Hangzhou’s AI Firms Embrace Standardization Amid Rapid Growth
In the heart of China’s digital economy, Hangzhou is emerging as a pivotal hub for artificial intelligence (AI) innovation, where a growing number of technology enterprises are not only developing cutting-edge applications but also actively engaging in the standardization process. A recent empirical study conducted by researchers from China Jiliang University, Hangzhou CPPCC Career Development Service Center, and Hangzhou Association of Standardization sheds light on how local AI firms are navigating the complex landscape of technical standards, revealing a strong willingness to shape industry norms despite existing challenges.
The research, led by Lu Hao, Mao Hai-jun, Yu Su-chun, Chen Xin, Chu Fei-fei, and Li Nan-yang, offers one of the most comprehensive snapshots to date of standardization practices within a regional AI ecosystem. Published in a peer-reviewed journal, the findings underscore a critical shift: while AI remains a nascent field globally, Chinese enterprises in Hangzhou are increasingly recognizing that participation in standard-setting is no longer optional—it is a strategic imperative for competitiveness, market access, and long-term sustainability.
Standardization has long served as the invisible architecture underpinning technological progress. From telecommunications to manufacturing, common technical specifications ensure interoperability, safety, and quality across products and services. In the context of AI, where algorithms, data pipelines, and hardware architectures vary widely, the need for coherent standards is even more pressing. Without them, fragmentation can stifle innovation, create barriers to adoption, and raise concerns about ethical use, privacy, and system reliability.
Globally, major economies have acknowledged this urgency. The United States launched its “American AI Initiative” in 2019, emphasizing the development of universal AI standards as a core pillar. The European Union, through its AI Strategy unveiled in 2018, has prioritized the creation of an ethical and legal framework alongside technical benchmarks. In China, the establishment of the National Working Group on AI Standardization in January 2018 marked a formal commitment to building a domestic standardization roadmap, supported by policy directives such as the Three-Year Action Plan for Promoting the Development of New-Generation Artificial Intelligence Industries (2018–2020).
Against this backdrop, the Hangzhou study provides granular insights into how these macro-level policies translate into micro-level corporate behavior. Drawing on a robust dataset collected from 98 AI enterprises—via both face-to-face interviews and online surveys—the research team employed rigorous statistical validation methods, including Cronbach’s Alpha (0.846) and KMO tests (0.819), confirming high internal consistency and suitability for factor analysis.
One of the most striking findings is the demographic profile of the AI sector in Hangzhou. The vast majority of firms—86.74%—are either state-owned or privately owned, with private enterprises alone accounting for over 70% of the sample. This reflects the broader trend in China’s tech industry, where agile startups and privately funded innovators drive much of the AI advancement. Moreover, nearly 60% of the surveyed companies have been in operation for more than a decade, indicating a maturing ecosystem rather than a speculative bubble. These are not fly-by-night ventures but established players with substantial organizational capacity.
The scale of these enterprises further reinforces their potential influence on standardization. Over three-quarters of the respondents employ more than 100 people, with over 30% surpassing the 1,000-employee threshold. Such scale suggests that many of these firms possess dedicated R&D departments, legal teams, and strategic planning units—resources essential for engaging in standard-setting bodies, which often require sustained participation and technical expertise.
When it comes to the nature of their AI development, the study reveals a clear orientation toward application and platform layers rather than foundational technologies. While 46.94% of firms engage in what is classified as the “basic layer”—including data acquisition, intelligent chips, sensors, and operating systems—the majority focus on the “technology layer” (67.35%) and “application layer” (62.24%). The former encompasses cloud platforms, open-source frameworks, algorithm models, machine vision, speech recognition, and natural language processing. The latter includes smart robotics, autonomous vehicles, smart homes, virtual and augmented reality, healthcare AI, financial technology, and intelligent transportation systems.
This distribution aligns with global trends, where much of the commercial value in AI today lies in deployment rather than core infrastructure. It also implies that standardization efforts in Hangzhou are likely to prioritize interoperability between AI services, model performance evaluation, data formats, and API specifications—areas that directly affect integration and user experience.
Crucially, the study finds that half of the responding enterprises—50%—have already developed their own standards. This figure is significant, especially when contrasted with the 27.5% that reported not using any standards and the 22.5% who were unsure. Among those not adopting standards, a notable proportion had been founded within the past five years, suggesting that newer entrants may lack the maturity or awareness to engage in formal standardization processes.
Of those that do develop standards, the primary focus is on national and industry-level specifications. According to the data, 75% of standard-developing firms have contributed to industry standards, while 65% have participated in national standard formulation. Regional (local) standards were cited by 32.5%, and enterprise-specific standards by 35%. Notably, 45% have engaged with group standards—a relatively new category in China’s standardization system that allows consortia of companies, associations, or research institutions to develop voluntary standards more rapidly than traditional state-led processes.
Perhaps most symbolically, four companies reported involvement in international standard development. This is a small number in absolute terms, but it represents a meaningful milestone. International standardization, particularly within bodies like ISO, IEC, or IEEE, requires not only technical excellence but also diplomatic engagement, linguistic proficiency, and institutional support. The fact that even a handful of Hangzhou-based firms are participating at this level signals that China’s AI sector is beginning to project influence beyond its borders.
The adoption of standards follows a similar pattern. A full 92.5% of firms that use standards do so by developing them internally, compared to 42.5% who adopt existing ones. This preference for self-initiated standard creation may reflect a combination of factors: a desire for competitive differentiation, a lack of mature external standards in certain niches, or a proactive strategy to establish first-mover advantage in emerging markets.
Equally telling is the level of organizational engagement. More than half—51.25%—of the surveyed companies are members of formal standardization organizations. Another 23.75% admitted they were unaware of such groups, pointing to an information gap that could be addressed through better outreach and education. Membership in standardization bodies is more than symbolic; it grants companies voting rights, early access to draft specifications, and opportunities to shape the direction of technical consensus.
Product certification, another key indicator of standardization maturity, shows promising uptake. Among the 64 firms that responded to this question, 67.18% reported that their products had been certified against recognized benchmarks. Only 18.76% had not pursued certification, and 14.06% were uncertain. This suggests that a majority recognize the value of third-party validation in building customer trust and meeting regulatory requirements.
However, the geographic scope of certification remains predominantly domestic. Of the certified products, 91.89% carried Chinese certifications, compared to just 16.22% with international recognition. This imbalance highlights a current limitation: while Hangzhou’s AI firms are confident in the domestic market, they may not yet be fully positioned for global expansion, where compliance with international standards such as CE, FCC, or GDPR-related frameworks becomes essential.
Beyond observable behaviors, the study delves into attitudes and awareness. Over 80% of respondents indicated some level of familiarity with AI standardization, with only 18.37% admitting they were unfamiliar. This widespread awareness is encouraging, as cognitive recognition often precedes active participation.
Even more compelling is the expressed willingness to engage in standard development. Nearly 80% of firms—56.12% “very willing” and 22.45% “relatively willing”—expressed strong interest in contributing to AI standards. This enthusiasm far exceeds the actual rate of standard creation (50%), suggesting a latent capacity that is not yet fully realized. Barriers likely include resource constraints, technical immaturity, or uncertainty about return on investment.
The disconnect between willingness and action underscores a critical policy challenge: how to convert intent into implementation. The authors suggest that targeted support—such as funding for SMEs to join standardization committees, training programs on standard development processes, and incentives for early adopters—could help bridge this gap.
From a methodological standpoint, the study’s strength lies in its grounding in real-world enterprise data. Unlike theoretical analyses or policy reviews, this work captures the lived experience of AI developers and managers. The high response rate (95.1%) and statistical rigor enhance its credibility, making it a valuable reference for policymakers, industry leaders, and academic researchers alike.
The implications extend beyond Hangzhou. As a leading city in Zhejiang Province’s “Number One Project” for digital economy development, Hangzhou serves as a bellwether for national trends. Its AI ecosystem—dense with talent, capital, and infrastructure—offers a preview of where China’s broader technological trajectory may be headed.
Moreover, the findings resonate with global debates about the governance of AI. As nations grapple with questions of algorithmic accountability, bias mitigation, and cross-border data flows, technical standards are becoming a de facto regulatory tool. By shaping these standards early, countries and companies can exert outsized influence over the future of the technology.
For China, active participation in AI standardization is part of a larger strategy to transition from a follower to a leader in high-tech innovation. Historically, Chinese firms often adopted Western-developed standards in fields like telecommunications and computing. Today, in AI, there is a concerted effort to co-create—or even lead—the rules of the game.
Yet challenges remain. The study notes that standardization in AI is still in its infancy worldwide. Unlike mature industries with well-established norms, AI lacks a unified taxonomy, benchmarking methodologies, or universally accepted testing protocols. Concepts like “intelligence,” “learning,” and “autonomy” are still being defined, making it difficult to codify them into technical specifications.
Additionally, the rapid pace of AI innovation creates a moving target for standardizers. By the time a standard is finalized, the underlying technology may have evolved significantly. This tension between stability and agility necessitates flexible, modular approaches to standard development—such as living standards that can be updated iteratively.
The role of open-source communities also cannot be overlooked. Many AI breakthroughs originate in open-source projects like TensorFlow, PyTorch, or Hugging Face, where de facto standards emerge organically through widespread adoption. Formal standardization bodies must find ways to collaborate with these decentralized ecosystems, rather than attempting to override them.
Looking ahead, the researchers call for expanded studies with larger sample sizes and deeper cross-sectional analysis. Future work could explore correlations between standardization activity and firm performance, innovation output, or export success. Longitudinal tracking would also help assess how standardization engagement evolves as companies grow.
Ultimately, the Hangzhou study paints a picture of an AI industry coming of age. Firms are no longer just building algorithms in isolation; they are thinking systemically about how their technologies fit into broader networks, markets, and regulatory environments. Standardization is no longer a back-office function but a front-line strategic capability.
As AI continues to transform industries from healthcare to finance, transportation to education, the importance of shared rules will only grow. The choices made today—by engineers in Hangzhou, policymakers in Beijing, and standards bodies in Geneva—will shape the contours of the intelligent systems that will define the next decade.
The message from Hangzhou is clear: Chinese AI enterprises are ready to standardize. The world should take note.
Lu Hao, Mao Hai-jun, Yu Su-chun, Chen Xin, Chu Fei-fei, Li Nan-yang, China Jiliang University, Hangzhou CPPCC Career Development Service Center, Hangzhou Association of Standardization, Standard Practice, DOI: 10.3969/j.issn.1002-5944.2021.09.020