Frequently Asked Questions

‘Little Giant’ (小巨人) is Beijing’s term for small and medium-sized enterprises (SMEs) it judges capable of occupying leading positions in critical technologies and industries. It simultaneously helps China reduce its dependency on Western technology while making the rest of the world more dependent on Chinese technology.

The original 2015 implementation guidelines for Beijing’s signature Made in China 2025 strategy, which mapped out how China would achieve industrial dominance in key technologies by 2025, explicitly mention the Little Giant program.

Little Giant status signifies that China’s Ministry of Industry and Information Technology has formally assessed a company as possessing specialised technological capabilities, a strong position in a niche market and relevance to national industrial priorities. While foreign governments, businesses and universities have concentrated scrutiny on globally recognised companies such as Huawei and Alibaba, Little Giants are generally smaller, less visible and deeply embedded in specialised supply chains. Yet they are central to China’s long-term ambition to secure its place at the heart of global supply chains across a range of critically important technologies and sectors.

Designated Little Giants receive preferential benefits, such as tax credits and reduced certification fees, alongside indirect support, such as credit access and promotion at domestic and international trade fairs.

Status is not permanent: companies undergo triennial review and can be decertified if they fail to maintain qualifying performance. Successful companies can be elevated to ‘Manufacturing Champion’ status — a designation for mid-size companies recognised as established global market leaders in their technology area.

The Little Giant eligibility criteria have evolved through several official notices since the program was first established in 2018. The 2018 launch criteria, the 2022 unified national framework and the 2026 revised framework are the three principal benchmarks for the Little Giants program, with further adjustments in intermediate batch notices.

Across these benchmarks, five criteria have remained important:

• Focus — sustained, deep specialisation in a single narrow segment or product line. Companies must demonstrate a track record of successfully operating in niche markets.
• Market standing — a leading, recognised position within that niche with a demonstrated market share.
• Innovation — technical depth evidenced by R&D investment, a qualifying intellectual property portfolio (particularly patents) and participation in standard-setting.
• Strategic relevance — active operation in a state-prioritised technology domain and/or a critical position in a domestic or global supply chain.
• Business quality — financial health (revenue scale, growth trajectory, leverage ratios) combined with management maturity, including governance standards and digitalisation capability.

Two further requirements function as pass/fail gates rather than graded criteria:

• Since 2022, the firm must already hold recognition as a provincial-level specialised SME (专精特新中小企业), the tier below national Little Giant status in China’s innovation cultivation hierarchy.
• The firm must carry a clean compliance record, with no serious safety, quality, environmental, credit or tax violations in the years preceding nomination.

Little Giants are selected annually in batches by China’s Ministry of Industry and Information Technology (MIIT) but nominated by sub-national authorities (a mix of provinces, regions and cities — outlined in a separate FAQ response: ‘What are nominating authorities?’).

The selection process has undergone one significant structural change that affects how the Tracker’s data is sourced and verified.

Batches 1–3 (2019–2021) were announced centrally by MIIT as a single national list, without identifying which sub-national authority had nominated each company. From Batch 4 (2022) onwards, the process was decentralised: each of the 37 nominating authorities now announces its own list of successfully nominated companies.

The program’s first batch, announced in 2019, designated 248 companies. Seven rounds have now been completed, with the seventh batch (announced in 2025) designating an additional 3,482 companies. Batch 8 is expected in late 2026 and will be incorporated into the Tracker promptly upon released.

The program operates on a rolling basis. Companies are subject to triennial review, meaning a designation from Batch 1 (2019) would first come up for review in 2022, and again in 2025. This rolling structure means the Tracker’s database at any point in time reflects a mix of new, actively maintained and lapsed designations — all of which are tagged accordingly to ensure users can distinguish current from historical status.

Nominating authorities are the sub-national government bodies responsible for identifying companies within their jurisdiction that meet the Little Giant criteria and formally nominating them to the Ministry of Industry and Information Technology (MIIT) for designation. In each case, they are the provincial or municipal department responsible for industry and information technology in that administrative unit — the local counterpart to MIIT.

There are 37 nominating authorities in total:

• 22 provincial authorities
• 5 autonomous region authorities (Guangxi, Inner Mongolia, Ningxia, Tibet and Xinjiang)
• 4 provincial-level municipality authorities (Beijing, Chongqing, Shanghai and Tianjin)
• 5 authorities for cities with independent planning status (Dalian, Qingdao, Ningbo, Shenzhen and Xiamen)
• 1 authority for the Xinjiang Production and Construction Corps (the Bingtuan)

The Tracker identifies the nominating authority for over 99.5% of companies designated from 2022 onwards.

Yes. Little Giant status is not permanent. It is subject to triennial review, and companies that fail to meet the program’s ongoing requirements are decertified and lose access to the preferential government support — financing, subsidies, regulatory facilitation and state collaboration — that designation confers.

Grounds for decertification include a mismatch between the company’s technology focus and the state-identified chokepoint priorities it was designated to address; insufficient product competitiveness within the designated niche; and failure to maintain a qualifying patent portfolio aligned with those chokepoint technologies.

Decertification is not uncommon. Of the 1,744 companies designated in Batch 2 (2020), only 1,079 retained their status through the 2023 review — a removal rate of approximately 38%.

Companies that successfully navigate the review cycle and demonstrate sustained outperformance can be designated a ‘Manufacturing Champion’— a higher-tier designation for enterprises that have established global market leadership in their niche.

Because reviews are triennial rather than annual, each review batch encompasses companies from multiple earlier cohorts. The 2025 review, for example, covered Batch 4 companies (designated in 2022) alongside Batch 1 companies that had already passed their 2022 review. The figure below illustrates this overlapping cycle.

The Tracker currently covers 19,444 companies that have been designated as Little Giants, spanning designations announced between 2019, the program’s first year, and 2025.

Note, however, that the total number of companies identified across the seven batches to date is 19,308 — a difference of 136 entries. This is because these 136 entries each represent instances where a company appears more than once across batches.

There are two reasons for this.

First, a company can lose its Little Giant status following a failed triennial review and be renominated in a later round, generating two separate entries in the batch totals. This occurs 110 times in the database.

Second, a small number of companies appear to have been nominated twice without an intervening failed review, most likely reflecting announcement errors in officially published lists. This occurs 26 times in the database.

In both cases, the Tracker consolidates these to a single unique company record, with the full batch history retained. The distribution across batches is as follows:

Across the 19,308 companies, 89.9% are tagged as currently being ‘Active’, indicating that, as of 2025, they are designated as Little Giants. 3.4% are tagged as ‘Inactive’, indicating that they failed their most recent triennial review and lost their Little Giant designation. 6.7% are tagged as ‘Unknown’, indicating that the outcome of their most recent review is not known, and consequently, their designation status is also unknown. Because of how the earlier batches were announced, most companies (80%) with an ‘Unknown’ Little Giant designation are in fact inactive. But the incomplete source material makes it hard to separate this active minority from the inactive majority. In practice, when the Tracker labels a company’s Little Giant designation as ‘Unknown’, it is more likely that they lost their Little Giant designation than that they retained it. Users should verify this manually where relevant.

No. The Tracker only covers national-level Little Giants.

Beijing operates a multi-tiered system for identifying and supporting high-potential SMEs. At the base are provincial-level ‘innovative SMEs’ (创新型中小企业) — companies recognised by their local authority as having meaningful specialisation and innovation potential within their sector. Above them sit provincial-level ‘specialised SMEs’ (专精特新中小企业), which demonstrate stronger technical depth and market position.

National-level Little Giants sit above both: companies the Ministry of Industry and Information Technology has assessed as capable of addressing technology chokepoints of national significance. Since 2022, being recognised as a provincial-level ‘specialised SME’ is itself a mandatory gateway requirement for national-level Little Giant nomination — meaning every company that has gained Little Giant designation since then has already cleared the lower tier before being assessed at the national level.

The provincial tiers are substantially larger: the Tracker’s current scope of national-level Little Giants includes over 19,000 unique companies. The population of provincial-level innovative SMEs and specialised SMEs runs into the hundreds of thousands. Including that tier would dilute the Tracker’s signal-to-noise ratio and shift its focus away from the companies of greatest strategic and due-diligence relevance.

Inclusion in the Tracker means a company has been formally designated by the Chinese government as capable in a technology domain Beijing has identified as important to national interests. That designation is not simply honorary. It comes with access to preferential state financing, direct subsidies, regulatory facilitation and structured collaboration with state research institutions.

Designation does not, by itself, establish civil-military fusion involvement or malign activities. Many companies focus on commercial markets, such as consumer electronics, with only indirect, if any, connection to China’s military objectives. However, the Tracker includes a range of data points that assist in making such inferences.

For example, strong activity in advanced aircraft engines, combined with military-relevant patenting and a connection to Beihang University — an institution that conducts extensive defence research — would support an inference that a company is contributing to China’s defence efforts.

The nominating authority field is ‘Unknown’ where no explicit source links the company to a specific sub-national authority. This arises from two distinct causes:

• Batches 1–3 (2019–2021). The first three batches were announced centrally by the Ministry of Industry and Information Technology in national lists, without identifying which authority had nominated each company. The nominating authority for these companies cannot be verified from the announcement lists alone.
• Batch 4 onwards. From Batch 4 (2022), each of the 37 nominating authorities published its own list of successfully nominated companies, making attribution possible in the overwhelming majority (99.5%) of cases. The residual cases in this group — affecting fewer than 0.5% of Batch 4 onwards companies — reflect instances where a company’s successful nomination is confirmed but the attributing source cannot be located or verified.

In many cases where a company’s nominating authority is listed as ‘Unknown’, a nominating authority can be reasonably inferred — for instance, from the company’s name which often contains geographic markers such as city names, or from a later review announcement — but inferred attributions are not entered into the Tracker without an explicit source.

The China Little Giants Tracker is built on four primary data sources, each serving a distinct function in the dataset’s construction.

1. Official Little Giant company lists

The foundation of the Tracker is a consolidated set of Little Giant announcement and review lists, collected through systematic desktop research across all seven completed batches (2019–2025). These lists include initial designation announcements, triennial review outcomes, and company name updates, which are typically published together with announcement and review lists, and are cross-referenced to maintain an accurate and current record.

Where official government sources are available — principally successful selection lists published by either the Ministry of Industry and Information Technology or the relevant nominating authority — these are prioritised. Supplementary online sources are used where official records are incomplete or inaccessible. All lists are cross-checked against one another to confirm accuracy and identify inconsistencies.

2. Google Patents

Google Patents is used to make an initial identification of English-language names under which Chinese companies file patents — a non-trivial challenge given the variation in how Chinese companies transliterate and render their names in international patent filings. This matching process is essential for linking Chinese company records to their international patent activity.

3. Derwent World Patents Index

The Derwent World Patents Index (DWPI) is the primary source for building each company’s patent portfolio. DWPI is one of the world’s most comprehensive patent databases, covering filings across major patent offices globally. It is used to support several data points in the Tracker, including the technology activity rating, the presence of military patents and connections to defence universities. The quality and depth of a company’s patent portfolio directly determines the reliability of these data points.

4. Public corporate data

Open-source, publicly available corporate data is used to supplement the analytical data points derived through patents analysis. It provides an additional filtering and identification layer, making it easier for users to navigate to companies of interest. This layer includes data points such as the Unified Social Credit Code (USCC) and the sector and industry in which each Little Giant operates in.

5. Wikidata

Wikidata data is used to identify English names for Chinese prefectures and cities identified from the Little Giant’s public corporate data. This data is reviewed and modified to ensure correctness and consistency across the over 2000+ administrative division names used in the Tracker.

The Tracker uses seven tags to indicate what has happened to a company at each stage of its involvement in the Little Giant program (specifically when it joined the program, and the outcome of its triennial reviews):

• Selected: The company was selected to enter the Little Giant program.
• Passed: The company passed its triennial review.
• Failed: The company failed its triennial review and was subsequently removed from the Little Giant program.
• Unknown: The Tracker cannot confirm whether the company passed or failed its triennial review.
• Failed, Batch Unknown: The company failed one of its triennial reviews, but the Tracker cannot say which one.
• Presumed Pass: The Tracker has no direct confirmation that the company passed its triennial review, but given that it appeared in a later review, it must have passed the earlier one. Example: A company was designated a Little Giant in 2019. It therefore would have been reviewed in 2022, and if it passed, in 2025. If the Tracker has explicit confirmation that the company passed its 2025 review, then it must also have passed its 2022 review. In such a case, if explicit confirmation of the 2022 review outcome is missing, it will be tagged as a ‘Presumed Pass’.
• Presumed Fail: The Tracker has no direct confirmation that the company failed its triennial review, but given that it later re-entered the program, it must have been removed from the Little Giant program at some point, implying that it failed the review. Example: A company was designated a Little Giant in 2020. It therefore would have been reviewed in 2023. If it was re-designated as a Little Giant in 2024, then it must have failed its 2023 review, even if the Tracker lacks explicit confirmation of this fact. In such a case, if explicit confirmation of the 2023 review outcome is missing, it will be tagged as a ‘Presumed Fail’.

The last two tags are used to infer the likely outcomes of as many Little Giant review events as possible where the source data is incomplete.

The China Little Giants Tracker uses the technology taxonomy developed by ASPI’s Critical Technology Tracker (CTT), available here. For each technology in that taxonomy, an AI model was used to iteratively identify Cooperative Patent Classification (CPC) codes relevant to it. This classification was then audited and validated by human inspection. Over 600 codes are used to map the 74 technologies currently tracked in the CTT.

The mapping for advanced optical communication, for example, includes codes such as:

• H04B10/25 — Fibre-based transmission systems employing electromagnetic waves other than radio waves
• H04J14/00 — Optical multiplex systems
• H01S5/026 — Monolithically integrated semiconductor laser components
• G02B6/02 — Optical fibres with cladding

How the code hierarchy works

CPC codes operate in a hierarchy. When a broad group code is specified, the analysis pipeline automatically captures every specific subgroup beneath it. For example, specifying H04B10/00 also captures H04B10/03 (arrangements for fault recovery). At a subgroup level, literal string prefix matching is used. If only a specific part of a technology group is relevant to the Tracker, we explicitly list those exact subgroups instead of the parent code.

Because broad parent codes can implicitly cover many subgroups, the exact number of codes explicitly assigned to a technology varies considerably:

• Median: 8 codes per technology.
• Highest Density: Advanced integrated circuit design and fabrication (32 explicitly listed codes).
• Lowest Density (Categorisable): Adversarial AI, geoengineering, gravitational-force sensors and post-quantum cryptography (one explicit code each). Note on hierarchy: The single code for gravitational-force sensors (G01V7/00) acts as a parent code, implicitly capturing many subgroups below it. Conversely, the single code for adversarial AI (G06N3/094) is a highly specific child code with no further sub-categories. A direct comparison of code counts therefore obscures an important distinction.
• Uncategorisable: Hypersonic detection and tracking (no codes). Because it lacks applicable CPC codes, this technology cannot be categorised under the Tracker’s current methodology.

Patent classification codes are a standardised tagging system used by patent examiners to categorise patents by technical field. Two systems are in widespread use: the International Patent Classification (IPC) and the Cooperative Patent Classification (CPC). Developed jointly by the European Patent Office (EPO) and the United States Patent and Trademark Office (USPTO), the CPC system is a more granular extension of the IPC, the treaty-based system administered by the World Intellectual Property Organization (WIPO) that serves as the universal classification standard. We build our technology mapping on CPC rather than IPC because its finer resolution and continuously updated scheme distinguish emerging and critical technology subfields more precisely than the broader IPC categories, which are often too coarse to separate the technologies of interest.

When a patent is filed and examined, examiners assign one or more CPC or IPC codes reflecting the technical areas the invention relates to. A single patent routinely carries multiple codes — a sonar system patent, for example, might be tagged across signal processing, underwater acoustics and materials science. This multi-code structure means that a company’s full patent classification profile, aggregated across its entire patent portfolio, provides a granular map of its technological activity that is more precise and more defensible than self-reported technology descriptions or sector classifications.

CPC codes for Chinese patents in this dataset are mostly not assigned by China’s primary intellectual property regulator, the China National Intellectual Property Administration (CNIPA), which applies CPC codes to only a subset of the patent applications it examines. The CPC codes are primarily drawn from Derwent World Patents Index, which aggregates classifications from the issuing offices and from the EPO’s reclassification of Chinese patents, propagated across the set of patents that relate to a single invention (i.e. the patent family).

The CPC system comprises approximately 250,000 codes organised in a hierarchical structure: nine top-level sections (A through H, plus Y for emerging cross-sectional technologies), subdivided into classes, subclasses, groups and subgroups. The hierarchy runs from broad to highly specific — a main-group code such as H04B10/00 (transmission systems using electromagnetic waves other than radio waves) encompasses every more specific subgroup beneath it, while a subgroup code such as H04B10/2519 identifies a precise technical configuration within that family.

Because CPC coverage is incomplete for Chinese patents, we also score each portfolio under the IPC system — which nearly all patents reliably carry — and use the higher of the two scores. As the CPC is an extension of the IPC, the IPC version of each technology mapping is derived from its CPC codes rather than built separately, with codes that don’t cleanly map to the IPC system dropped. Codes are dropped either because they sit in the CPC-only Y section for codes relevant to new and emerging technologies or because the nearest IPC code is too broad to distinguish the technology from closely related technologies.

The Tracker’s technology activity ratings are built on patent classification codes. For each of the 74 critical technologies in ASPI’s Critical Technology Tracker taxonomy, ASPI has identified the specific CPC and IPC codes that define activity in that technology. A company’s patent portfolio is then analysed against these mappings: the proportion of the company’s patent classification code instances that fall within a given technology’s definition determines its alignment score, which feeds directly into its technology activity rating.

This approach is grounded in examiner-assigned classifications rather than company self-description, making it resistant to the kind of strategic framing that companies — particularly those with incentives to appear aligned with state priorities, or to obscure dual-use research — may apply to their own public communications.

The technology activity rating is the Tracker’s core analytical output for each company-technology pair in the database. It reflects the degree of alignment between a company’s patenting activity and a particular technology. It is not a measure of quality. A company with a ‘Very strong’ rating in radar is not necessarily stronger in that technology than a company with a ‘Weak’ rating. Rather, the company with a ‘Very Strong’ rating has clearer evidence that the technology forms a central part of its business operations.

The rating combines two distinct dimensions:

• Alignment — how closely a company’s patent portfolio maps to a given technology, as defined by the Cooperative Patent Classification (CPC) and International Patent Classification (IPC) codes ASPI has identified for that technology area. A company whose patents are concentrated in classification codes associated with radar or post-quantum cryptography, for example, will carry a high alignment score in that technology. A company with patents spread across many unrelated technical fields will carry a lower alignment score in any single technology.
• Confidence — how much weight the alignment score can bear given the size of the underlying patent portfolio. A company with three patents, all in advanced aircraft engines, has perfect alignment in that technology, but three patents is a thin evidential base. A different company with 200 patents, of which a similar proportion relate to advanced aircraft engines, provides a far more reliable signal. The confidence dimension captures this: it rises with portfolio size, reaching its halfway mark at 50 patents and continuing to climb steadily for larger portfolios.

The rating starts from the alignment score and then caps it at the confidence score. In plain terms, the rating reflects the share of a company’s patent activity in a technology — held back only when the company has too few patents to trust that share. A company with deep alignment but few patents is capped at a low rating by its thin evidence; a company with a large portfolio but little alignment to the technology simply carries a low alignment score, and so a low rating.

This rating is then mapped to one of five rating bands:

The technology activity rating is derived from a composite score that combines two underlying values — an alignment score and a confidence score — for each company-technology pair in the database. The composite score is scaled to 100 and mapped to one of five rating bands. This runs separately for each classification system (IPC and CPC); the higher of the two scores is used.

Step 1: Alignment score. The alignment score measures the share of a company’s patent classification code instances that fall within the definition of a given technology. Specifically, it is calculated as:

(number of a company’s patent classification code instances matching the technology) ÷ (total patent classification code instances in the company’s portfolio)

A company whose patent activity is heavily concentrated in a single technology will carry a high alignment score for that technology. A company with a broad and diversified patent portfolio will carry lower alignment scores across multiple technologies.

Step 2: Confidence score. The alignment score alone can overstate confidence in identifying a company’s activity in a technology. A company with two patents, both squarely in a target technology, has perfect alignment but a thin evidential base. The confidence score caps the alignment score by portfolio size. It rises from zero as the patent count increases and climbs smoothly towards 1. It is calculated as n² ÷ (n² + 50²), where n is the number of the company’s patents carrying classification codes: five patents give about 0.01, 50 patents give 0.50 and 100 patents give 0.80. By 500 patents, the confidence score is 0.99.

Step 3: Composite technology activity score. The rating takes the smaller of the alignment and confidence scores, scaled to a value out of 100:

Because a company’s confidence is usually higher than its alignment in any one technology, the score is normally just the alignment — the share of the company’s patent activity in that technology. Confidence only changes the result when a company has so few patents that its confidence falls below its alignment; then it acts as a ceiling, pulling the score down to reflect the thin evidence.

Step 4: Rating bands. Each company-technology pair is assigned to one of five bands:

A minimum activity score of 2 filters out the long tail with very limited activity in a technology. In practice, to receive a rating a company must meet both of these conditions:

• It has been assigned at least seven patents carrying patent classification codes.
• At least 2% of those code occurrences match the codes defining a given technology.

Worked example:

Consider a fictious company, Galactic Light Solutions Co. Ltd. (GLS). GLS has six patents assigned to it, with simplified CPC codes listed in each as shown below (multiple codes on a single patent are separated by ‘ | ’):

This gives 12 CPC code instances across GLS’s portfolio. We then count how often each CPC code appears across GLS’s patents, and identify the technology (if any) it falls under:

Grouping by technology and calculating the alignment score, confidence score and technology activity score:

GLS’s patents substantially align with advanced optical communication — more than half its CPC activity maps to that technology — but its portfolio of only 5 patents with CPC codes produces a confidence score of 0.01. This caps the activity score at 1, meaning GLS would not have received an activity rating for either of the two technologies.

Had GLS maintained those same alignment scores, but with 40 patents (corresponding to a confidence score of 0.39) instead of 5, the ceiling would sit higher, and advanced optical communication would be capped at 39 while nanomaterials and nanoscale manufacturing, whose 17% alignment is already below that ceiling, would score its activity directly at 17. This would have placed the two activity ratings at ‘Strong’ and ‘Moderate’ for advanced optical communication and nanomaterials and nanoscale manufacturing respectively.

A technology activity score calculated from a large patent portfolio is more reliable than one inferred from a handful of patents, so confidence rises with portfolio size and gives more weight to better-evidenced profiles.

Turning portfolio size into a confidence weight requires a curve with a few specific properties: it should be zero when a company has no patents, rise steadily as the portfolio grows and approach full confidence smoothly rather than jumping at a cut-off. A simple ratio, n² ÷ (n² + 50²), has exactly these properties, and a single setting — the patent count at which confidence reaches its halfway point, here 50 — controls how quickly confidence builds.

Yes. A company might be highly active in a technology space without that being reflected in the Tracker. There are three primary reasons for this discrepancy:

• Taxonomy scope. The Tracker relies on ASPI’s Critical Technology Tracker taxonomy, which focuses exclusively on 74 technologies chosen for their defence, economic and societal importance. It does not cover the entire technological landscape. For instance, a company could hold a massive patent portfolio in petroleum refining, plastics or wastewater treatment, but because these fall outside the Tracker’s strategic scope, the company’s classification in these technologies would be absent.
• Patent classification methodology. The Tracker uses a conservative methodology for assigning patent classification codes. It only counts codes definitively linked to the latest, cutting-edge advances in each technology, ignoring superseded technical approaches or tangentially related codes. Consequently, some technologies are systematically underrepresented. The most extreme example is hypersonic detection and tracking: because there no codes that meet this strict criterion, no company will be classified as active in this technology in the Tracker.
• Patent coverage. The Tracker measures the public patent record, which is not a complete audit of a company’s actual capabilities. A company might be making significant progress in early-stage research that is not yet ready to patent, or it might rely on unpatented trade secrets and proprietary processes. And if a company files patents through corporate structures that the Tracker’s data sources do not monitor, such as through subsidiaries, those patents will not be counted.

Because of these limitations, the Tracker’s technology activity ratings should be treated as an indicator rather than an absolute fact. While a strong activity rating indicates a well-documented patent footprint, an absent rating is evidence of an absence of meaningful activity. In high-stakes due diligence, the Tracker’s output should be just one piece of a broader assessment.

Defence patents are identified using Derwent Manual Codes, a classification system separate from the IPC and CPC codes used to determine a company’s technology focus areas. While IPC and CPC are assigned by patent examiners and describe a patent’s technical content, manual codes — assigned by Clarivate, which maintains the Derwent World Patents Index — describe an invention’s novelty and application: what a patent could actually be used for.

It is worth emphasising that this does not indicate who uses the patent. Classifying a patent as a defence patent does not in itself mean it is actively being used by the PLA. The following Manual Codes have been used:

In the Tracker, Manual Codes are prefix-matched the same way classification codes are: ‘W07’ matches every Manual Code beginning with ‘W07’. Most defence patents in the dataset match via the W07 manual code.

Links to institutions in the China Defence Universities Tracker (CDUT) are identified by being a patent co-assignee with Little Giants. If a university has jointly filed a patent with a company, that is a strong signal that there is some connection between the two entities.

A CDUT institution must be identified at least twice as a co-assignee on a Little Giant’s patent before that connection is established in the Tracker. This is to reduce the likelihood of noise in the data leading to the suggestion of a connection that doesn’t exist.

Note that this is a narrow way to define connections, and there are many more types of connections that could connect Little Giants to CDUT institutions, such as commercial relationships, informal R&D support, or shared lineage where a Little Giant was spun out of a CDUT institution. Patent data was chosen as the initial basis for identifying connections because it is tractable and scales across the entire set of Little Giants tracked.

CDUT links should therefore be taken as a floor rather than a ceiling. The links shown are those the Tracker identifies from patenting behaviour, and it is very likely that further connections to CDUT institutions exist.

In general, no. Each company in the Tracker is treated as its own entity, classified according to the name recorded in official Little Giant announcements. Companies are not consolidated across subsidiaries, parent entities, merged entities and restructured corporate groups — each legal entity is assessed on a standalone basis.

Systematically mapping the full ownership structures, restructuring histories and corporate group relationships of the 19,000 plus companies in the Tracker is not feasible, and partial mapping would make consistent classification impossible to maintain at scale. The standalone entity approach ensures the Tracker’s coverage is uniform, reproducible and grounded in verifiable official sources.

We do, however, attempt to indicate when a company is part of a larger corporate structure. This classification is incomplete and reflects the corporate relationships that existed at the time the data was collected.

A company’s Tracker entry — including its technology activity ratings, patent portfolio and sector — reflects its own recorded activity and identity, not that of its corporate group. A state-owned enterprise with multiple Little Giant subsidiaries will appear as several distinct entries rather than a consolidated group profile.

The standalone approach means the Tracker will not automatically surface the full corporate context around a company. Users conducting due diligence should independently verify beneficial ownership, parent-entity relationships, subsidiary structures and any relevant restructuring history.

The Tracker is designed to be a starting point for company-level analysis, not a substitute for it. For high-stakes applications — investment screening, export licensing, sanctions compliance, supply chain risk — the Tracker’s output should be treated as an initial signal that directs further investigation, not a definitive characterisation of a company’s full corporate context.

Classification is based on company names as they appear in official Little Giant announcement and review lists, with two categories of adjustment applied to ensure the database accurately reflects the underlying population of designated companies.

Category 1: Transcription error corrections. Where a name in an official source appears to contain a transcription or OCR error — and where no company can be located under that name — a minimal correction is applied, subject to strict criteria. All four of the following conditions must be satisfied before any correction is made:

• No company can be found under the original name in any available source
• The corrected name is independently confirmed to be that of a Little Giant presently absent from the dataset
• The correction affects no more than two characters
• The substituted characters are visually similar to those replaced, consistent with a plausible OCR misread

For example, the official Batch 2 list published by the Ministry of Industry and Information Technology contains 宁波佳尓灵气动机械有限公司 — a company that could not be located in any desktop research. The near-identically named 宁波佳尔灵气动机械有限公司, differing only in the character 尓 versus 尔, was confirmed as a Little Giant absent from the dataset. All four criteria being satisfied, the corrected name was adopted.

This type of correction is rare, affecting 0.1% of companies in the dataset. It is applied conservatively because the alternative — retaining an unverifiable name — would introduce phantom entries into the database that cannot be linked to any real company.

Category 2: Name variation consolidation. More commonly, a company’s recorded name varies across different official lists — between its initial designation announcement, subsequent batch announcements and triennial review publications. These variations are consolidated to a single canonical record. Three causes account for most cases.

Genuine name changes. A company may formally change its registered name during its time in the program. Luoyang Longsheng Technology Co. Ltd., a Batch 6 Little Giant, changed its registered Chinese name from 洛阳隆盛科技有限责任公司 to 中航隆盛（洛阳）科技有限责任公司 — a change that also signals a shift in its corporate affiliations, with 中航 (AVIC, Aviation Industry Corporation of China) appearing in the new name.

Changes in corporate form. A common source of apparent name variation is the conversion of a company from a limited liability company (有限责任公司 or 有限公司) to a joint-stock company (股份有限公司), which is a standard step in preparing for a stock exchange listing. Henan Zhongyuan Roller Co. Ltd. (河南中原辊轴股份有限公司), a Batch 3 Little Giant, was previously registered as 河南中原辊轴有限公司. The underlying entity is the same; only the corporate form designation has changed.

Minor transcription or listing differences. Some variations reflect inconsistencies in how official lists record the same company name rather than any formal change. Baoji Safe Petroleum Machinery Co. Ltd. (宝鸡市赛孚石油机械有限公司), a Batch 3 Little Giant, appeared in its review list without the character 市, which marks 宝鸡 (Baoji) as a city-level administrative unit. Its omission does not indicate a different company.

Name variation consolidation is applied throughout the dataset and is materially more common than transcription corrections. At least 5.7% of companies in the database have changed their officially registered name over the course of their participation in the Little Giants program, and over 50% of companies have changed their registered name over their lifetime (including before they were designated as a Little Giant).

Patent count measures the number of Derwent-curated patent families that have been assigned to the corresponding company, with a priority date no earlier than 1 October 2009.

A number of follow-up questions may stem from this explanation:

1. What are patent families? A patent is a single legal grant in one jurisdiction. A patent family is the set of patents across jurisdictions that all stem from the same original invention — so when a company files for the same invention in China, the US and the EU, that’s three filings but one family. Families also absorb patent documents that add no new invention — translations, amendments, divisional and continuation filings — which would otherwise be counted as separate patents.
2. What does ‘Derwent-curated’ mean? Patent families are conceptually straightforward but messy to construct in practice, so several competing methods exist. The China Little Giants Tracker uses Derwent families, rather than the more widely used INPADOC families maintained by the European Patent Office. INPADOC families are built algorithmically from shared priority claims, producing large families that can span multiple inventions depending on exactly how patents are filed. Derwent families add editorial judgement, splitting families where subject matter diverges and merging clear equivalents, which gives tighter, more invention-focused groupings.
3. What is a priority date? The priority date is the date of the earliest application in a patent family. Later filings for the same invention, whether in other jurisdictions or as continuations, claim priority back to that first application and are judged against the state of the art as it stood on that date rather than on their own filing date. Because it marks when the invention was first disclosed to a patent office, the priority date is generally the best available proxy for when the underlying R&D took place.
4. Why is 1 October 2009 the cutoff? That is when the Third Amendment to China’s Patent Law took effect, shifting the country from a relative to an absolute novelty standard. Before the change, an invention already known or in use outside China could still be patented there, so patents with pre-2009 priority dates were not required to clear an international novelty bar and are therefore not directly comparable to later patents. The cutoff restricts the Tracker to patents granted under a consistent novelty standard, supporting fairer comparison between companies.

The Tracker does not distinguish between invention patents and utility models. Invention patents require a novel solution to a technical problem and undergo substantive examination; utility models need only demonstrate an advantage over an existing solution and receive weaker legal protection. Both are counted equally in the Tracker, so the patent count should be read as a measure of patenting activity rather than of technical sophistication.

Companies with no identifiable patents are included in the database, but no-patent derived fields are populated for them. No technology activity rating is generated, and no links to institutions engaged in defence or security-related research are identified. That is because both the technology activity rating and links to institutions are currently derived entirely from patent data. This applies to 0.2% of companies in the Tracker. Core registry information — including the company’s name, nominating authority, batch designation, current status and any available corporate details — remains available.

There are several reasons why a company may appear in the database without identifiable patents.

The most common is that the company has not filed patents under corporate identities captured by the Tracker’s patent data source. Some Little Giants operate in sectors where trade secrets, proprietary processes or regulatory exclusivity are the primary forms of IP protection rather than patents. Others — particularly smaller or earlier-stage companies — may hold patents that are pending, recently filed or registered under a related entity not yet linked to the company’s canonical record in the dataset. In each case, the absence of patent data reflects the limits of the sources rather than a confirmed absence of IP activity.

Sector and industry classifications come with the underlying corporate data and follow China’s national industry classification standard. They are recorded attributes of each company rather than assessments made for the Tracker. Unlike the technology focus — which uses the taxonomy developed for ASPI’s Critical Technology Tracker together with a custom patent-assignment methodology — these classifications are not devised or applied by ASPI.

Instead, the standard used in the Tracker is the industrial classification for national economic activities (GB/T 4754—2017, as amended 2019), maintained by China's National Bureau of Statistics. It is hierarchical, and the Tracker uses its top two levels: industry corresponds to the standard’s division level (大类), and sector to the section level above it (门类), so each industry sits within exactly one sector. This standard is used because it is the one Chinese companies are classified against by their own regulators, so it reflects each company's classification of record rather than an inference drawn by ASPI.

Under the standard, a company’s classification is determined by its principal economic activity. Where operations span activities falling in different industries, the activity accounting for the largest share of value added is taken as the principal activity; where value added cannot be determined, the standard allows it to be identified by share of sales revenue, operating revenue or employees. Classification proceeds from the top down, with the sector determined first and the industry within it second, rather than by selecting the single largest activity outright.

An industry code is recorded against each registered company in China’s market entity registration system. Registration authorities were required to apply the 2017 standard from 1 June 2018, and to convert the codes of all previously registered companies to it by the end of that month. The classification is therefore an administrative record of a company’s registered principal activity and may not capture its current or full range of activities.

The China Little Giants Tracker is available only as part of an ASPI Tracker product combo licence, which includes both the Critical Technology Tracker and the China Defence Universities Tracker. It is not presently available as a standalone product or as an add-on to a single product licence.

Combo licences provide access to all three trackers under a single subscription agreement. Customers with an existing licence to either the Critical Technology Tracker or the China Defence Universities Tracker can upgrade to a combo licence at a discounted bundled rate.

Each seat provides access for one individual licence. Licences can be purchased individually or in volume packages, with per-seat pricing decreasing at higher licence tiers.

An organisation-wide licence provides unlimited access for every user within the licensed organisational unit. You choose the scope that matches how your organisation is structured, and pay only for the coverage you need.

All organisation-wide licences include Single Sign-On (SSO) integration with your IT system. Where one email domain is shared across multiple departments or divisions within your organisation, access is configured to match the unit covered by the licence.

Enterprise-wide licences extend unlimited access across all units of an organisation. Consortium licences allow multiple organisations to pool a single subsidised arrangement for their users.

To scope an organisation-wide, enterprise or consortium licence, contact sales@aspi.org.au.

ASPI is an independent non-profit. Licence revenue funds the research it provides access to.

Data export from the platform is not currently available. Organisations requiring programmatic access or bulk data extraction for integration with internal systems, screening tools or risk workflows should contact the sales team at sales@aspi.org.au to discuss API access options.

You can pay annually upfront by card or invoice or choose monthly instalments (card only) with an annual commitment. All the subscriptions/licences are annual commitments.

Currently, we are only offering demos to customers interested in the unlimited licence tier or with custom requirements. If this suits you, please contact us at sales@aspi.org.au with any requirements you have for a demo.

We do not offer free trials of the tracker. We have made 10 company profiles freely available however so users can understand what the Tracker looks like.