Economic Observer reporter Zhang Ling

In September 2026, an AI pharmaceutical company emerging from the ByteDance won an angel round financing of $290 million, becoming the first AI pharmaceutical company in China with a first round financing valuation of 10 billion yuan.

This company is called Anew Labs, which has been independent from ByteDance for only three months. It has no mature product and the drug pipeline is still in the pre clinical stage, but it has attracted many star institutions to enter the market. This has made many AI pharmaceutical startups envious.

Investor Li Mi, who participated in this round of financing, told the Economic Observer that the competition for financing was very fierce, and he "snatched" the investment quota from more than 100 institutions.

For ByteDance, this is a rare business split, and the last time it received such high attention was back to the end of 2023. Before the split, Anew Labs had been operating inside ByteDance for five years. After the split, ByteDance was still the controlling shareholder.

Anew Labs is not an isolated case of Internet giants entering life sciences. In the past few years, Baidu Group (BIDU. NASDAQ/09888. HK), Tencent Holdings (00700. HK), Alibaba (BABA. NASDAQ/09988. HK), Huawei, and others have all brought AI capabilities into drug research and development. Some have built their own teams, some have invested in industrial companies, and some have tried to use big models to enter drug discovery. In this crossover, ByteDance has taken a deeper step, not only as an AI model, but also as a drug pipeline.

Behind the enthusiasm of capital is an unproven cross-border experiment: can ByteDance computing power, talents and capital really be transformed into pharmaceutical capabilities? When Anew Labs moves from model making to drug making, can it establish its own competitiveness in a research and development system that is different from the Internet?

The Economic Observer sent a letter to Anew Labs regarding the reasons for the split, the significance of financing, and future development strategies, but as of press time, no formal response has been received. Liu Kai, the head of Anew Labs, responded that the company currently hopes to focus on research and development and will not accept interviews at this time.

Unknown Visitor

In April 2026, at the annual meeting of the American Society of Immunologists, ByteDance Anew Labs first disclosed the data of a small molecule candidate drug targeting IL-17, which is the core pipeline of its AI pharmaceutical business.

In this venue, which has been dominated by global biomedical enterprises for a long time, ByteDance are beating like strange visitors. This Chinese company, famous for its Internet and AI business, came here for the first time with a candidate drug designed with AI.

Liu Kai, the head of Anew Labs, later forwarded the relevant message on his social media and wrote, "I hope to use AI to do something that truly promotes the development of the pharmaceutical industry. Although it is difficult, I believe this is also where the true value of AI lies. ”

In the first few years of AI pharmaceutical layout, ByteDance was quite low-key. Team members rarely participate in pharmaceutical industry forums, and they rarely take the initiative to promote their papers after they are published. As a result, some people in the industry once believed that this was just a trial run for big companies. Founder of Jiaojing Biotechnology, Zhou Yiming, also stumbled upon a paper published by Anew Labs and made his first serious attempt at its algorithm.

It's both good and beautiful. "This was his feeling after using it for the first time.

The list of investors for Anew Labs' first round of financing is quite luxurious. Renowned investment institutions such as Sequoia China, IDG Capital, Hillhouse Capital, Wuyuan Capital, Gaorong Venture Capital, Chunhua Venture Capital, and Boyu Capital entered the market, while China Biopharmaceutical (01177. HK) and Shanghai Future Industry Fund participated as strategic investors.

Anew Labs can achieve its current results within 5 years, quite quickly. If their efficiency is not high, we will not participate in the investment. ”Li Mi said.

Li Mi has long been interested in AI pharmaceuticals and brain computer interfaces, and he has found that both tracks are equally popular, but the financing landscape is vastly different. Many brain computer interface projects are being pursued, but there are few AI pharmaceutical projects that allow institutions to concentrate their bets, and Anew Labs is one of them.

High financing has made some peers envious and surprised. Zhao Yu, co-founder of AI pharmaceutical company Zheyuan Technology, frankly stated that he "can't understand" and believes that the halo of big companies accounts for a large part.

Many people still cannot understand AI pharmaceuticals, and there is no particularly clear answer on how to do it and to what extent it can be achieved. ”Zhou Yiming believes that in the track where the answer is still unclear, the advantages of capital, technology and talent of ByteDance itself constitute a source of trust.

However, Zhou Yiming is not surprised by the current volume of Anew Labs.

As early as 2024, Jiaojing Biotechnology began using Anew Labs' protein model and was one of the earliest AI pharmaceutical companies to be exposed to its technology. In Zhou Yiming's opinion, the algorithm capability of ByteDance is very strong, even compared with the algorithms of foreign top AlphaFold, Diffusion and other companies, it is not inferior. More importantly, these models are completely open-source.

Nowadays, Anew Labs has launched biological AI models such as Protenix and PXDesign, covering drug development processes such as protein structure prediction and molecular design, and further launching AnewMind, a scientific reasoning model for drug development decision-making.

In Zhou Yiming's view, AI pharmaceutical tools can assist in the early stages of drug development, but it is difficult to support a large company solely with tools. The ultimate determinant of the value of an AI pharmaceutical company is whether it can sustainably create competitive pipelines.

Anew Labs is also attempting to move from model capabilities to drug pipelines. At present, it has 4 preclinical pipeline assets, with the core pipeline being a small molecule candidate drug targeting IL-17, which is in the preclinical lead compound optimization stage. According to Anew Labs, this is the world's first full spectrum oral small molecule inhibitor that targets both IL-17A and IL-17F simultaneously.

Unlike the perspective of AI pharmaceutical practitioners, Li Mi prefers to observe Anew Labs in the global market. In his opinion, the scale and valuation of Anew Labs' financing this time are not high. The financing scale of overseas companies such as Iso morphicLabs, as well as the investment of technology giants such as Microsoft in AI pharmaceuticals, has reached billions of dollars.

If we want to compete with globally competitive big models, an equal amount of funding is still necessary, "said Li Mi.

For industrial capital, there is another question behind the $1.5 billion valuation: whether these models can ultimately enter real drug development.

Cao Fenze, assistant vice president of Chia Tai Tianqing and head of the Digital Innovation Center, told the Economic Observer that in addition to financial returns, China Biopharmaceutical hopes to accelerate the application of cutting-edge technologies in real pipelines, find new drug opportunities with differentiation potential, and form research and development capabilities that can be reused in multiple projects in cooperation with technology giants such as ByteDance.

In Cao Fenze's view, the advantages of technology giants and their incubation teams lie in algorithm talents, computing power resources, and engineering systems. We will focus on how these advantages can be combined with drug development, whether they can produce replicable results in real projects, and whether there is a professional team and investment mechanism for continuous promotion

When AI pharmaceuticals leave big factories

Anew Labs is not an emerging team.

At the end of 2020, ByteDance began to look for talents in AI pharmaceutical field. In 2021, Liu Kai led a team to form an AI pharmaceutical team with about 50 core members, including both AI4S algorithm talents and senior pharmaceutical experts. In the following years, this team has been hiding in ByteDance until June 2026, when it officially went out of the big factory system and became an independent Anew Labs.

Liu Kai is not a typical technology entrepreneur. Before joining ByteDance, he successively worked in IDG Capital and Volcanic Stone Investment, and was responsible for the investment business in medical and biotechnology fields. Now, IDG Capital has also appeared on the list of investors for Anew Labs' first round of financing.

After the split, ByteDance injected the core team, algorithm platform and existing pipeline assets into Anew Labs as a whole, with Shanghai Yangpu as the domestic headquarters, and set up offices in San Jose and Singapore. ByteDance still holds the majority and continues to provide computing support for volcanic engines. According to Li Mi's understanding, Anew Labs currently maintains a small team of less than 100 people.

Why go out of ByteDance at this time?

Li Mi believes that the separation of AI pharmaceutical business by large companies is primarily a strategic priority choice. For tech giants, big models and general artificial intelligence are the first battlefield, while AI pharmaceuticals are in the second battlefield. After independence, teams can actually have greater strategic and execution space.

What Zhao Yu saw was another layer. AI and pharmaceuticals themselves are two vastly different business logics, and independent operation also facilitates the introduction of external capital. AI pharmaceutical research and development requires a large amount of funding, and the maturity of the technology has not been fully verified. Introducing external capital can diversify the risks borne by the group itself.

Zhou Yiming believes that Internet companies are accustomed to pursuing efficiency and rapid iteration, and feedback can be seen soon after a product is launched; Drug development is a typical industry of long termism. The decision made today may take a year or even ten years to know the outcome, accompanied by a high failure rate. This difference makes it difficult for the cultures and growth logic of the two types of companies to be compatible.

In addition to ByteDance, Baidu, Tencent, Huawei, Alibaba and other major manufacturers are also laying out in the field of biomedicine, but the paths are different. Currently, companies such as Tencent, Alibaba, and Huawei are keeping their related businesses within the group; ByteDance will develop AI pharmaceutical business independently after several years of internal development; Since its establishment, Baitu Biotechnology has built a team, technology, and research and development system according to the logic of a life science company.

These paths reflect the different strategic priorities and investment depths of large companies. Companies that are independent from the internal structure of large companies can focus more on their work, measure their main business more clearly, and recruit more personnel. "Liu Wei, co-founder and CEO of Baitu Shengke, analyzed the Economic Observer.

When founding Baitu Biotech, Baidu founders Robin Lee and Liu Wei did not consider placing AI pharmaceutical business inside Baidu. Liu Wei said that if life science reaches a certain depth, it will encounter parts that are unfamiliar to Internet giants. The underlying equipment and data of life science need long-term investment, and it often takes several years to see the results. It is difficult to follow the iterative method of Internet companies in this R&D rhythm.

In Liu Wei's view, the AI capabilities of large companies can radiate to various industries. If we only focus on AI empowerment, the core technology, solutions, team, and culture can share a set of systems. If we continue to delve deeper into the core issues of life sciences, we will encounter another set of logic and must construct a separate system.

There is also a practical problem behind the two different paths, which is how to calculate the return.

Lin Ze works at a large investment institution and has long been interested in the AI pharmaceutical industry. He observed that AI applications such as navigation, speech recognition, and shopping recommendations usually show quantifiable results within one or two months. If they are done well, they continue; if they are not done well, they stop. But the value of AI pharmaceuticals is difficult to calculate in this way. Even if AI helps design a good molecule, it is difficult to form a clear return mechanism unless the company sells the molecule or breaks down the project or company for financing. This is an important reason why some large companies have separated their AI pharmaceutical business.

Making medicine cannot rely solely on computing power

Resources and talents are the first advantages that appear after ByteDance enters AI Pharmaceuticals.

Li Mi introduced that in the algorithm stage, start-up companies usually receive an investment of $100 million to $200 million, corresponding to a team of dozens of people and a hundred card level computing power; ByteDance can invest hundreds of millions of dollars, use kilocalorie computing power, and have the ability to form a larger team. In addition, ByteDance originally has a mature AI talent and technology system. The large model team, algorithm talent reserve, and the attractiveness of the platform itself can all be transferred to AI pharmaceuticals.

However, resource advantages cannot automatically translate into pharmaceutical capabilities.

In Li Mi's view, the discourse systems of AI and biomedicine are different. For Anew Labs, which came out with the genes of a big company, it needs to maintain a sufficiently open attitude in order to continue attracting top biomedicine talents. After recruiting people, the original AI oriented evaluation system also needs to be iterated.

This difference will eventually fall onto specific pipelines.

Providing services for drug development may only earn 1 cent, while drug development itself may earn 9.99 yuan. "In Zhou Yiming's view, many of the tools currently provided by Anew Labs are essentially just a small step in helping drug early detection. It is difficult to support a large company solely by making tools, and greater development space needs to be obtained through pipeline development.

Zhou Yiming believes that at the pipeline level, Anew Labs is on the same starting line as other AI pharmaceutical and biotechnology companies, and even has a lower starting line than other companies. If there are no biopharmaceutical talents who can truly control the pipeline, relying solely on algorithm advantages may gradually lose competitiveness.

In the field of AI pharmaceuticals, different companies have chosen different development paths. One type is based on self-developed pipelines and directly participates in drug research and development; Another type focuses on life science big models and platform capabilities, providing underlying AI capabilities for biopharmaceutical companies. Liu Wei believes that the two types of paths correspond to different business logics and success conditions. For companies with self-developed pipelines as their core, the ultimate success depends on the selection of pipeline projects, targets, and indications, as well as luck. If the pipeline is selected incorrectly, no matter how good the AI is, it cannot be effective.

After AI pharmaceuticals reach the pipeline stage, judgments beyond algorithms become particularly important, which is a more unfamiliar challenge for Anew Labs. Choosing what targets to target, what molecules to make, what projects to advance, and what projects to stop losses, requires truly knowledgeable people to make judgments.

Several interviewees mentioned the importance of biopharmaceutical talent, including Huashen Zhiyao. In the past year and a half, this company has reached 5 partnerships with domestic and foreign pharmaceutical giants, with a total potential value of over 7 billion US dollars, making AI pharmaceutical peers around the world envious.

Zhou Yiming believes that the algorithm of Huashen Intelligent Pharmaceutical is not particularly outstanding in the industry. It can continuously make multiple heavyweight business development (BD) transactions, and one important reason is the introduction of Zhu Zhenping, the former R&D president of Sansheng Pharmaceutical (01530. HK), to be responsible for the pipeline.

Zhu Zhenping is a senior scientist in the field of new drug development, known for his expertise in target selection. During his time at San Sheng Pharmaceutical, he led the development of PD-1/VEGF bispecific antibodies. This pipeline was almost sold at a low price, but it was ultimately authorized to Pfizer for over $6 billion, shocking the industry.

In Zhou Yiming's opinion, Anew Labs needs to grow and strengthen, and finding a "Zhu Zhenping style" talent is more important than continuing to increase computing power and expand the algorithm team. But such talents are extremely scarce and hard to come by.

Anew Labs is also aware of the importance of biopharmaceutical talent. At present, it has formed a scientific advisory team consisting of several experts in the field of biomedicine, including Liu Yongjun, former president of Xinda Biotech, Ma Ji, former chief scientist of Anjin, and Zou Hua, director of protein chemistry science at Takeda Pharmaceuticals in California.

Finding someone who understands medicine is just the first step. How these types of talents and algorithm talents collaborate is also a challenge that Anew Labs needs to face. Zhao Yu introduced that these two types of talents naturally have high communication and collaboration costs. Algorithmic talents are accustomed to solving problems with clear boundaries, rules, and systematic logic, while biomedical talents rely more on understanding complex biological phenomena. If two teams lack chemical reactions, they may fall into internal friction and slow down business.

Computational chemistry has developed quite maturely, with a deep accumulation of methodological systems and talent supply. However, when computing and biology are truly combined, the situation is completely different. Zhao Yu said that whether Anew Labs can form a real technological barrier depends on whether it can transform computing power into the ability to understand life and disease. At present, there is still a scarcity of talents in the industry who can understand complex biological problems and transform them into computational problems, and a mature talent training system has not yet been formed.

Why Make Medicine

ByteDance is an unexpected and reasonable crossover.

The belief of a generation of young scientists was that the 21st century is the century of life sciences. Over the past decade, Internet entrepreneurs such as Ma Yun, Robin Lee, Huang Zheng and Wang Xiaochuan have all focused on life science.

In 2013, shortly after resigning as CEO of Alibaba Group, Jack Ma became a donor to the Global Life Science Breakthrough Prize Foundation. Later, Ma Huateng, Ding Lei, Robin Lee and others participated in the donation of the Future Science Award. Life science is an important field supported by the award.

In 2020, Robin Lee and Liu Wei jointly founded Baitu Shengke, and Liu Wei served as the CEO of Baitu Shengke. This company received hundreds of millions of dollars in financing in its second year of establishment. At that time, Robin Lee recalled that he was interested in life science when he was a programmer on Wall Street more than 20 years ago, and even applied for the work of Merck research team.

In 2021, Huang Zheng, founder of Pinduoduo (PDD. NASDAQ), expressed his desire to enter the field of life sciences in a shareholder letter; In the same year, after stepping down as CEO of Sogou, Wang Xiaochuan also positioned his dreams for the next 20 years in life sciences and medicine.

It was also in this year that Zhang Yiming began to look further ahead. He said in an internal letter that the impact of virtual reality, life sciences, and scientific computing on human life has shown the dawn, and to seize this opportunity, innovators need to break through business inertia and explore. For this reason, he decided to step down as CEO of ByteDance and participate in this wave of technological change for a period of ten years.

Looking back, the ByteDance layout of AI pharmaceutical business was exactly in this year.

In Zhou Yiming's view, once people have sufficient wealth, their demand for health and longevity will naturally further increase, leading to attention and layout in related fields.

Lin Ze said that the biomedicine industry has a great imagination and attractive market space, which is also the reason why Internet tycoons have paid attention to biomedicine for a long time.

Zhao Yu also believes that large companies entering the AI pharmaceutical field have seen the long-term industrial value of life sciences. Life sciences are closely related to everyone, with a large market space and certain countercyclical properties.

Beyond emotions and industry imagination, the advancement of technology provides a suitable opportunity for big shots to step into the game.

In 2018, when OpenAI released GPT-1, the large model had only over 100 million parameters, and the actual effect was not significant. By 2020, GPT-3 had reached 175 billion parameters, and a basic model began to be able to handle hundreds of natural language tasks. In Liu Wei's view, this means that there has been a leap in the ability of the basic model. In the same year, DeepMind developed Al-phaFold2, which solved the protein folding problem that had plagued researchers for nearly 50 years. Afterwards, AI began to delve deeper into more complex scientific problems from the information world.

Why has AI entered this field more frequently recently? Because although this field is large, it was not easy to enter before. ”Liu Wei said that in the past, life sciences relied more on traditional experiments, and the data that could be utilized by machine learning was limited, or even non-existent. In recent years, with the development of various omics technologies, life sciences have begun to generate and accumulate more and more data, and the physical world is constantly being digitized, giving AI greater opportunities for application.

Liu Wei noticed that the landing of AI in the field of life sciences is at an important turning point. In the past two years, a series of awards in the scientific community have been presented to AI+life sciences, representing the recognition of the classical scientific community. At the same time, the FDA is now accelerating and even strongly demanding the use of AI to replace some traditional clinical and animal experiments, providing various acceleration channels for AI designed drugs, which represents the attitude of regulatory authorities.

Li Mi judged that AI pharmaceuticals are still in the stage of going from 0 to 1, and the current investment is only the "first money". With the continued development of technology, there may be sustained huge investments in AI in the field of life sciences, similar to the large-scale model industry. As long as humans want to live longer and solve diseases, the pharmaceutical industry will exist for a long time. Therefore, life sciences will be a long-term industry that runs through the 21st century

(At the request of the interviewee, Li Mi and Lin Ze are given pseudonyms in the article)

(Author Zhang Ling)