Xuanjia Technology's B1 and B2 funding rounds landed in quick succession; founder Qin Lin says commercialization in the AI video track is accelerating

Recently, Xuanjia Technology completed its B1 round of financing; about 30 days later, the B2 round followed, jointly invested by Shum Yip Capital (Shum Yip Cloudwalk AI Fund), Fachuang Capital, and Jiangbei New Area Science and Technology Investment Group. The company is currently advancing its B3 round of financing, with cumulative financing exceeding 200 million yuan and an enterprise valuation expected to surpass 2 billion yuan.

From the perspective of investors, Shum Yip Capital can bring in Shenzhen state-owned assets and AI industry fund resources, Jiangbei Sci-Tech Investment is Xuanjia's long-term co-builder in the Yangtze River Digital Intelligence Street district, and Fachuang Capital is a market-oriented financial investor. What this capital is densely betting on is the qualifications, implementation, and technology layout this company—transformed from the large-screen audiovisual sector—has formed in the field of AI video large models.

Xuanjia Technology was founded in 2013 and has long been deeply engaged in the intelligent audiovisual field, with its traditional large-screen business covering 27 provinces nationwide and about 160 million household users. In recent years the company has pushed into the AI video sector; in May and July of this year, its self-developed Kino-AIGC video large model and Zhibiaobao text large model successively passed the generative artificial intelligence service filing of the Cyberspace Administration of China.

At a recent media open day, Xuanjia Technology founder and CEO Qin Lin shared his judgment on the stage of the AI video industry. In his view, the industry has bid farewell to the early stage of competing over demo effects and leaderboard scores, and the focus of industrial competition has shifted to AI video industrialized mass-production capability; whether finished films can be delivered stably, at low cost, and in a controllable manner on a continuous basis is the true productivity yardstick for video large models.

Around this judgment, Qin Lin further broke down the industry chain landscape: the AI video industry chain has clearly differentiated into a three-tier structure of computing power foundation, video large models, and content production and distribution, commercialization in the sector is accelerating toward viability, and short dramas are the optimal scenario at the current stage for validating the mass-production chain and realizing commercial value. "AI has restructured the cost structure of audiovisual content; relying on batch production tools, the production cost of a 120-minute short drama can be compressed from 300,000 to 500,000 yuan to about 5,000 yuan, a mature team can produce hundreds of short dramas per day in batches, and an assembly-line content production model has already been implemented."

The large-scale implementation of the mass-production model has also given rise to business models suited to the AI video industry. "Considering the high cost of video large model training and inference, subscription will become the mainstream business model in the sector; a company's positioning should not be merely that of a model provider, but that of an AI video infrastructure service provider, building an integrated product system of foundation models, mass-production toolchains, and industry solutions," Qin Lin said. In his view, the final competition in the AI video sector lies not in the stunning quality of a single sample film, but in a comprehensive contest of controllable consistency, mass-production cost, and compliant delivery capability.

This set of industry judgments is also corroborated by Xuanjia's own business growth curve. Since the beginning of this year, the company's AI business has rapidly scaled up. According to Qin Lin, Xuanjia's AI business revenue is growing at about 50% per month; as of the media open day, the platform had more than 100,000 online registered users and more than 10,000 registered enterprise entities.

In the process of advancing commercialization, compliance is a core threshold that cannot be ignored. Qin Lin pointed out that audiovisual content has strong regulatory attributes and the filing threshold for large models is high; compliance qualifications themselves are a prerequisite threshold for large-scale commercialization. Building on a compliance foundation, Xuanjia is also planning a longer-cycle technology evolution roadmap. In the KinoAI platform architecture released by Xuanjia, the intelligence layer, in addition to the Zhibiaobao text large model and the Kino video large model, also reserves a world model section, positioned as "simulation, interaction, prediction."

According to the R&D plan, Kino World Model 1.0 will build on video generation capabilities to explore world state memory, spatiotemporal consistency, and interactive generation. The "stage-adaptive world model" the company is developing attempts to maintain long-term consistency of characters and scenes when the picture is stable, strengthen geometric and motion modeling during moments such as rapid movement and collisions, and dynamically allocate computing power accordingly.