NetEase, Inc.

NetEase, Inc., one of China's largest internet technology companies, has assembled a broad and technically ambitious portfolio of digital human capabilities that spans gaming, education, enterprise communications, metaverse platforms, and entertainment. What began as research into intelligent game characters at the company's dedicated artificial intelligence laboratory has grown into a sprawling ecosystem of virtual human products, platforms, and services deployed across dozens of industries and involving partnerships with global technology leaders including NVIDIA and Xiaoice. The following report examines every known NetEase-affiliated entity and product contributing to this ecosystem, covering their leadership, physical locations, technologies, deployments, and commercial relationships.

NetEase was founded in June 1997 by William Ding Lei, who continues to serve as chief executive officer and director of the company. Ding Lei, born in 1971 in Ningbo, Zhejiang Province, holds a bachelor's degree from the Chengdu Institute of Radio Engineering, now known as the University of Electronic Science and Technology of China, and directly controls approximately 45.78 percent of the company's shares. The company's principal headquarters is located at the NetEase Building, No. 599 Wangshang Road, Binjiang District, Hangzhou, Zhejiang Province, 310052. Its Beijing office occupies Building No. 7, West Zone, Zhongguancun Software Park Phase II, No. 10 Xibeiwang East Road, Haidian District. Additional major offices are situated in Guangzhou and Shanghai, and international operations extend to Tokyo, Seoul, Montreal, and the San Francisco Bay Area. NetEase is listed on NASDAQ under the ticker NTES and on the Hong Kong Stock Exchange as 9999, and it employs more than 32,000 people globally. Charles Zhaoxuan Yang has served as chief financial officer since June 2017.

The corporate structure that houses digital human work is organized around four major reporting segments: Online Game Services, Youdao (the education subsidiary, separately listed on the New York Stock Exchange under the ticker DAO), Cloud Music (listed on the Hong Kong Stock Exchange), and Innovative Businesses and Others. Within and across these segments, several entities drive the digital human agenda. The most important are the Fuxi AI Lab, which conducts foundational research; Yaotai, the immersive metaverse platform; the enterprise services division formerly known as Zhiqi and now rebranded as Shuzhi, which includes the Yunxin communications platform; and Youdao, which deploys digital humans in educational products. The Leihuo, or ThunderFire, business group serves as the organizational home for the Fuxi Lab and for several of the game titles that showcase digital human technology in production environments. NetEase's overarching approach treats digital humans not as a single product line but as a cross-cutting capability woven into gaming, communications infrastructure, education, enterprise services, and cultural production.

NetEase Fuxi, formally known as the NetEase Fuxi AI Lab, was established in September 2017 as the first professional game-focused artificial intelligence research laboratory in China. It is headquartered in Hangzhou alongside the parent company, within the Binjiang District campus, and maintains a presence in Beijing. The laboratory's chief advisor is Harry Shum, a foreign member of the United States National Academy of Engineering and formerly an executive vice president at Microsoft, who also serves as chairman of Xiaoice and council chairman of the Hong Kong University of Science and Technology. Li Renjie, a graduate of the University of Science and Technology of China's elite Young Class program who holds a doctorate from the University of Rochester, has served as head of the laboratory. Li Renjie was previously employed at Blizzard Entertainment, where he worked on World of Warcraft, and at Riot Games, where he built the data and AI team for League of Legends, making him reportedly the first Chinese professional to contribute to both titles.

The laboratory is organized into two main divisions and a commercial arm. The research division pursues frontier work in reinforcement learning, computer vision and graphics, natural language processing, speech synthesis and music generation, user profiling, and virtual human technologies. The commercial arm, Fuxi Technology, handles external sales of game AI, virtual human solutions, and the Yaotai immersive platform. A separate robotics unit, now branded as NetEase Lingdong, develops autonomous construction machinery and has deployed solutions across more than fifty real-world projects in over ten Chinese provinces, led by Yang Xinwei.

Key researchers within the virtual human group include Yi Yuan, director of the Visual Computing Group, who holds a doctorate from Wuhan University and whose research spans three-dimensional face, hair, and body reconstruction, expression and motion generation, and differentiable rendering. Yi Yuan's Character Auto-creation system, which generates three-dimensional game avatars from a single selfie photograph, was used more than one million times by Chinese gamers within its first years and won NetEase's internal Best Technology Innovation Award in 2020 and the Thunder Fire Innovation Award in 2021. Yu Ding led the Virtual Human Group from 2018 until 2023, when he departed to become director of AI research and development at Happy Elements; he holds a doctorate from Télécom ParisTech and completed postdoctoral work at the University of Houston and CNRS. Li Lincheng presented work on neural implicit surface modeling for the Yaotai platform at NVIDIA's GTC conference in 2023. Tao Jianrong, who holds a doctorate in computer science from Zhejiang University, co-founded the laboratory with Li Renjie in 2017 and led game AI commercialization, authoring over fifty top-tier AI papers and forty patents and speaking twice at the Game Developers Conference, before transitioning to NetEase Zhiqi's game division. Changjie Fan served as a principal researcher from 2017 onward, and Runze Wu leads the user profiling group, having earned a doctorate from USTC in 2018 and received best paper awards at UbiComp 2016, CoG 2020, and SMDS 2020. Lü Tangjie serves as head of the AI Algorithm Department and represented Fuxi at the Arm Unlocked 2025 conference. Hu Jingzhen heads large language model products, and Wu Yunong leads large language model and embodied intelligence products.

The laboratory's virtual human research has produced a series of published systems and competition victories. The Fuxi Virtual Human Group won first place in the Expression Classification challenge and the Facial Action Unit Detection challenge at CVPR 2022, CVPR 2023, and ICCV 2021, and secured second place in the Valence-Arousal Estimation challenge at CVPR 2023 and ICCV 2021. Published work includes StyleTalk, a system for one-shot talking head generation with controllable speaking styles presented as an oral paper at AAAI 2023; DINet, a deformation inpainting network for realistic face dubbing on high-resolution video, also at AAAI 2023; FlowFace, a semantic flow-guided face swapping technique; and Emotional Voice Puppetry, published in IEEE Transactions on Visualization and Computer Graphics in 2023. The group maintains three GitHub organizations, FuxiVirtualHuman, fuxiAIlab, and NetEase-FuXi, collectively hosting dozens of open-source repositories. Overall, the laboratory has produced more than two hundred papers at leading venues including CVPR, ICCV, ECCV, AAAI, SIGGRAPH, TPAMI, KDD, WWW, IJCAI, and ACM Multimedia, and has filed more than six hundred patent applications with over two hundred granted. In December 2024, the laboratory publicly released an embodied intelligence solution for humanoid robots at the Second Annual Humanoid Robot Scenarios Conference in Suzhou.

Fuxi independently developed the Yuyan family of large language models for Chinese text generation. Yuyan is a decoder-only architecture in the GPT-2 lineage, trained on a large, high-quality Chinese novel dataset. The model family includes yuyan-10b, a natural language understanding model suitable for fine-tuning on benchmark tasks; yuyan-11b, a generative causal language model; and yuyan-dialogue, a multi-turn dialogue model fine-tuned from the 11-billion-parameter base. These models are publicly available on Hugging Face under the FUXI organization. Alongside Yuyan, Fuxi released the Danqing multimodal text-to-image model, trained entirely on domestically sourced Chinese cultural data and capable of generating imagery in response to classical poetic descriptions, and the Yuzhi knowledge model. The laboratory also developed EET, the Easy and Efficient Transformer inference acceleration framework, which was published at NAACL 2022's industry track and open-sourced on GitHub. The Danqing model has been integrated into the Yaotai platform for AI-generated clothing on avatars, and Yuyan powers the intelligent NPCs and text-to-face features in the mobile game Ni Shui Han, where players can type poetic descriptions and receive a matching three-dimensional character face within three seconds.

The Fuxi Lab additionally contributes anti-cheating systems and AI gaming bots. The NGUARD+ framework, published in ACM Transactions on Knowledge Discovery from Data, uses attention-based models to detect bot behavior in massively multiplayer online games, reportedly avoiding fifty percent of unnecessary economic losses and improving operations team efficiency by eighty percent. Reinforcement learning-trained competitive bots have been deployed in Ni Shui Han, Qian Nü You Hun, Tian Yu, Naraka Bladepoint, and several other titles, where they develop offensive, defensive, and harassment tactical styles through self-play.

The laboratory's official website is accessible at fuxi.163.com, the Virtual Human Group's research page is at fuxivirtualhuman.github.io, and the Youling AI open platform is at yl.163.com.

NetEase Yaotai is an immersive virtual event platform developed by Fuxi Lab and launched in October 2020. Its origins trace to the second Distributed AI International Conference, DAI 2020, which was held inside NetEase's Ni Shui Han game world due to the COVID-19 pandemic, drawing more than three hundred global AI scholars and prompting the team to create a dedicated platform for virtual gatherings. Yaotai is based in Hangzhou, co-located with Fuxi and the NetEase headquarters, and its official website is yaotai.163.com.

The platform is led by Liu Bai, who previously oversaw NetEase's big data product efforts within ThunderFire. Liu Bai has described Yaotai as an open system progressing from "meta rooms" and "meta cities" toward an interconnected virtual universe, and has represented the platform in interviews with the South China Morning Post, Yicai, and 36Kr, and at the Zhongguancun Forum. Hu Zhipeng, senior vice president of NetEase and head of the Leihuo business group, oversees the broader metaverse and technology strategy and participated in a public dialogue with metaverse analyst Matthew Ball on the Yaotai platform in September 2022. Ruan Liang, vice president of NetEase in Hangzhou and general manager of the enterprise services division, has characterized Yaotai as a gateway toward a virtual world comparable to the Oasis from the novel Ready Player One.

Yaotai's technical foundation relies on NetEase's proprietary game engines, Messiah and NeoX, and an ACE distributed engine capable of supporting more than ten thousand simultaneous on-screen real-time interactions and one hundred thousand virtual characters simultaneously online through a distributed service framework and area-of-interest management. Rendering is handled in the cloud so that users can access the platform through a web browser without downloading software, following a cloud gaming approach, and the platform also offers a dedicated PC client, a mobile application launched in mid-2022, and emerging virtual reality support. The avatar system permits over two hundred dimensions of face customization and supports one-photo AI avatar generation powered by Fuxi Lab's face-creation technology. AI-generated clothing is created by the Danqing text-to-image model from text prompts, and neural implicit surface modeling accelerated by NVIDIA CUDA reduces three-dimensional scene reconstruction from phone-captured video to as little as fifteen minutes, a twenty-fold speed improvement. The platform supports walk-around three-dimensional environments, embedded screen sharing and presentation playback, real-time translation via Youdao, video chat, barrage commentary, and full event management including registration, venue deployment, meeting controls, identity-based permissions, and data analytics with visitor heatmaps.

Yaotai's public deployments constitute some of the most visible demonstrations of NetEase's digital human technologies. In December 2021, the platform hosted the Hong Kong IPO ceremony for NetEase Cloud Music, in what was described as the world's first metaverse IPO event, during which CEO Ding Lei appeared simultaneously as two virtual avatars representing himself at ages twenty-nine and fifty, both banging a virtual gong. That same month, a virtual investor communication meeting drew more than two hundred investors from the United States, United Kingdom, Canada, France, and South Africa. In March 2022, the Henan Smart Tourism Conference became the platform's first metaverse tourism application, and in August 2023, the Yuan Yu Zhou cultural tourism metaverse was formally launched at the Henan Smart Cultural Tourism Conference as China's first ultra-realistic digital cultural tourism space, featuring ten major Henan tourist destinations including Laojun Mountain, Shaolin Temple, Longmen Grottoes, and Yingtian Academy. The platform built the first metaverse venue for the 2023 CCTV Online Spring Festival Gala, constructing a thirty-million-square-foot cyber-Chinese-style virtual space in thirty days, and it created the permanent metaverse venue for the 2023 Zhongguancun Forum, China's national-level science and technology innovation platform, designed as an outer-space permanent base with a main Star Hall. Additional deployments include the 2022 China International Big Data Industry Expo, brand campaigns for Nayuki Tea, Sanofi, Audi, and Alipay, the Asian Games metaverse fireworks event for the 2023 Hangzhou Asian Games under the direction of the Zhejiang Provincial Culture and Tourism Department, and TEDxBund Women's Conference.

By mid-2023, Yaotai had hosted hundreds of large-scale events with millions of participants across more than thirty industries. The platform was selected for the Ministry of Industry and Information Technology's New Information Consumption Demonstration Project in 2022 and won the Cailian Press Metaverse Industry Application and Pioneer Technology Gold Award. NetEase has filed more than one hundred metaverse-related patents, ranking fourth in China behind Baidu, Tencent, and SenseTime. The platform is commercially operational with paid services for online exhibition booth decoration, enterprise metaverse spaces, and digital tourism partnerships, and its development continues through version upgrades including the Yaotai 3.0 avatar system.

NetEase Zhiqi was established in 2015 as a one-stop enterprise services platform consolidating internal technology capabilities across communications, content security, and business growth tools. It was led by Ruan Liang, a Zhejiang University computer science graduate who joined NetEase in 2006 and previously served as general manager of Yixin, a messaging product jointly developed with China Telecom. In March 2024, NetEase announced the merger of Zhiqi with the Shufan data platform division into a new unified business-to-business brand called NetEase Shuzhi, expanding the portfolio to five core domains: digital content risk control, converged communications and cloud-native platform-as-a-service, service-marketing integration, full-chain data development and governance, and intelligent application development through the CodeWave platform. Ruan Liang serves as vice president of NetEase and general manager of the combined Shuzhi entity, and also holds the title of deputy secretary of NetEase Group's CPC Committee. The enterprise platform is headquartered in Hangzhou with operations across Beijing, Guangzhou, and Shanghai, and it has partnered with Guanshanhu District in Guiyang to build a security and compliance innovation center. The platform serves more than one million enterprise customers.

Within the Shuzhi structure, three core business units handle specific functions. NetEase Yunxin provides converged communications as a platform service. NetEase Yidun handles digital content risk control including content security, business security, and application security. NetEase Yunshang offers business growth and service-marketing integration, including the Qiyu intelligent customer service system. The official portals are grow.163.com for the enterprise platform and yunxin.163.com, which redirects to netease.im, for the communications service.

NetEase Yunxin is the primary vehicle through which the enterprise division delivers digital human functionality to developers and business clients. The platform provides instant messaging, signaling, SMS, audio and video calling, interactive live streaming, video on demand, interactive whiteboard, and enterprise video conferencing. Its global network, branded WE-CAN for Communications Acceleration Network, comprises more than two thousand acceleration nodes delivering sub-one-hundred-millisecond latency to hundreds of countries, a ninety-nine-point-nine percent stutter-free call rate, and support for more than ten million concurrent connections. By late 2021, Yunxin had served over 1.2 million enterprise developers, registered more than 480,000 applications, processed over 18.6 billion smart terminal SDK installations, and handled 1.6 trillion messages reaching more than one billion users. The platform has appeared in multiple Gartner CPaaS market reports and offers SDKs across Android, iOS, macOS, Windows, web, Flutter, WeChat Mini Programs, Node.js, Electron, and HarmonyOS.

Yunxin's digital human integration operates through both its instant messaging and real-time communication SDKs. Within instant messaging, the platform supports three types of AI digital humans: an AI chat digital human for conversational companionship, an AI translation digital human for real-time multilingual translation, and an AI text-search digital human for selection-based search assistance. These digital humans function as special user accounts configured through the Yunxin developer console and can participate in one-on-one conversations, be invoked through at-mentions during human-to-human chats, or join group conversations. The SDK exposes dedicated API classes including V2NIMAIService with methods for retrieving AI user lists, proxying model calls, and managing AI event listeners. Documentation is available at doc.yunxin.163.com.

For real-time voice and video interaction, Yunxin provides a pipeline in which user audio passes through automatic speech recognition, is processed by a large language model, converted to speech through text-to-speech synthesis, and streamed back as audio and video. Developers can configure the ASR, LLM, and TTS parameters through open API endpoints and choose between a visual agent development mode on the Yunxin platform or a flexible parameter configuration mode via API. Real-time captioning functionality is included. This real-time conversational agent system supports applications in e-commerce live streaming with virtual human presenters, virtual customer service in banking and financial services, immersive virtual meetings, telemedicine, remote bidding and procurement, and social entertainment.

The virtual human solution jointly developed by Yunxin and Fuxi Lab, accessible at netease.im/virtual-human, was first announced in November 2021 at the NetEase Innovation Enterprise Conference and represented the industry's first integrated virtual image and real-time communication SDK. The system uses a dual rendering architecture described as real-time capture on the device and real-time drive in the cloud. Ordinary monocular cameras on smartphones or computers detect facial expressions, head posture, eye movements, tongue position, and body gestures, and the captured motion data drives cloud-rendered avatars with lip-sync accuracy stated at ninety-six percent. The system works on budget devices priced around one thousand renminbi and supports cartoon, realistic, anime, and mixed avatar styles. A comprehensive three-dimensional face modeling capability allows one-click face customization and outfit changes without manual modeling, and AI-driven three-dimensional animation synthesis generates character lip sync, expressions, and body gestures from text or voice input alone.

Confirmed financial services clients using Yunxin's communications and digital human capabilities include Ping An Bank for remote banking transformation, Nanjing Bank for a video banking hall with video face-signing and video customer service, Hangzhou Bank and Changsha Bank for remote banking services, Bank of Communications for customer service livestreaming in its credit card application, and Yong'an Futures. Huatai Securities and CITIC Bank use the affiliated Yidun content security platform. Additional enterprise clients across the broader Shuzhi portfolio include Xiaomi, New Oriental Education, Zhilian Recruitment, OPPO, Vivo, Didi, Zhihu, Bilibili, People's Daily, and the Ministry of Foreign Affairs.

Youling, meaning "intelligent spirit," is not a single SDK but rather a brand ecosystem under NetEase Fuxi encompassing several related platforms and products. The Youling AI open platform, accessible at yl.163.com, provides end-to-end solutions for virtual digital human creation spanning character design, art and animation customization, voice customization, and integrated online and offline operations. Its core capabilities include virtual human animation driving and intelligent voice technology. A subsidiary art production service, Huimeng Tiangong, available at tiangong.yl.163.com, provides professional art services for virtual human and game asset creation and received a MUSE Silver Award in the second quarter of 2023.

The Youling Crowdsourcing Platform, accessible at zb.163.com and available as an iOS application developed by Guangzhou NetEase Computer System Co. Ltd., operates as a human-AI collaboration task platform for data collection and annotation. It supports reinforcement learning from human feedback training workflows and serves projects in robotics, gaming, augmented and virtual reality, and the metaverse. The platform actively recruits remote workers and has received a MUSE Award for its disability employment support initiatives.

The Youling Agent Platform is an AI agent system built on Fuxi's proprietary Agent-Oriented Programming framework, which integrates active learning, truth inference, and reward decomposition algorithms for applications in game AI, smart manufacturing, construction machinery, and data labeling. The virtual human SDK product commonly referred to as the Youling Virtual Human SDK is technically the Yunxin-Fuxi integrated Virtual Image and RTC SDK described above, co-developed across the two units and first demonstrated in late 2021 as part of NetEase's broader metaverse strategy. CEO Ding Lei stated at the time that NetEase was ready at all levels of technology and planning for the metaverse.

NetEase Youdao was founded in 2006 as a subsidiary of NetEase and listed on the New York Stock Exchange in October 2019. Its registered address is No. 399 Wangshang Road, Hangzhou, though its primary operational headquarters is in Beijing, where the developer entity Beijing NetEase Youdao Computer System Co. Ltd. is registered. Youdao serves over one hundred million users through its dictionary and translation products and employs between four thousand and six thousand people.

Youdao's chief executive officer is Feng Zhou, who joined the subsidiary in 2007 and has led it since 2018. Zhou holds a doctorate in computer science from the University of California, Berkeley, completed between 2002 and 2007, and bachelor's and master's degrees from Tsinghua University. He previously served as a senior vice president at NetEase. Zhou has articulated a vision in which education-focused AI progresses from proactive learning assistance toward fully autonomous virtual teachers within approximately two years. Renlei Liu serves as senior vice president and represented Youdao at the Consumer Electronics Show in January 2025, where he spoke on the company's edge model strategy and AI education globalization. Zhang Yi heads Youdao's Intelligent Applications Division.

Youdao's most prominent digital human product is Hi Echo, described as the world's first digital human oral English tutor, which launched in October 2023 and was upgraded to Hi Echo 2.0 in early 2024. Hi Echo presents as a virtual human avatar with vivid facial expressions that coaches spoken English with an American accent through real-time dialogue, personalized speaking exercises, instant pronunciation and grammar feedback, and comprehensive scoring reports. It covers eight dialogue scenarios and more than sixty-eight topics. The product is powered by Youdao's Zi Yue large language model, which was released in July 2023 as China's first educational large language model and received the highest accolade in the first evaluation series by the China Academy of Information and Communications Technology. Zi Yue 2.0 followed in January 2024 with improvements in knowledge question-answering, spoken dialogue, and text processing. By July 2024, Hi Echo had surpassed one million registered users. The product is available as a standalone mobile application, a WeChat mini-program, and as an embedded feature in Youdao's Dictionary Pen X7 hardware. A collaboration with IELTS added a built-in speaking preparation module, and Hi Echo 2.0 introduced oral level setting functions and an expanded range of virtual human roles. The Hi Echo website is hiecho.youdao.com.

Youdao also developed Mr. P AI Tutor, known as Xiao P Laoshi, described as China's first all-subject AI tutor, which integrates large language model and retrieval-augmented generation technology to provide guided explanations and multi-turn interactive teaching across all grades and subjects. It was implemented in the Youdao AI Study Machine X20 in January 2024 and in the Dictionary Pen X7. In February 2025, Youdao released SpaceOne, described as the first AI-native educational hardware to deeply integrate the DeepSeek-R1 inference model, featuring visual step-by-step problem-solving deduction. Additionally, Youdao has developed a small-sample digital human platform that allows users to create personalized virtual avatars supporting photo-to-speech and singing features.

The company's lightweight AI model strategy, articulated by Renlei Liu at CES 2025, emphasizes smaller models running on edge hardware for targeted training, enhanced precision, faster responses, and offline capability, in contrast to the large-and-comprehensive approach of general-purpose models. Youdao developed what it describes as the industry's first edge large language model based on Zi Yue, enabling its Dictionary Pen X7 to run AI features offline on consumer-grade hardware. An open-source lightweight inference model called Confucius-o1, with fourteen billion parameters, deploys on a single GPU card and uses chain-of-thought technology for step-by-step reasoning, achieving performance that Youdao claims matches OpenAI's o1 model for single-model inference. It is available on Hugging Face under the identifier netease-youdao/Confucius-o1-14B and on ModelScope. Youdao reports that its math problem-solving accuracy has reached ninety-six percent.

Youdao is also the developer of EmotiVoice, an open-source emotional text-to-speech engine released in approximately November 2023 under the Apache 2.0 license. The project is hosted at github.com/netease-youdao/EmotiVoice and accumulated more than 4,200 GitHub stars within its first week. EmotiVoice supports more than two thousand distinct pretrained voices in both English and Chinese, with prompt-controlled emotional synthesis covering happiness, excitement, sadness, anger, and other affective states, adjustable pitch, speed, and energy parameters, forty-eight-kilohertz WAV output quality, and voice cloning capabilities added in December 2023. The system is based on the PromptTTS architecture and trained on the LibriTTS and HiFi TTS datasets. It offers an OpenAI-compatible TTS API, streaming API support, and Docker deployment. Its roadmap includes Japanese and Korean language support.

NetEase Leihuo, known internationally as ThunderFire, was founded in 2007 as NetEase's first game development studio in Hangzhou. It has grown into a full business division with more than two thousand employees and serves as the organizational home of the Fuxi Lab. Leihuo develops several of NetEase's most prominent titles including Naraka Bladepoint, Ni Shui Han (also known as Justice), Qian Nü You Hun (Chinese Ghost Story), and All-Star Street Ball Party. Its motion capture center uses the world's highest-resolution Vicon cameras with Shogun software, supporting multi-person simultaneous full-body capture with direct binding to game models for real-time output. A professional scanning center produces high-fidelity assets for realistic digital humans. The CCF-NetEase Leihuo Joint Fund, a research partnership with the China Computer Federation established in 2023, supports academic research in game AI and related fields, with second-phase results announced in January 2025.

Digital human technology is most visibly deployed in the mobile version of Ni Shui Han, launched on June 30, 2023, which features text-to-face creation driven by the Yuyan language model, where poetic Chinese descriptions generate matching three-dimensional character faces in three seconds. The game also incorporates highly intelligent NPCs powered by Fuxi that possess personality, memory, goals, and emotional responses, enabling open-ended conversation. The title earned three billion renminbi in profit during July 2023 alone and was featured on CCTV as a showcase of Chinese AI innovation. Naraka Bladepoint employs AI expression synthesis for multi-language lip synchronization across Chinese, English, Japanese, and Korean, and AI teammates jointly developed with NVIDIA were presented at GDC 2025. The Chinese Ghost Story game franchise introduced conversational AI as early as 2018, making it reportedly the world's first MMORPG with such a feature, and debuted the virtual idol A Chu at its May 2019 conference, alongside an intelligent child-rearing system with AI children. Where Winds Meet, a newer title, incorporates AI-powered facial modeling, voice-to-appearance generation, and NPC interactions. The ThunderFire Universe X Studio, a sub-studio founded in 2021, focuses on console and overseas game development.

NetEase Cloud Music, separately listed on the Hong Kong Stock Exchange, has pursued virtual singer and digital human applications through a strategic partnership with Xiaoice, the AI company spun off from Microsoft China in 2020 and valued at approximately one billion dollars at its Series A round, in which NetEase was a direct investor alongside Hillhouse Capital Management, GGV Capital, IDG Capital, and Northern Light Venture Capital. On June 15, 2023, NetEase Cloud Music and Xiaoice jointly launched NetEase Cloud Music X Studio, a free AI singer music creation tool that uses deep neural network voice synthesis to generate natural singing voices. Alongside this, they created the WOWAIDO! virtual singer label, initially featuring twelve AI virtual singers spanning pop, folk, Chinese style, electronic, and rock genres. The virtual singer Chen Shuiruo performed at the People's Network 828 AI Night event. By 2023, the platform had expanded to more than twenty-five virtual singers. CEO Ding Lei observed that AI can create very mature musical works, particularly in the domain of arrangement. In a separate but related earlier event, the popular Chinese virtual singer Luo Tianyi performed at a NetEase Cloud Music concert at the National Stadium in Beijing on November 3, 2018, using performance technology provided by LATEGRA. NetEase also operates Tianyin, an AI music creation platform that automatically generates music from melody, theme, and lyric inputs.

Harry Shum's dual position as chief advisor to Fuxi Lab and chairman of Xiaoice creates a notable personal link between the two organizations, connecting NetEase's internal digital human research with the external AI company whose technology powers the virtual singer collaboration.

Several other NetEase units have deployed digital human technologies in more focused applications. NetEase Yanxuan, the company's e-commerce brand, created a virtual spokesperson named Lin Yaoyao using Fuxi and Youling technology for the brand's sixth anniversary celebration. The NetEase Interactive Entertainment AI Lab launched a virtual spokesperson called i.F., known as Yi Fu. NetEase Youdao developed a digital human study companion named Keke for online learning, co-developed with Fuxi and demonstrated at the 2021 Mobile World Congress in Shanghai with mixed-reality glasses. In the gaming space, the Onmyoji franchise has pursued a virtual idol project since May 2018, when characters from the game were repurposed as virtual performers. The character Otengu debuted at the Bilibili Macro Link Visual Release and Vsinger Live concert in July 2018, performing his exclusive character song.

NetEase Yidun, the content security arm within Shuzhi, provides AI anti-cheat capabilities as a software development kit that performs real-time analysis of player behavior to detect aim bots, wall hacks, and speed hacks, supporting multiple languages and listed on the AWS Marketplace for global SaaS distribution. While not a digital human product in itself, its behavioral detection models share foundational technology with the motion and behavior analysis systems underlying digital human animation.

A virtual digital human referenced as Linghua appears in research as a NetEase-affiliated virtual idol positioned within gaming and entertainment ecosystems, though with a smaller following than competitors from other major Chinese technology companies.

NetEase's partnership with NVIDIA centers on the NVIDIA Avatar Cloud Engine, or ACE, a suite of technologies for speech, animation, and AI-driven NPC interactions. At GDC 2024 and GTC 2024 in March 2024, NVIDIA announced NetEase Games as an official ACE partner alongside miHoYo, Tencent, Ubisoft, and Perfect World. The partnership encompasses NVIDIA Audio2Face for facial animation from audio, NVIDIA Riva for automatic speech recognition and text-to-speech, NVIDIA NeMo for large language model deployment, and NVIDIA RTX for ray-traced rendering. NetEase's Yaotai platform also uses NVIDIA Omniverse and CUDA for three-dimensional scene reconstruction, achieving the twenty-fold acceleration in neural implicit surface modeling described earlier. Hu Zhipeng, speaking on behalf of NetEase, stated that NVIDIA is making games more intelligent and playable through the adoption of gaming AI technologies. At GDC 2025, a joint presentation by NVIDIA and FuxiLab examined AI teammates in the mobile version of Naraka Bladepoint.

Beyond NVIDIA, NetEase's direct investment in Xiaoice's Series A round represents a significant financial commitment to external digital human technology. Other investments in the virtual human space include positions in IMVU, a virtual reality social and avatar platform; Cishidai Culture, a virtual human animation and content creation company; and Shiyue Xingcheng, a virtual human technology provider. These investments were identified in securities research reports from Tianfeng Securities. NetEase's broader acquisition strategy has also expanded its creative technology base through the full acquisition of Quantic Dream, the French interactive entertainment studio, in August 2022, and stakes in multiple Japanese studios including Grasshopper Manufacture, Ouka Studios, GPTRACK50, and Nagoshi Studio, as well as a capital and business alliance with the anime studio Satelight announced in August 2020.

Academic collaborations form a significant pillar of Fuxi's research output. The laboratory's work with the China Computer Federation through the CCF-NetEase Leihuo Joint Fund supports academic research across game AI and digital human topics. Digital twin replicas of Zhejiang University's Qiushi Conference Hall have been constructed on the Yaotai platform, and Peking University's Guanghua Management School has hosted events through the system. The DAI 2020 conference hosted inside the Ni Shui Han game world with Nanjing University's AI Academy provided the original impetus for Yaotai's creation.

Government and state media partnerships have given NetEase digital human technology high-profile visibility within China. The CCTV Online Spring Festival Gala metaverse venue in January 2023, the Zhongguancun Forum permanent metaverse venue, the Henan Provincial Department of Culture and Tourism's Yuan Yu Zhou project, and the Asian Games metaverse fireworks event all represent state-endorsed deployments. The laboratory's Intelligent Therapy Robot R1 was selected as a 2024 Typical Robot Application Scenario and won the 2024 iF Design Award. In November 2020, virtual digital humans based on Ni Shui Han characters were deployed at Hangzhou China Silk City as virtual guides, poets, and dancers, marking an early public demonstration of the technology in a cultural tourism setting.

List of subsidiaries:

1. NetEase Fuxi AI Lab (网易伏羲 AI 实验室): NetEase’s game-AI research lab that produces the company’s core digital human technologies, including avatar creation, facial performance, and interactive character intelligence.

2. NetEase Fuxi Technology (网易伏羲科技): The commercialization arm associated with Fuxi that packages and delivers digital human and game-AI solutions outside individual NetEase titles.

3. NetEase Yaotai (网易瑶台): NetEase’s immersive virtual event platform that deploys avatars and digital humans as the interface for meetings, ceremonies, exhibitions, and tourism-style virtual spaces.

4. NetEase Shuzhi (网易数智): NetEase’s enterprise services division that provides the business platform and customer distribution layer for digital human deployments in enterprise workflows.

5. NetEase Yunxin (网易云信): NetEase’s communications platform under Shuzhi that exposes digital human functionality through messaging and real-time audio/video SDKs.

6. NetEase Yidun (网易易盾): NetEase’s content and business security unit under Shuzhi that supports safe, compliant digital human operation in live, user-facing, and regulated contexts.

7. NetEase Yunshang (网易云商): NetEase’s service-marketing and growth unit under Shuzhi where digital humans can be deployed as customer-facing service and sales interfaces.

8. NetEase Youdao (网易有道): NetEase’s education subsidiary that deploys digital humans as tutors and study companions in consumer learning products and educational hardware.

9. NetEase Leihuo (ThunderFire) (网易雷火): NetEase’s game development group that serves as a primary production environment for shipping digital human features in live games.

10. NetEase Cloud Music (网易云音乐): NetEase’s music platform business that applies digital humans through virtual singers and AI-assisted music creation tools in partnership-led programs.

List of products:

1. Yaotai (瑶台): An immersive virtual event and venue platform where users participate through avatars and where NetEase showcases large-scale digital human and virtual world capabilities.

2. Yunxin Virtual Human Solution (网易云信虚拟人解决方案): A developer-facing module that combines avatar embodiment with real-time audio/video so a digital human can speak, emote, and interact live in communications scenarios.

3. Yunxin AI Digital Humans (IM Accounts) (网易云信 AI 虚拟人账号): Preconfigured AI digital human account types inside Yunxin messaging that can participate in 1:1 or group chat as conversational, translation, or search assistants.

4. Youling AI Open Platform (幽灵 AI 开放平台): An end-to-end virtual human creation and operation platform covering character production, animation driving, voice configuration, and deployment workflows.

5. Huimeng Tiangong (绘梦天工): A professional art and asset production service under the Youling ecosystem that supplies production-ready virtual human and related digital content assets.

6. Youling Crowdsourcing Platform (幽灵众包平台): A human-in-the-loop data collection and annotation product that supports training and refinement of digital human and adjacent AI systems.

7. Youling Agent Platform (幽灵智能体平台): An agent framework product used to build goal-directed behaviors and operational workflows for interactive characters and digital human applications.

8. Hi Echo (有道 Hi Echo): A digital human oral English tutor that delivers real-time dialogue practice with feedback and scoring through a virtual human persona.

9. Mr. P AI Tutor (Xiao P Laoshi) (小P老师): An all-subject tutoring product that provides guided explanations and interactive teaching and can be presented through a digital human tutor interface.

10. SpaceOne (有道 SpaceOne): An AI-native education hardware product line designed to run advanced tutoring features on-device and serve as a dedicated endpoint for digital human learning experiences.

11. Youdao Small-Sample Digital Human Platform (有道小样本数字人平台): A personalization product that enables user-specific avatar creation from limited input and supports photo-to-speech and related features.

12. EmotiVoice (情感语音合成 EmotiVoice): An emotional text-to-speech engine that provides expressive voice output suitable for digital human tutors, companions, and presenters.

13. NetEase Cloud Music X Studio (网易云音乐 X Studio): An AI singer creation tool that generates synthetic singing voices and supports virtual singer content production for creators.

14. WOWAIDO! (哇喂哆!): A virtual singer label that packages multiple AI singers into a managed catalog for releases, performances, and brand collaborations.

15. Tianyin (天音): An AI music creation platform that generates music from structured inputs and can supply or accelerate content production for virtual singer projects.

16. Lin Yaoyao (Yanxuan Virtual Spokesperson) (严选虚拟代言人 林瑶瑶): A brand-facing virtual spokesperson used for NetEase Yanxuan marketing and promotional communication.

17. i.F (Yi Fu) Virtual Spokesperson (网易互娱 AI Lab 虚拟代言人 i.F／易甫): A NetEase Interactive Entertainment virtual spokesperson product used as a public-facing digital character for brand communication.

18. Keke (可可): A digital human study companion concept used in education demonstrations as an avatar-based learning support persona.

19. Ni Shui Han Text-to-Face Character Creation (逆水寒 文本捏脸): An in-game digital human feature that generates a 3D character face from user text descriptions to accelerate avatar creation.

20. Naraka: Bladepoint AI Facial Performance and Teammate Features (永劫无间 AI 表情与队友功能): In-game digital human-related systems that support avatar performance behaviors and AI-assisted companion functionality in gameplay contexts.

21. Onmyoji Virtual Idol Program (阴阳师 虚拟偶像计划): A character-to-performer program that repurposes game IP characters as virtual performers for concerts and media events.