CloudWalk Technology
Last reviewed
Jun 5, 2026
Sources
33 citations
Review status
Source-backed
Revision
v6 · 5,507 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 5, 2026
Sources
33 citations
Review status
Source-backed
Revision
v6 · 5,507 words
Add missing citations, update stale details, or suggest a clearer explanation.
CloudWalk Technology Co., Ltd. (Chinese: 云从科技, stock code: 688327.SS) is a Chinese artificial intelligence company headquartered in Guangzhou, Guangdong Province. Founded in April 2015 by Zhou Xi, the company specializes in facial recognition, computer vision, and human-machine collaborative intelligence. CloudWalk is one of China's "Four AI Dragons" (AI 四小龙), a group that also includes SenseTime, Megvii, and Yitu Technology. On May 27, 2022, CloudWalk became the first of the four to list on the Shanghai Stock Exchange's Science and Technology Innovation Board (STAR Market).
The company has built core technology spanning face detection, face comparison, 1:N face recognition, pedestrian re-identification, natural language processing, and speech recognition. CloudWalk provides AI solutions across finance, public security, civil aviation, smart cities, and governance. It has also developed the Congrong Large Model (从容大模型), a multimodal generative AI system that entered the competitive Chinese large language model landscape in 2023. From 2025 onward the company repositioned itself around its CloudWalk Operating System (CWOS) and AI agents, describing this combination as the core of its future business and beginning a shift away from one-off project delivery toward subscription and operations revenue.[19][24]
CloudWalk Technology was established in April 2015 in the southwestern municipality of Chongqing by Zhou Xi, Yao Zhiqiang, and Wen Hao. Unlike the other three AI Dragons, which were founded as private enterprises, CloudWalk was incubated at the Chinese Academy of Sciences (CAS). Zhou Xi had been a professor at the CAS Chongqing Institute of Green and Intelligent Technology from 2011 to 2015 before departing to form the company. This CAS origin gave CloudWalk a distinct profile among China's AI startups, providing it with access to academic research networks and government relationships that would prove useful in winning public-sector contracts.
The company quickly gained traction in the financial sector. Within its first two years, Haitong Securities, Xi'an Bank, China Construction Bank, and other financial institutions adopted CloudWalk's facial recognition systems for identity verification. By 2017, the company had begun supplying facial recognition technology to the Bank of China, one of China's largest state-owned commercial banks. CloudWalk's early focus on financial identity verification gave it a foothold in a sector where demand for biometric authentication was growing rapidly due to the expansion of mobile banking and digital finance in China.
In 2018, CloudWalk signed a strategic cooperation framework agreement with the government of Zimbabwe to provide facial recognition technology for a smart financial service network and intelligent security applications at airports, railway stations, and bus stations. The deal, backed by China's Belt and Road Initiative, would see CloudWalk help establish a national facial recognition database in Zimbabwe. The agreement drew international attention and criticism from digital rights advocates concerned about privacy and civil liberties implications, particularly in a country with a poor human rights record.
CloudWalk generated revenue of approximately CNY 807 million in 2019 and CNY 755 million in 2020. Revenue climbed to roughly CNY 1.08 billion in 2021, the company's highest annual revenue figure in its history at that point. Despite growing revenues, the company accumulated significant net losses during this period, with aggregate net losses of approximately CNY 2.58 billion from 2018 to 2020. Research and development costs consumed a large share of total expenditures, accounting for 56.25% in 2019 and rising to 76.59% in 2020. This heavy R&D spending reflected the capital-intensive nature of building competitive AI systems, including investment in deep learning infrastructure, training data acquisition, and algorithm development.
In December 2020, CloudWalk filed a prospectus with the Shanghai Stock Exchange, seeking to raise CNY 3.75 billion (approximately US$573 million) through an initial public offering on the STAR Market. The IPO process extended through 2021 as the company navigated regulatory review. During this period, CloudWalk was described by Caixin Global as a "Chinese state-backed AI unicorn," reflecting the significant role government investment played in the company's development.
CloudWalk also received a CNY 2 billion (approximately US$301 million) investment from the Guangzhou municipal government to set up an AI image center in the city, and the company relocated its headquarters from Chongqing to Guangzhou. This move to Guangzhou, one of China's major technology and manufacturing hubs in the Pearl River Delta, positioned CloudWalk closer to hardware supply chains and a broader talent pool.
On May 27, 2022, CloudWalk Technology officially listed on the Shanghai Stock Exchange under the stock code 688327. The opening price was CNY 24 per share, representing a 56.1% increase over the issue price of CNY 15.37. By the close of the first trading day, the stock settled at CNY 21.40, up 39.2%, giving the company a total market capitalization of approximately CNY 15.85 billion.
However, CloudWalk raised only CNY 1.73 billion from the listing, significantly less than the planned CNY 3.75 billion. The shortfall reflected cautious investor sentiment toward AI companies that had not yet achieved profitability. Still, the IPO marked an important milestone as CloudWalk became the first of the Four AI Dragons to enter China's domestic A-share stock market, beating SenseTime (which had listed in Hong Kong in December 2021) to the mainland exchange.
In subsequent years, the company's revenue fluctuated considerably. Revenue for 2022 was approximately CNY 526 million, a sharp decline from 2021, attributed in part to macroeconomic headwinds and the impact of COVID-related disruptions on project deployment. Revenue recovered to approximately CNY 628 million in 2023, before declining again to approximately CNY 398 million in 2024 (a 36.6% year-over-year drop). The company continued to report net losses across all of these years, though losses narrowed from CNY 869 million in 2022 to CNY 630 million in 2023. The 2024 net loss attributable to shareholders was approximately CNY 696 million according to the company's audited annual report.[20]
By the third quarter of 2025, CloudWalk reported revenue of CNY 355 million, representing a year-on-year increase of 56.81%, with third-quarter revenue alone reaching CNY 186 million (a 75.68% year-on-year increase). These figures suggested a recovery trajectory, driven in part by growing demand for the company's large model solutions and AI agent products.
CloudWalk's results in 2025 and early 2026 showed faster revenue growth from its large model and agent business set against continued, though narrowing, losses and a sharply downsized organization.
For the first half of 2025, the company reported revenue of approximately CNY 169 million, up 40.21% year on year, with a net loss of about CNY 230 million, narrower than the CNY 356 million loss a year earlier. CloudWalk attributed the revenue growth to the expansion of its AI solutions business and the loss reduction to tighter cost control, noting that period expenses fell 33.83% year on year. The company also disclosed that its lower-margin "pan-AI" (泛AI) segment had grown to 55.78% of revenue, which pulled the half-year gross margin down to about 37.7%.[21]
In the first three quarters of 2025, revenue reached CNY 355 million (up 56.81%) and the net loss attributable to shareholders was about CNY 310 million, with third-quarter net loss narrowing 47.64% year on year. Management said operating cash flow for the nine months turned positive at roughly CNY 313 million, which the company described as its first positive operating cash flow since listing.[22][24]
For the full year 2025, CloudWalk reported operating revenue of about CNY 500.6 million (CNY 5.01 billion measured in hundred-millions, i.e. roughly CNY 501 million), an increase of 25.88% year on year, and a net loss attributable to shareholders of about CNY 555.6 million, narrowing 20.14% from 2024. Basic loss per share was CNY 0.54, and the gross margin was 28.62%, down 7.19 percentage points, which the company linked to the rising share of lower-margin pan-AI work. The audited annual report was published in April 2026.[23][25]
The 2025 results capped a multi-year contraction in headcount and research spending that Chinese financial media characterized as "survival through downsizing." Press reports based on company filings noted that R&D investment for the first three quarters of 2025 fell about 43.85% year on year to roughly CNY 162 million, that the research staff had been cut from a peak of around 997 in 2020 to about 228 by the end of 2024, and that total headcount had fallen from about 1,799 in 2020 to about 453 by the end of 2024. Cumulative net losses since 2017 exceeded CNY 4.5 billion, consuming most of the proceeds raised at the 2022 IPO.[26]
For the first quarter of 2026, CloudWalk reported revenue of about CNY 23.23 million, down 37.61% year on year, which the company attributed to the timing of project delivery and acceptance rather than weaker demand. The net loss attributable to shareholders narrowed sharply to about CNY 13 million, an improvement of roughly 89.5% from a year earlier, although the company noted that the result still relied on non-recurring items.[27]
In April 2026, alongside its annual report, CloudWalk announced a plan to raise up to CNY 300 million through a simplified-procedure private placement of shares to specified investors, capped at 20% of the prior year-end net assets, with proceeds earmarked for its main business and working capital.[28] On the same occasion the company said its AI agent business had reached commercial deployment across several industries and was making a meaningful contribution to results, with product lines spanning manufacturing, energy, finance, and governance.[29] By 2026 the company's shares were increasingly grouped by Chinese market commentators under the "physical AI" and embodied-intelligence themes, reflecting its agent-and-robotics positioning.[30]
Zhou Xi (周曦) is the founder, chairman, and president of CloudWalk Technology. Originally from Sichuan Province in southwestern China, Zhou graduated from the University of Science and Technology of China (USTC) with a degree in artificial intelligence, specializing in pattern recognition. He went on to earn a doctoral degree from the University of Illinois at Urbana-Champaign (UIUC), where he focused on image processing and computer vision. He completed a postdoctoral fellowship at UIUC from October 2010 to November 2011.
During his academic career, Zhou built an impressive record in competitive AI research. Between 2007 and 2011, he led university research teams to victory six times in international pattern recognition competitions, establishing himself as one of the leading young researchers in the field. He also gained industry experience through internships at NEC, Microsoft, and IBM between 2007 and 2009, working on speech recognition and computer vision projects at these companies.
Zhou's early research at USTC began with voice recognition before shifting to visual pattern recognition. He spent time at the Microsoft Research Asia Institute working on speech recognition before moving to UIUC for his doctoral studies. During his time at UIUC, Zhou developed a cluster server for image processing that was considered ahead of its peers in terms of computational efficiency.
In 2011, Zhou was selected for the "CAS Hundred Talents Program," a prestigious recruitment initiative that brings top researchers back to China. He joined the Chongqing Institute of the Chinese Academy of Sciences, where he spent four years developing research infrastructure and AI systems. By 2015, he left academia to found CloudWalk with co-founders Yao Zhiqiang and Wen Hao. The company maintained ongoing research collaboration with labs at UIUC and in Silicon Valley even after its founding.
CloudWalk's original face recognition algorithm holds industry-leading capabilities in face detection, face comparison, 1:N face recognition, and face attribute analysis. The company's 3D structured light face recognition technology achieves an accuracy rate above 99% at an error rate of one in one million, making it suitable for high-security applications in banking and law enforcement. CloudWalk's Re-ID (pedestrian re-identification) technology broke three world records in academic benchmarks, advancing the field from simple "face scanning" to full "human recognition" that identifies individuals based on body shape, gait, clothing patterns, and other features beyond the face alone.
Beyond visual recognition, CloudWalk has built capabilities in natural language processing, voice technology, and video structuring. The company's video structuring technology can analyze surveillance footage in real time, extracting structured data about people, vehicles, and events from unstructured video streams.
At the center of CloudWalk's product strategy is the CloudWalk Operating System (CWOS), described as a "human-machine collaborative operating system" and positioned as an "operating system for the intelligent era." CWOS integrates large models across vision, speech, and NLP domains. It serves as a central management platform for data, models, and knowledge, enabling cross-scene, cross-industry, and personalized AI services.
CWOS is designed to function as underlying infrastructure software, bridging the digital and physical worlds. The system covers a closed loop of "perception, cognition, and decision-making" capabilities, bringing together facial recognition, object detection, voice technology, video structuring, and analytics into a unified platform. CWOS also serves as the operational base for all types of "digital humans" that CloudWalk develops for customer-facing applications.
The main products built on top of CWOS include AI agents, industry-specific large models, and foundational large models (hybrid models). These products are deployed through a combination of cloud services, edge computing devices, and on-premise installations depending on the client's security and latency requirements. From 2025 the company began describing CWOS as the basis for moving customers from one-off hardware purchases toward continuous subscription and operations of agent modules, a model it argued would improve margins and cash flow quality.[24]
On May 18, 2023, CloudWalk officially launched the Congrong Large Model (从容大模型), its self-developed generative AI system. The Congrong model is a language model-based AI system that can learn and analyze text and voice data and output relevant answers. It supports graphic understanding, copywriting, logical reasoning, multi-round dialogue, and programming. The model targets industry-specific applications in finance, government affairs, manufacturing, healthcare, elderly care, and gaming.
In October 2023, CloudWalk partnered with Huawei and Tianjin Port Group to launch the PortGPT large model for port operations, with planned applications in port production, logistics, and customer management. CloudWalk and Huawei also jointly announced a Congrong large model training-and-inference integrated solution that allows enterprises to train, build, and manage their own large models on Huawei's computing infrastructure.
The model achieved notable benchmarks. CloudWalk's foundational large model for pedestrian analysis set new world records on multiple datasets, reaching a fine-tuning accuracy of 92.89% on the PA100K dataset using over 2 billion data points, including unlabeled datasets and multimodal data. In multimodal evaluation, the updated version of the Congrong model (CongRan V2.0) scored 80.7 on the OpenCompass multimodal leaderboard, ranking first globally and surpassing models from Google, OpenAI, and other leading teams. An earlier version had scored 65.5 on OpenCompass, surpassing Gemini-1.5-Pro and GPT-4V and ranking third globally behind GPT-4o and Claude 3.5 Sonnet.
The Congrong model has been deployed through AI agents in multiple sectors, including financial risk control, e-commerce customer service platforms, and government administrative services.
The 80.7 OpenCompass result was achieved by the CongRong-v2.0 multimodal model, which CloudWalk said topped the global OpenCompass multimodal leaderboard around late May 2025, ranking first in China and first globally on that board. The company reported the model led on component tests including the MMBench Chinese evaluation, the MMMU subject benchmark, and the MMStar reasoning leaderboard, and said its composite score was several points ahead of leading international models. CloudWalk credited a training approach that combined direct preference optimization (DPO) and group relative policy optimization (GRPO) alignment techniques.[31]
On May 27, 2025, at the Kunpeng Ascend Developer Conference 2025, CloudWalk released the Congrong Large Model "intelligent-use" all-in-one appliance (从容大模型智用一体机). The product is built on Huawei's Ascend AI hardware and software stack and integrates CloudWalk's Congrong multimodal model together with the DeepSeek language model and an associated toolchain, packaging model training and inference into a single on-premise system aimed at government, finance, education, and healthcare customers. CloudWalk described the appliance as combining a multimodal large model with AI agents and shipping with more than 200 preset industry models so that large and small models can be used together.[32] In December 2025, CloudWalk received Huawei Ascend "Preferred Level" partner certification, and over the course of 2025 the company said the appliance had been deployed at scale with partners including the national cybersecurity base, Tianjin Port Group, China Telecom, and State Grid Shandong.[19][32]
In January 2026, CloudWalk launched a trading-focused AI agent, extending the agent product line that the company by then described as commercially deployed across manufacturing, energy, finance, and governance.[19][29]
CloudWalk is the largest AI supplier in China's banking and financial industry, with a reported market share of 82.8% in AI-powered facial recognition for the sector. Its technology has been deployed at major financial institutions including the Bank of China, Chongqing Bank, China Construction Bank, and Haitong Securities.
In banking scenarios, CloudWalk's facial recognition is used for identity verification at counters, ATMs, and direct banking platforms. The system enables real-time comparison between a customer's live face and their registered biometric data, reducing the risk of identity fraud. In consumer-facing settings such as supermarkets, smart bank branches, healthcare facilities, scenic areas, and rail transit stations, CloudWalk's face-payment system allows users to complete transactions without carrying bank cards or mobile phones. Chongqing Bank was an early adopter, applying CloudWalk's cloud-based facial recognition across counter services, ATM operations, and direct banking for identity verification.
CloudWalk has established a joint laboratory with the Civil Aviation Management Institute of China. The company's technology covers 9 of China's 10 major airport hubs, serving approximately 1 million passengers daily. Its software and hardware products are certified by China TravelSky Holding Company Limited, the dominant IT service provider for China's aviation industry.
CloudWalk's smart airport solutions include face registration, self-service seat selection, face-scan payment, and security check acceleration using integrated ID card readers, barcode readers, passport readers, and face recognition gate machines. These systems allow passengers to pass through security checkpoints and boarding gates without presenting physical documents after an initial registration step. The company has partnerships with China Eastern Airlines, Shanghai Airport Authority, Baiyun Airport, Shenzhen International Airport, Wuhan Tianhe International Airport, and Capital Airport Holding Company.
CloudWalk provides AI solutions for urban governance and smart city initiatives. Using CWOS, the company has deployed "Park City" digital city solutions, including in the Tianfu New Area of Chengdu. These solutions integrate video surveillance, pedestrian and vehicle analysis, and intelligent decision-making systems across urban infrastructure, providing city administrators with real-time situational awareness and automated incident detection.
The company's products for smart governance include facial recognition terminals, facial scanning door entry systems, infrared binocular scanning machines, and video structuring platforms. CloudWalk's smart governance solutions have been adopted by municipal governments across China for public security applications, administrative service centers, and community management systems.
In June 2025, CloudWalk signed an agreement with the Wuhan Linkonggang (Lingang) Economic and Technological Development Zone for a large model training-and-inference project at the country's national cybersecurity base in Wuhan. The company said the project, pursued in cooperation with Huawei, was intended to build a leading AI large model training-and-inference center for central China and an AI "security base" serving content moderation, government approval, and financial risk-control scenarios. CloudWalk also stated that it planned to incubate at least ten industry-specific large model and agent solutions with local partners over the following one to two years.[33]
CloudWalk is one of China's "Four AI Dragons" (AI 四小龙), a collective term coined in the Chinese technology press for four AI companies that rose to prominence in the mid-2010s on the strength of their computer vision and facial recognition capabilities. The four companies became symbols of China's ambition to lead the world in artificial intelligence, attracting billions of dollars in venture capital and government funding during the AI investment boom of 2016 to 2019.
| Company | Founded | Headquarters | IPO Status | Notable Focus |
|---|---|---|---|---|
| SenseTime | 2014 | Hong Kong | Listed on HKEX (December 2021) | Largest of the four; broad AI platform |
| Megvii | 2011 | Beijing | Listed on SHSE (March 2024) | Creator of Face++; open-source computer vision |
| Yitu Technology | 2012 | Shanghai | Withdrew IPO application (2021) | Smart cities, healthcare AI |
| CloudWalk Technology | 2015 | Guangzhou | Listed on SHSE STAR Market (May 2022) | CAS-incubated; finance and aviation AI |
All four companies have been placed on U.S. sanctions lists in various forms due to their alleged involvement in surveillance activities in Xinjiang. They shared a common origin story in China's push to lead globally in AI, particularly in facial recognition, and all four have struggled to achieve consistent profitability. Collectively, they accumulated billions of yuan in net losses through the early 2020s: SenseTime lost approximately CNY 24.3 billion over three and a half years, Megvii lost around CNY 13 billion, Yitu lost roughly CNY 7.2 billion, and CloudWalk lost approximately CNY 2.3 billion.
CloudWalk is the smallest of the four by revenue and headcount, but it has a unique position as the only one that was incubated from a government research institution (the Chinese Academy of Sciences) rather than being started by private entrepreneurs. This distinction has given it particularly strong government and state-owned enterprise relationships.
As of the mid-2020s, the landscape has shifted. SenseTime and CloudWalk have pivoted toward becoming large model companies, investing in GPU clusters and generative AI capabilities. Yitu and Megvii have taken more conservative strategic positions, focusing on specific verticals. All four face growing competition from a new generation of AI companies, sometimes referred to as the "Six Little Tigers," which focus on large language models and generative AI. By 2025 and 2026 CloudWalk had recentered its strategy on the Congrong large model, agent products, and the CWOS platform, while continuing to carry the legacy facial-recognition business that originally defined the group.[24][26]
CloudWalk Technology has been subject to two separate rounds of U.S. government sanctions, each administered by a different federal agency.
On May 22, 2020, the U.S. Department of Commerce's Bureau of Industry and Security (BIS) announced the addition of CloudWalk Technology to the Entity List. The designation, which took effect on June 5, 2020, was part of a broader action targeting nine Chinese entities for their alleged roles in enabling China's "campaign of repression, mass arbitrary detention, forced labor and high-technology surveillance" against Uyghurs, ethnic Kazakhs, and other Muslim minority groups in the Xinjiang Uyghur Autonomous Region (XUAR).
The Entity List designation restricts CloudWalk's ability to obtain U.S.-origin technology, components, and software without a special license from the Bureau of Industry and Security. In practice, this means American companies cannot sell semiconductor chips, software tools, or other technology products to CloudWalk without explicit government approval, which is generally denied for entities on the list. The restriction on U.S.-origin computing hardware is one reason CloudWalk has built its 2025 large model appliances on Huawei's domestic Ascend platform rather than on imported accelerators.[32]
On December 16, 2021, the U.S. Department of the Treasury's Office of Foreign Assets Control (OFAC) added CloudWalk Technology to the Non-SDN Chinese Military-Industrial Complex Companies (NS-CMIC) List. This action, imposed pursuant to Executive Order 13959 as amended by E.O. 14032, targeted eight Chinese companies for "actively supporting the biometric surveillance and tracking of ethnic and religious minorities" in China, particularly the Uyghur minority in Xinjiang.
The NS-CMIC sanctions, which became effective on February 14, 2022, prohibit U.S. persons from purchasing or selling publicly traded securities or derivatives of the designated entities. This effectively bars American investors from holding shares in CloudWalk Technology. The timing of this sanction, just months before CloudWalk's Shanghai IPO in May 2022, added regulatory complexity to the company's public listing, though the STAR Market listing was targeted at domestic Chinese investors rather than international ones.
Before its public listing, CloudWalk completed multiple funding rounds totaling hundreds of millions of dollars. The company's investor base included a mix of state-owned financial institutions, government-backed funds, and private venture capital firms. Notable investors included the Industrial and Commercial Bank of China (ICBC), Shanghai Guosheng Group, Shunwei Capital, and Puhua Capital.
In November 2017, Shunwei Capital, Oriza Holdings, and Puhua Capital led a CNY 500 million (approximately US$75 million) Series B round. Additional participants included Yuexiu Financial Holdings Group, Shenzhen Qianhai Xingwang Investment, and a fund co-invested by Galaxy Holding and Zhangjiang Group.
The Guangzhou municipal government invested CNY 2 billion (approximately US$301 million) in the company to establish an AI image center in the city. This was one of the largest single government investments in any of the Four AI Dragons and reflected the Guangzhou government's ambition to build the city into a national AI hub.
In April 2026, CloudWalk announced a post-IPO equity financing of up to CNY 300 million through a simplified-procedure private placement to specified investors, with the amount capped at 20% of the prior year-end net assets. The company said the proceeds would support main-business projects and working capital.[28]
The table below summarizes CloudWalk Technology's annual revenue and net income/loss figures (in CNY) based on public filings and reports:
| Year | Revenue (CNY) | Net Income/Loss (CNY) | Notes |
|---|---|---|---|
| 2019 | ~807 million | ~-1.76 billion | R&D costs = 56.25% of spending |
| 2020 | ~755 million | ~-720 million | R&D costs = 76.59% of spending |
| 2021 | ~1.08 billion | Loss (not disclosed individually) | Peak pre-IPO revenue |
| 2022 | ~526 million | ~-869 million | First full year as listed company; revenue decline |
| 2023 | ~628 million | ~-630 million | Losses narrowed year-over-year |
| 2024 | ~398 million | ~-696 million | Revenue declined 36.6% YoY; audited net loss attributable to shareholders [20] |
| 2025 | ~501 million | ~-556 million | Revenue +25.88% YoY; loss narrowed 20.14%; basic LPS CNY 0.54; gross margin 28.62% [23][25] |
For interim 2025 periods, CloudWalk reported first-half revenue of about CNY 169 million (up 40.21%) with a net loss of about CNY 230 million, and nine-month revenue of about CNY 355 million (up 56.81%) with a net loss attributable to shareholders of about CNY 310 million. The company said nine-month operating cash flow turned positive at roughly CNY 313 million, its first positive operating cash flow since listing.[21][22][24] In the first quarter of 2026, revenue fell to about CNY 23.23 million (down 37.61% on project-timing effects) while the net loss narrowed about 89.5% year on year to roughly CNY 13 million.[27]
The company's persistent losses reflect the high cost of AI research and development, the challenge of scaling enterprise AI deployments in a competitive market, and the impact of U.S. sanctions on access to advanced computing hardware. Chinese financial media reported that CloudWalk cut R&D spending for the first three quarters of 2025 by about 43.85% year on year to roughly CNY 162 million, reduced its research staff from a peak of about 997 in 2020 to around 228 by the end of 2024, and shrank total headcount from about 1,799 in 2020 to about 453 by the end of 2024, with cumulative net losses since 2017 exceeding CNY 4.5 billion.[26]
CloudWalk Technology has received several notable recognitions over the course of its history:
CloudWalk Technology has been accused by the U.S. government and human rights organizations of providing facial recognition technology that enables mass surveillance of the Uyghur population in the Xinjiang Uyghur Autonomous Region. These allegations led directly to the company's placement on the U.S. Commerce Department Entity List in May 2020 and the Treasury Department's NS-CMIC sanctions in December 2021. The U.S. Treasury Department specifically cited CloudWalk for "actively supporting the biometric surveillance and tracking of ethnic and religious minorities" in its sanctions announcement.
The 2018 agreement between CloudWalk and the Zimbabwean government drew criticism from digital rights advocates and human rights groups. Under the agreement, CloudWalk would help establish a national facial recognition database in Zimbabwe, with the technology deployed at airports, railway stations, and bus stations. Critics raised concerns about the potential for unwarranted surveillance and the collection of citizens' biometric data without adequate safeguards, especially given Zimbabwe's human rights record.
Chinese state media reports indicated that CloudWalk planned to use the collected photographs to improve its facial recognition systems' performance on darker skin tones. Training data diversity has been an acknowledged challenge in facial recognition technology globally, but the approach of collecting biometric data from a foreign population without clear informed consent drew scrutiny regarding data ethics.