AIGC unicorn "Silicon Intelligence" has completed its Series D financing. Its digital human business has achieved hundreds of millions in revenue and served over 500,000 users | Exclusive report by Yingke.

Author | Huang Nan

Editor | Peng Xiaoqiu

Yingke has learned that Nanjing Silicon Intelligence Technology Group Co., Ltd. (hereinafter referred to as "Silicon Intelligence"), an AIGC unicorn, recently completed a Series D financing of hundreds of millions of yuan. The investor is Jiaxing High - tech Zone Industrial Fund (Jiaxing Xiuzhou District). The funds from this round will be mainly used for R & D innovation investment, promoting the implementation of technologies, and accelerating the marketization of products.

Since its establishment in 2017, "Silicon Intelligence" has completed more than 1 billion yuan in financing over 10 rounds. Its investors include Tencent, Sequoia China, CMBC International, Zhongcai Qihoo, Haisong Capital, etc. The company focuses on the R & D and commercial application of AIGC digital human technology. It proposes to build a core around silicon - based labor force and has constructed four product lines: the DUIX intelligent interaction platform, the virtual live - streaming platform, the video production platform, and the intelligent customer service.

With the rapid development of technologies such as large AI models, computer vision, and speech synthesis, the digital human industry has rapidly emerged in China. The virtual avatars generated through AI technology can simulate human appearance, voice, and expressions and have interactive capabilities. They are widely used in fields such as live - streaming, customer service, education, and entertainment.

According to a report from iResearch, the market size of digital humans in China exceeded 10 billion yuan in 2023 and is expected to reach 30 billion yuan in 2025, with a compound annual growth rate of over 50%.

Currently, a large number of AI Agents have emerged in the market. As the inference cost of large models continues to decline, the technical threshold and commercial barriers of those simply providing logic and process automation capabilities will quickly disappear. "Purely instrumental Agents will definitely end up in a fierce price war," Sima Huapeng, the founder of "Silicon Intelligence", told Yingke.

Sima Huapeng, the founder of Silicon Intelligence, attended the 2024 36Kr WISE Business King Conference

Sima Huapeng pointed out that the intelligent level of many industries still remains at a primary stage. In fact, there is a wide - spread demand for high - quality Agents. AI naturally has the ability of a thinking chain based on logical reasoning and can solve problems in an absolutely rational way. Based on this, the digital human agents derived from it have also begun to enter more scenarios that require in - depth reasoning.

"For example, a doctor's inference of a disease is essentially a process of using professional knowledge to conduct a thinking - chain derivation based on information such as the evidence chain and patient characteristics," Sima Huapeng said. "There are a large number of thinking - chain application nodes in all industries. In the future, these nodes may gradually be taken over by AI, forming a 'human - machine collaboration' work model. That is, AI processes standardized processes based on thinking - chain logic, while humans focus on more core areas such as creative decision - making and emotional interaction."

The algorithm technology platform is the underlying support for the digital human product system of "Silicon Intelligence". In November 2024, "Silicon Intelligence" launched a new - generation multimodal large model, DUIX ONE, in the form of an API. This model integrates core algorithm capabilities such as speech recognition, speech synthesis, NLP, and computer vision and can provide multimodal capabilities such as seeing, hearing, speaking, high emotional intelligence, and having a real - time human - like appearance.

Its advantage lies in greatly enhancing the sense of realism in interaction, effectively solving the problem of traditional digital humans being "similar in form but different in essence". At the same time, it also has stronger universality and application potential in the market - implementation stage.

Silicon Intelligence's DUIX ONE multimodal large model

In terms of speech technology, DUIX ONE can accurately recognize speech content with different accents and speaking speeds, ensure stable operation in noisy environments, and has high accuracy and strong noise - resistance ability. Then it can generate natural, smooth, and emotional human voices, achieving an effect close to that of real - human voices.

In terms of digital human image construction, "Silicon Intelligence" can quickly and accurately create highly realistic facial models through a complete facial modeling technology. The virtual avatar driving algorithm can drive the facial expressions and body movements of digital humans in real - time based on inputs such as speech and text, ensuring the vividness and naturalness of their actions.

In the interaction environment, DUIX ONE can deeply understand the semantics and emotions behind human language and provide functions such as intelligent dialogue, text classification, and knowledge Q&A. During the process, the digital human will generate rigorous and content - rich responses according to the dialogue scenario and user needs.

In terms of commercialization, "Silicon Intelligence" adopts a differentiated product strategy and provides customized service solutions for enterprise customers and individual users. For the enterprise market (the 2B field), the company focuses on serving customers such as operators, e - commerce platforms, and media institutions. Through standardized API interface services, it realizes the modular output of AI capabilities, significantly reducing the technical application threshold for enterprises.

Taking the e - commerce live - streaming scenario as an example, enterprise customers only need to provide about 5 minutes of video materials to quickly complete the personalized customization of a digital human anchor on the AI training platform. The platform supports the modeling of all - dimensional features such as appearance, voice, and language style, highly restoring the characteristics of real - human anchors. The generated digital human anchor also has real - time intelligent interaction capabilities, can accurately understand viewers' questions, and respond according to the comments, enhancing the interactivity of the live - streaming room and increasing the viewer retention rate.

The implementation and application of Silicon digital humans in various industries

Currently, the revenue of the digital human business of "Silicon Intelligence" has reached hundreds of millions of yuan. It has provided digital human anchor solutions for more than 5,000 cooperative brands, helping merchants reduce costs by about 80%. In the financial industry, it has cooperated with more than 30 banks and insurance companies and deployed intelligent customer - service digital humans. It has also cooperated with mainstream TV stations to develop virtual hosts, etc.

For C - end users, "Silicon Intelligence" has launched a digital human "cloning" service to help users create their exclusive digital avatars and realize personalized functions such as virtual avatar customization and intelligent companionship. At the same time, on the DUIX intelligent interaction platform and the video production platform, ordinary users can easily complete professional - level creations such as "digital human video generation" and "digital human real - time interaction", further lowering the threshold for digital content creation.

Yingke has learned that "Silicon Intelligence" has reduced the price of its digital human cloning service to a few thousand yuan. The production process uses a fully automated process. With just a 1 - second video, it can complete the cloning of the appearance and voice. While ensuring quick delivery within one working day, it also significantly reduces the technical application cost.

The latest official data shows that the platform of "Silicon Intelligence" serves more than 500,000 users, with the daily average user - generated content exceeding 500,000 pieces. The cumulative AIGC live - streaming duration is over tens of millions of hours. Through the in - depth application of digital human technology in scenarios such as short - videos and live - streaming, more and more ordinary users are transforming from passive content consumers to active content creators.

To further promote the popularization of AIGC technology, "Silicon Intelligence" has fully open - sourced the core technology of the DUIX platform. The currently released DUIX - Mobile version supports Android and iOS platforms and provides a minimalist integration solution for developers. By simply calling three lines of core code, they can quickly embed digital human functions into their own product systems, shortening the development cycle and accelerating the large - scale implementation of digital humans in more industries such as entertainment, law, finance, education, and healthcare.

As digital humans that integrate the capabilities of large AI multimodal models and can have real - time conversations gradually penetrate into more fields and enter people's lives, this is not only a technological advancement but also redefines the way of emotional transmission. "Digital immortality" is expected to move from concept to reality. When human emotions and memories are preserved across time and space through digital human technology, this also expands a brand - new dimension for the form of life.