this cycle. Next edition: Q3 2026.

english_web_undercoveragebelow_full_crp_v02_coverage

What to do with a visibility score

A visibility score is something a company here can move, because every signal behind it is something it already controls. Three steps separate it from a record that shows what it is actually doing.

Start with the company’s own row

Every scored company in China appears in the dataset with its five signal scores and an evidence grade. A grade C or D row is usually a company that has not said much, not a company that has not done much.

Then measure what this cannot see

This index reads the outside only. Leadership alignment, governance, data quality and workforce capability are invisible to it by design, and that is where the real answer sits.

Correct the record on purpose

Proof of capability is the region’s weakest signal, because companies announce intent faster than they show work. Job posts, named leads and dated case studies move the score honestly.

None of this requires a new AI programme. It requires publishing the one that already exists.

The full China dossier, as published.

66 of the 216 ranked companies in China reach the published APAC top 1,000.

Macau has no benchmark block and is reported as contextual. The dossier counts companies on a wider definition than this page does and the two reconcile: how the company counts fit together.

Assess readiness