CarbonSilicon AI Completes RMB 50 Million Angel Round to Advance AI-Driven Drug Discovery Platform

VCBeat learned第一时间获悉，近日，CarbonSilicon AI Technology Co., Ltd. (hereinafter referred to as "CarbonSilicon AI") has completed an angel round of financing worth 50 million RMB.This round of financing was jointly led by Legend Capital and Lenovo Star.The financing funds will be mainly used to build a one-stop new drug design platform with independent intellectual property rights, covering the entire process of new drug discovery, and to create innovative drug design services driven by data and rapid iteration of dry and wet experiments.

CarbonSilicon AI Focuses on New Drug Development, Aiming to Become an Artificial Intelligence Infrastructure and Service Provider in the Field of New Drug Research. The company aims to deeply integrate advanced life sciences with information technologies such as artificial intelligence, driving progress through AI, physical computing, and automated hardware and software technologies. By enhancing capabilities in producing, managing, and AI modeling of data within new drug development, it seeks to digitize and intelligize every aspect of the process, addressing key challenges in new drug research.

In the AI pharmaceuticals field, what new vitality will the establishment of CarbonSilicon AI inject into the industry? VCBeat interviewed Deng Yafeng, founder of CarbonSilicon AI, and Professor Hou Tingjun at the first opportunity.

CarbonSilicon AI CEO Deng Yafeng and Chief Scientist Hou Tingjun each have 20 years of deep expertise in their respective fields.

Deng Yafeng's work experience has continuously evolved with the development and iteration of AI technology. He has witnessed the technological advancements of artificial intelligence over the past two decades, as well as the industrial implementation of AI in China over the past ten years.There are not many people who can persist in artificial intelligence in the industrial sector and eventually enter the core field of AI, and Deng Yafeng is one of them.

In 2013, deep learning emerged as a promising field for large-scale applications in artificial intelligence. Deng Yafeng joined Baidu's IDL (Institute of Deep Learning) — the earliest institution in China dedicated to deep learning research and development, which later became known as the "Huangpu Military Academy" of Chinese AI. During his three years at Baidu, he and his team applied deep learning technology to various subfields of facial recognition, achieving world-leading results in facial recognition and face detection. However, compared to leaderboard achievements, Deng Yafeng was more eager to see AI technologies widely implemented in the industry. In 2016, he joined DeepGlint as CTO, helping build its deep learning technology system and advancing facial recognition, vehicle recognition, and person re-identification to industry-leading levels. DeepGlint later became the first AI company listed on China's STAR Market. By then, he had accumulated 14 years of R&D experience in artificial intelligence, particularly in computer vision, and obtained 95 authorized patents during his previous work.

After DeepGlint's development got on track, in 2020, Deng Yafeng joined the 360 Group as the Group Vice President, Dean of the Artificial Intelligence Institute, and General Manager of the Search Business Unit, leading a product and technology team of about 400 people. During his time at 360, he promoted the implementation of deep learning technologies such as pre-trained large models, multi-modal representation, and knowledge graphs in areas like the Internet and smart hardware. He once led his team to achieve nearly a 20% improvement in user metrics for search products—a significant enhancement for a mature product with hundreds of millions of users. In 2021, Deng Yafeng was named one of the Top Ten Influential Figures in China’s Artificial Intelligence Industry of the Year.

As AI technology gradually matured in fields such as the internet, mobile internet, smart cities, and intelligent commerce, Deng Yafeng yearned for greater challenges, hoping to gain more accomplishment and a sense of value. He began exploring new directions, seeking a field that could genuinely help improve individual lives and be driven by AI.

AI + New Drug Discovery Entered the Vision of Deng Yafeng. He Found That Life Science Research Is a Complex and Precise Information System, Many Modules Can Be Modeled with AI, and New Drug Discovery Is a Field Where Efficiency Can Be Significantly Improved Using Artificial Intelligence Technology."This is exactly what I've been looking for," he said. "It's a field with tremendous social value, enormous commercial potential, and remains a blue ocean in terms of technology. This gives us the opportunity to do work that gets closer to the essence, driving social progress through fundamental technological innovation."

In the exploration of AI-driven new drug discovery, Tingjun Hou became the best partner for Yafeng Deng.

Hou Tingjun is highly regarded in the field of AI-assisted drug design. He completed his bachelor's, master's, and Ph.D. at Peking University. As a leading talent in scientific and technological innovation under China’s "Ten Thousand Talents Plan," he has been awarded the Elsevier Highly Cited Chinese Researcher, the WuXi AppTec Life Chemistry Research Award, and recognized as one of the "Top 1%" Highly Cited Chinese Authors by the Royal Society of Chemistry (UK). He has published over 400 SCI academic papers, including 30 ESI Highly Cited and Extended ESI Highly Cited papers, with a total citation count (Google) exceeding 20,000, an H-index of 70, 19 software copyrights, 43 invention patents, of which 23 have been authorized.

As a specially appointed professor of the College of Pharmaceutical Sciences at Zhejiang University, Tingjun Hou has over 20 years of experience in methodology and application research in drug design. In the recently released 2022 Global Scholar Academic Influence Rankings, he ranked third among leading figures in China's pharmaceutical discipline. The research team he leads is also one of the best teams in China in the field of AI-assisted drug design.

A shared vision inspired Deng Yafeng and Hou Tingjun to consider collaboration; their humble and pragmatic personalities turned them into good friends, while their complementary strengths accelerated their joint entrepreneurial journey.

Currently, at CarbonSilicon AI, the two have clear divisions of labor: Deng Yafeng, as Chairman and CEO, is responsible for the company's strategic planning, operation management, and the research and development of artificial intelligence software and hardware products; Hou Tingjun, as Chief Scientist, focuses on the company’s research and development and layout in the pharmaceutical field, as well as the exploration of cutting-edge directions.

In addition, the founding team of CarbonSilicon AI also includes Shi Hui, who has served as the Vice President of Haodaifu Online and the General Manager of Alibaba Health Internet Hospital, acting as the COO to oversee the company's operations management. Professor Cao Dongsheng from Central South University, a long-time academic research partner of Hou Tingjun, is participating in the development of pharmaceutical + AI algorithm models in the capacity of Chief Algorithm Scientist.

The establishment of CarbonSilicon AI resulted from the consensus reached by Deng Yafeng and Hou Tingjun after multiple exchanges. They believed that,In the field of new drug discovery, the efficiency and success rate of research and development have been relatively low. The core reason is the over-reliance on expert experience. Human experts struggle to fully grasp all the knowledge involved in the drug development process and find it difficult to completely digest and utilize all experimental data, making drug development full of uncertainty.They believe that the future will surely usher in an era of artificial intelligence-driven life sciences. Specifically,In the field of new drug discovery, the model will certainly shift from relying on experts + experiments to AI computation + automated experiments + expert collaboration.Under the new paradigm, AI is not meant to replace human experts but to deeply collaborate with them. In the process of this paradigm shift, whoever builds the design platform based on AI modeling and automated hardware and software first will seize the initiative in new drug development.

Building such a design platform, on the one hand, requires deep expertise in artificial intelligence modeling. This is because the algorithms in the entire new drug discovery field are still relatively in their early stages and undergoing rapid changes. Unlike fields like computer vision or natural language understanding, there aren’t mature frameworks or models that can be directly applied, demanding very strong algorithm modeling capabilities from the team. On the other hand, these models and algorithm strategies need to form a complete workflow. Drug design is a long chain composed of multiple stages, similar to EDA tools in chip design, requiring a platform that allows drug design experts to see the full picture of drug design while forming a closed loop for data and model iteration. This approach differs from the traditional toolbox mindset in the field, which involves stacking individual tools rather than creating an integrated software platform covering the entire process. Based on data iteration models generated by automated hardware, this kind of platform incorporates experts into the workflow, ultimately forming a unified design platform driven by data and human-machine collaboration, significantly improving the certainty of drug research and development.

This shared vision became the original intention for Deng Yafeng and Hou Tingjun to establish CarbonSilicon AI.They hope to start from the underlying core technology and eventually form a one-stop design platform covering the entire process of new drug discovery, improving the success rate and speed of drug research and development, and solving the difficulties in new drug research and development.

The screening of苗头化合物, optimization of先导化合物, in vitro experiments, biological experiments, and clinical trials (Phases I, II, and III)—so many complex steps determine the long cycle and high costs characteristic of the pharmaceutical industry, which is also the fundamental reason for the high failure rate in new drug development. The traditional "expert + experiment-driven" working model heavily relies on expert experience, but humans are not adept at handling vast amounts of data across numerous stages, with each expert only being familiar with knowledge in a specific subfield. Artificial intelligence, however, has the potential to understand and model all data within the drug development domain, integrate human knowledge, collaborate with experts through human-machine cooperation, and enhance the success rate of new drug development across the entire chain and multiple stages.

"Tasks that are less suitable for experts have the potential to be enhanced or replaced by AI. For example, experiments—currently, manual experimentation is highly inefficient, with a Ph.D. student only able to synthesize about 100 compounds in five years, which is not very productive. Through automated experimentation, both efficiency and consistency can be improved. Furthermore, the data generated from these experiments can better feed back into models, continuously improving their capabilities, forming a closed loop, and reducing reliance on experts." In Deng Yafeng's view, the new model of 'intelligent computing + automated experimentation + experts' will provide new momentum to the pharmaceutical industry. From this perspective, CarbonSilicon AI will focus more on algorithm modeling based on AI and physical computing, building a one-stop software process based on models, and ultimately integrating automated experimentation to form a data-driven closed-loop system.

Currently, CarbonSilicon AI has established DrugFlow, a one-stop new drug discovery SaaS platform with independent intellectual property rights. This platform includes modules such as target evaluation, virtual screening, molecule generation and optimization, and drug-likeness prediction, which can help medicinal chemistry experts find potential drug-like molecules more efficiently and conveniently.Among them, several modules, such as the ADMET2.0 module based on Molecular Graph Analysis (MGA), the IGN algorithm for molecular protein activity prediction based on graph models, and the MCMG molecular generation algorithm based on reinforcement learning and deep generation, are at the leading edge of the industry and hold significant innovative value.

Even though AI is believed to have tremendous development potential in the pharmaceutical field, Deng Yafeng and Hou Tingjun remain rational and calm in their understanding of this area.

Deng Yafeng believes that although the AI + new drug discovery track is very popular, it is still in the early stage and remains an unsolved problem.In any industry, only when AI technology reaches or even surpasses the level of human experts does it truly reach the moment of explosion. In the field of AI + new drug discovery, this most important milestone has not yet been reached.

"Now, looking back, no single technology has been disruptive," said Hou Tingjun, who does not want people to have overly high expectations for AI technology in the short term. "Take AlphaFold2, for example, which has drawn significant attention in the industry. It has convinced more people of AI's capabilities and is a milestone, but from the perspective of drug development, it has not yet solved the core issues in this field. There is still much work to be done."

AI Pharmaceutical Track: Financing Events Often Reach Tens of Millions of Dollars, Filling Many with High Expectations for AI Technology, Even Bestowing It with the Halo of Revolutionizing the Pharmaceutical Industry, But It Is Also Often Questioned. In Deng Yafeng's View, People Prefer to See to Believe; Most Will Only Choose to Believe When They See Results. And He Believes That the Companies Which Will Stand Out in the Future Are Those That Deeply Integrate AI Design Platforms with Pharmaceutical Experts—This Process Is Inevitable, It Just Takes Time.

For the AI pharmaceuticals sector, which gained significant attention as early as 2020, CarbonSilicon AI is not among the first batch of entrants. However, it represents an innovative company that has chosen a more rational and pragmatic development path as the industry evolves to a certain stage, moving away from bold claims of disruption. Deng Yafeng believes that the impact of AI on the field of new drug discovery and even the life sciences has only just begun. The current state is very similar to how computer vision technology was in 2013 and 2014—change is only just beginning.

Legend CapitalRepresentation: The life science industry is currently at the cusp of an explosion, and introducing technologies such as AI and automation into fields like pharmaceuticals has become an inevitable trend. CarbonSilicon AI is transforming the traditional new drug R&D model, which was "expert + experiment-driven," into a novel model that is "intelligent computing + automated experimentation + expert-driven." Its newly launched drug discovery platform, DrugFlow, enables machine learning modeling for target evaluation, virtual screening, lead compound optimization, and drug-likeness prediction. By empowering traditional pharmaceutical companies with a SaaS product combined with services, it breaks through key digital barriers in the pharmaceutical field. Legend Capital will continue to focus on CarbonSilicon AI and leverage its rich CVC 2.0 industrial ecosystem to help CarbonSilicon AI accelerate the arrival of the AI-driven life science era.

Legend StarUnder the current framework system of pharmaceutical R&D and regulation, the wide gate has closed. In the future, the pharmaceutical industry will no longer differentiate between AI-based and non-AI-based pharmaceutical companies. The AI-driven pharmaceutical concept, built on the foundation of algorithms, computing power, data, and high-throughput experimental iteration, will become a required course for all participants in the innovation drug competition. As the low-hanging fruits in the pharmaceutical field have been almost entirely harvested, participants must explore every possible new technology or mechanism to empower themselves to pass through the narrow gate. CarbonSilicon AI is composed of senior AI experts and pharmaceutical experts who have accumulated extensive engineering and practical experience in AI and pharmaceuticals. We look forward to them playing the role of a ferryman in this era.

Legend Capital is Lenovo Group's global technology industry fund, focusing on investing in early-stage core technologies. It has invested in more than 200 outstanding technology companies, including over 10 IPO companies such as CATL, Meituan, NIO, Cambricon, SUPCON, Zhuhai CosMX, Shenzhen Urban Transport Consultants, SmartSens, and Hygon, as well as over 40 unicorns in various subfields like Megvii, 4Paradigm, BYD Semiconductor, CloudMinds, and SemiDrive. At the same time, it has incubated more than 10 subsidiaries and innovative businesses from within the group, such as SHAREit, which has 2.4 billion users, and National Identity. Moreover, Legend Capital pays attention to the development of "specialized, refined, distinctive, and innovative" small and medium-sized enterprises (SMEs). Currently, it has invested in more than 40 national/provincial and municipal-level "specialized, refined, distinctive, and innovative" SMEs, leading the CVC industry with the highest proportion of such enterprises among its total investments.

Legend Star is one of the earliest early-stage investment institutions to systematically layout in the healthcare sector, having prioritized healthcare as a key investment focus since 2010. In this field, Legend Star has established its strategic layout in two major sectors: "Biopharmaceuticals" and "Digital Healthcare." Biopharmaceuticals include innovative drugs, gene technology and services, diagnostics, and high-value medical consumables. Digital healthcare refers to empowering and upgrading drug research and development, medical devices, medical services, insurance payments, and health management through technologies such as artificial intelligence, big data, and robotics. Currently, Legend Star has invested in over 100 high-quality healthcare projects, including Kintor Pharmaceutical (9939.HK), Burning Rock Biotech (NASDAQ: BNR), Axonics (NASDAQ: AXNX), Mabwell (2162.HK), Pagatech, Cadyx, Qihan Biotech, HiFiBiO, Deepwise Healthcare, Jingfeng Surgical Robotics, Exegen Bio, Changmugu, Stronglink Intelligence, Yihong Health, and Huihe Medical.