Three Roads to the Virtual Human: Japan, China, and South Korea

Over three decades the virtual human — a digitally constructed figure performing cultural, commercial, or communicative work once reserved for living people — has traveled from a curiosity of computer graphics to a consequential global industry. Three countries did more than any others to carry it there, and each took a different road. Japan supplied the concept and the creative culture that made it durable. South Korea fused virtual-character technology to the entertainment export machine that already carried its influence abroad. China industrialized virtual-human production at a scale no one else has matched, and is now moving to govern it more comprehensively than any jurisdiction on earth. Read together, the three trajectories reveal what is singular about the Chinese model: its distinctive contribution is not invention or artistry but volume, standardization, and the coordinating hand of the state.

Japan's claim to primacy runs back to the mid-1990s, when the talent agency Horipro unveiled Kyoko Date, a fully computer-generated three-dimensional idol built to occupy the same promotional ecosystem as flesh-and-blood pop stars. Her 1996 debut is widely regarded as the first real-life virtual idol, and it was a commercial failure, her lone single selling fewer than thirty thousand copies amid complaints about the uncanny valley her appearance produced. The concept survived its own false start. The transformative moment came with Vocaloid, Yamaha's singing-voice synthesizer, and above all with Hatsune Miku, released by Crypton Future Media in 2007. What set Miku apart from earlier virtual performers was not her voice but the participatory culture that erupted around her: users created well over a hundred thousand original songs and more than a million derivative artworks, encouraged by a licensing framework that deliberately permitted non-commercial reuse. Nearly a decade later, in late 2016, Kizuna AI coined the term "Virtual YouTuber" in a single self-introduction video, naming an entire industry into being. The agencies followed — Hololive, operated by Cover Corporation, and Nijisanji among them — and by the mid-2020s the format had become serious business, with Cover reporting revenue near 43.4 billion yen for fiscal 2025 and setting a medium-term target of roughly a hundred billion yen by 2030. Japan's road, across all of it, has run through the depth of its creative ecosystem and the cultural legitimacy it conferred on virtual performers over thirty years — a lineage that by early 2026 had produced the format's strangest milestone yet in Neuro-sama, an AI-driven VTuber whose near-continuous broadcasts made it the most-subscribed streamer on Twitch, ahead of every human on the platform.

China entered later and followed foreign models first. HiPiHi, a three-dimensional virtual world comparable to Second Life, was founded in Beijing in 2006, partnered with IBM, struggled with weak engagement, and shut its servers in 2012. What came afterward, however, had no parallel anywhere. By 2024 the country was home to roughly 1.144 million companies engaged in digital-human work, with more than 174,000 of them registered in the first five months of that year alone; the research firm iiMedia valued the core market at 339.2 billion yuan and projected it toward 935.6 billion by 2030, with the broader driven market running into the trillions. Leaders emerged across every niche. XiaoIce, spun out of Microsoft as an independent Chinese company in 2020, accumulated 660 million registered users and generated tens of millions of digital humans. SenseTime's SenseAvatar platform could produce a speaking, gesturing avatar from as little as thirty seconds of footage. Shanghai's Ranmai Technology created AYAYI, widely recognized as China's first hyper-realistic metahuman, who went viral on Xiaohongshu in 2021. Silicon Intelligence, working the e-commerce end of the field, had cloned more than half a million avatars and was running upward of fifty thousand livestreams a day. The scale was concentrated in identifiable hubs; a 2025 city-competitiveness index placed Shenzhen first, ahead of Beijing, Guangzhou, Shanghai, and Hangzhou, mapping the industry onto the same coastal technology corridor that had already produced the country's internet giants. And unlike its neighbors, China folded the entire field into its industrial-planning apparatus. The Three-Year Action Plan for the metaverse, issued jointly in 2023 by five ministries and agencies, called explicitly for the cultivation of benchmark digital-human products and a handful of globally influential firms, while Beijing's municipal government set itself the target of a fifty-billion-yuan digital-human industry.

Governance followed the same centralized logic. The 2022 provisions on deep-synthesis internet services imposed watermarking and labeling requirements on synthetic media, a category that expressly covered digital-human generation, face replacement, and voice synthesis. Then, on April 3, 2026, the Cyberspace Administration of China released a far more targeted draft, the Digital Virtual Human Information Service Management Measures. It would mandate a visible "digital human" label for the entire duration a virtual figure is displayed, prohibit the use of avatars resembling specific real people without consent, establish special protections for digital recreations of the deceased, restrict companion services that might induce minors to overspend, and require real-name authentication, all under oversight coordinated across telecom, police, cultural, health, financial, and broadcasting authorities. No other country is attempting to govern the field this comprehensively, and the draft is of a piece with a national approach that treats the virtual human as infrastructure to be planned and regulated rather than a cultural phenomenon to be watched.

South Korea took the third road, routing virtual humans straight through the pop-culture machinery that had already made it a cultural power. Firms such as DeepBrain AI and Pulse 9 built the production tools, the latter creating Eternity, the country's first virtual K-pop girl group. The breakout, though, was Plave, a five-member virtual boy band whose hidden performers wear motion-capture suits that translate their movements in real time into animated avatars rendered in Unreal Engine. Its February 2025 mini-album sold more than a million copies in its first week, and the group became the fastest act in history to reach a billion streams on the Melon platform, overtaking a record set by NewJeans. Korea also produced the sharpest legal test the field has yet seen. In May 2025 a district court ruled that insulting a virtual avatar constitutes defamation of the real person behind it, reasoning that in the age of the metaverse an avatar functions as a means of self-expression and identity rather than a mere image — an early precedent for how the law might adapt to characters that are commercially real but personally anonymous. That ruling arrived alongside a broader regulatory shift: Korea's Framework Act on artificial intelligence, passed by a lopsided margin in December 2024 and effective from January 2026, made the country the second jurisdiction after the European Union to adopt comprehensive AI legislation.

The three roads are not parallel but crossing, and where they cross they sometimes collide. In September 2020 two Hololive VTubers mentioned Taiwan while reading YouTube analytics data — the platform lists it as a separate region — and the remark detonated among Chinese viewers on Bilibili. Cover Corporation suspended both performers and issued statements that diverged by language, the Chinese-language version endorsing the One-China principle while the English and Japanese versions did not, a discrepancy that angered audiences on every side. Within months the entire six-member Hololive China branch had dissolved. The episode revealed that virtual performers, for all their engineered controllability, inherit the political frictions of the humans and markets they move between; a figure designed to be flawless and scandal-proof could still become the flashpoint of an international dispute.

By the mid-2020s all three industries had entered a phase of consolidation, regulatory maturation, and rising ambition, but each along the road it had chosen. Japan's strength remains the depth of its creative ecosystem and three decades of accumulated legitimacy, from Kyoko Date's tentative debut to Cover's march toward a hundred-billion-yen business. Korea's is the power of an existing cultural juggernaut, capable of producing in Plave a virtual act whose commercial numbers rival those of the biggest human idol groups. China's is scale and coordination — more than a million firms, a national standards apparatus, and a governance framework moving toward the most comprehensive anywhere in the world. The trajectories are increasingly interdependent, bound by shared technologies and overlapping markets, and they converge on a single conclusion: the virtual human is no longer a novelty but a permanent and consequential feature of the global digital economy, and China, having industrialized it, is now positioned to set the terms on which it is governed.