AI pharmaceuticals are waiting for super giants.

On June 10, it was reported that ByteDance split its AI pharmaceutical business and started independent financing. The core team, algorithm platform and existing pipeline assets were moved into the new entity as a whole. Byte continued to hold shares and the volcanic engine continued to provide computing power.

Looking at the entire industry, this scene has long been seen, because this is not the first time that big companies have appeared in the narrative of AI pharmaceuticals.

Moreover, with biopharmaceuticals being explicitly listed as a national pillar industry, more and more players are deciding to invest their future in this field, let alone AI pharmaceuticals which are currently the hottest industry.

However, compared to the past, ByteDance's recent split adjustment has made the industry realize that super giants are really here. AI pharmaceuticals are no longer an "exploratory project" for tech giants, but an industry that can be done and is worth cultivating.

Regarding this, Zhang Jiang, founder and CEO of LongRiver Jiangyuan Investment, analyzed to Xieyi Jun that "if this news is true, it essentially means that AI pharmaceuticals have moved from pure algorithm research and development to an industry stage that requires independent operation and can also be commercialized independently. ”

 

1、byte entry, come for real

The layout of ByteDance in the AI pharmaceutical field is far earlier and deeper than the public awareness. As early as 2020, ByteDance has quietly laid out the AI drug discovery track internally, launched the recruitment of special talents, and advanced the cutting-edge field.

In 2021, a team dedicated to AI pharmaceuticals was officially established, led by Liu Kai, with a core of about 50 members, including AI4S (AI for Science) algorithm experts and senior pharmaceutical industry talents.

In the early stages of the team's establishment, the core goal was to focus on polishing the underlying infrastructure, deeply cultivating basic model research, and building core underlying capabilities for protein structure prediction. At this stage, ByteDance's AI pharmaceutical business is more focused on building scientific research infrastructure, with a focus on strengthening the technological foundation, and there is still a certain distance to go before it can truly produce pharmaceuticals.

The core prerequisite for independent business financing is to have mature technological confidence and industrial strength. If rumors of a spin off come to fruition, it means that the entity has mature development conditions. According to publicly available information, the technical foundation and preclinical verification stages of each research and development pipeline have been laid out, and the overall technical system is becoming more complete, "Zhang Jiang pointed out to Xieyi Jun.

Therefore, looking back at the past year, the pace of ByteAI's pharmaceutical business has accelerated comprehensively. In 2025, ByteDance will successively launch two core models, Protenix and Seedfold, and complete the v2 version iteration in 2026; Synchronize the landing protein design tool PXDesign to build a comprehensive protein research and development tool matrix. In addition, Byte also launched Anew Labs, an AI pharmaceutical platform for the whole process of real drug research and development, to upgrade from a single model tool to an integrated research and development platform.

The key turning point occurred at the annual meeting of the American Society of Immunologists in April this year, when the ByteAI pharmaceutical team publicly disclosed a preclinical IL-17 small molecule inhibitor, officially completing its pipeline debut.

According to its introduction, this is not a conventional Fast Follow pipeline, but the world's first "full spectrum" IL17 small molecule inhibitor. This project demonstrates ByteDance's use of internal AI technology to achieve comprehensive blockade of the IL-17 family (AA/AF/FF) at the small molecule level for the first time.

For ByteDance, this industry debut has become a new starting point for its AI pharmaceutical business, not only outputting models and tools, but also outputting real pipelines.
According to the information, the main entity behind this pipeline is Anew Labs. According to the official website, the development pipeline of IL17AA/AF/FF is currently in the preclinical lead compound optimization stage. In addition, IL4R has entered the Hit to Lead stage, while the other two undisclosed targets are in the Hit discovery stage.

The technological foundation is mature, and the pipeline is starting to advance. It seems natural to separate and promote independent financing. Zhang Jiang further pointed out to Xi Yijun that "the pace of the pharmaceutical business is completely different from that of traditional Internet companies. The clinical pipeline will run for five to ten years, and talent incentives should follow the way of Biotech. The investors introduced should also have VC's with medical insurance resources or biomedical background, rather than just TMT capital. To dismantle it is to provide it with a governance structure that is more in line with industry norms. ”

Simply put, drugs are not software, and the schedule of clinical trials will not be adjusted for anyone. ByteDance's split this time is a clear statement through an organizational action that this business should follow the rules of pharmaceuticals and be done more deeply.

It is reported that after the split, ByteDance will still hold the controlling stake in the new company, while Volcano Engine will continue to provide computing power - structurally independent, and the resource link has not been cut off.  

2、talent outflow sends signal earlier than splitting

In fact, before the business split was implemented, the departure of core talents had already laid the groundwork for the organizational change of ByteDance AI Pharmaceuticals, and also revealed the hidden currents behind the entry of large companies into the AI pharmaceutical track.

On June 2, in 2023, Gu Quan joined ByteDance and announced his resignation through overseas social media officials. Later, many media described him as "the head of AI pharmacy and big model pre training business of the ByteDance Seed team", which attracted wide attention.

Regarding the claim of related titles, on June 8th, the person in charge of ByteSeed denied the rumors to the public, stating that Gu Quanquan had only participated in part of the work related to biological molecular structure prediction and LLM pre training, and was not the person in charge of AI pharmaceuticals and large model pre training business.

The departure of core members of the team, coupled with the divergence of opinions between the company and the outside world regarding their job titles, has made the industry keenly aware of the subtle changes within the ByteAI pharmaceutical team. The day after ByteDance's official response, Gu Quanquan publicly stated on Xiaohongshu that the official statement was "inconsistent with the facts" and confirmed in the comment section that he had been appointed as the head of AI pharmaceuticals since July 2023.

He also added that during his tenure at ByteDance, his work responsibilities and achievements in the field of AI pharmaceuticals have been widely recognized within the company. He is deeply puzzled by the false information circulating from the outside world, and also believes that this has left the team members who fought side by side feeling puzzled. As of now, the two sides have not reached a consensus on the relevant statements, and the dispute has not yet come to an end.

However, it can be made clear that regardless of the outcome of the dispute, the integrity of the resigned company and the split of ByteDance continue to intensify their efforts in the AI pharmaceutical industry. It is reported that Gu Quanquan has officially started his entrepreneurial journey after leaving, and his new project has successfully obtained multiple rounds of financing from top investment institutions in the United States, receiving strong endorsement from top tier capital.

A key talent deeply involved in the core work of ByteAI pharmaceuticals quickly gained the favor of top capital after leaving. This result itself has significant signaling significance, as it confirms the capital heat and industry talent scarcity in the AI pharmaceutical field.

Regarding this, Zhang Jiang admitted that "the most difficult part of AI for science is not technology selection, but organizational design. The interdisciplinary talents needed for AI pharmaceuticals are extremely scarce - they need to understand AI algorithms, have a background in structural biology, understand pharmaceutical chemistry, and have automation engineering capabilities. These groups of people also need to be able to understand each other. This is very scarce globally. ”

The mismatch of incentive rhythm is another structural contradiction. AI people are accustomed to producing results and iterating quickly within six months; Pharmaceutical professionals are accustomed to a five to ten year cycle and a regulatory driven rhythm. These two types of people are placed in the same organization, and what is needed is not just adaptation, but a set of incentives and management systems specially designed for this industry. ”Zhang Jiang added.

From this perspective, ByteDance's independent financing is not just a capital level action, but also proves that if big companies want to do a good job in AI pharmaceutical business, they need a new organizational container.  

3、The entry of giants not only changes the competitive landscape

ByteDance is not the only big company betting on this track. In recent years, there have been many cross-border players, including Huawei, Baidu, and Tencent, all of whom have entered the game and have their own paths of focus.

Huawei is taking the lowest level path, not building pipelines or targeting, but integrating computing power, AI toolchain, and industrial software into the entire pharmaceutical process.
In June 2026, the "Pharmaceutical Industry Digital Transformation Promotion Center" under the guidance of the Consumer Goods Industry Department of the Ministry of Industry and Information Technology was officially established. Huawei, together with more than ten enterprises such as Hengrui Pharmaceutical, Fosun Pharma, and Jingtai Technology, participated in the digital transformation of pharmaceutical companies to deeply embed their supply side roles. Wide coverage and risk diversification, but maintaining a certain distance from upstream pipeline research and development.

Baidu's Baitu Biotechnology is deeply involved in computational biology and life science models, focusing on platform technology empowerment and exporting AI drug development tools and solutions to the outside world. Tencent focuses on financial investment and ecological resource integration as its core entry point, without personally developing technology and pipelines. Through precise layout of high-quality track resources, Tencent has participated in the two benchmark AI pharmaceutical companies in the track - Jingtai Technology and Yingsi Intelligence, relying on capital ties to deeply bind industry leaders and maintain its discourse power in the track.

After years of development, the player structure of the AI pharmaceutical industry is now clear, and the industry ecology is becoming more perfect, which can be divided into three core camps as a whole.

One is the underlying infrastructure players represented by Huawei and Baidu (Baitu Shengke), who focus on computing power, self-developed large model bases, and tool platforms, without laying out specific drug pipelines; The second is AI native startup companies represented by Jingtai Technology and Yingsi Intelligence, some of which focus on end-to-end pipelines, while others specialize in underlying algorithms or vertical scenarios; The third is the direct line of technology giants represented by ByteAnew Labs and Isomorphic Labs, which have both powerful model bases and self built real pipelines, following a dual track path of "technology self research+pipeline self support".

The entry of large companies will increase talent costs and compress the recruitment space of small and medium-sized companies in the short term. However, the public use of open-source toolchains and basic models is also simultaneously lowering the engineering threshold of the entire industry, allowing teams focused on vertical disease fields or specific targets to concentrate limited resources in truly differentiated areas. Competition pressure and infrastructure dividends are coming at the same time, "Zhang Jiang said frankly to Yi Yi Jun.

With the continuous rise of industry heat and the increasing number of players entering the field, the competitive landscape of AI pharmaceuticals is still undergoing iterative restructuring. From the perspective of capital, the differentiation of track value has already emerged.

Standing at the forefront of the trend, Zhang Jiang believes that there are currently two types of enterprises that are being monitored by investors. "One type has its own pipeline and clinical data, and investors are buying clinical progress and the ability to cash in BD; The other type is more like the CRO model, providing intermediate layer services. Both can make a good company, but the most crucial thing is whether several core competencies are in place. ”

In Zhangjiang's view, the construction of industry barriers mainly relies on five core dimensions:

One is the dry wet closed loop, which requires not only algorithms, but also their own experimental verification capabilities, or deep binding with the laboratory. Only algorithms without experimental data feedback will quickly reach the ceiling of the model;

The second is that the pipeline strategy can be validated, either by pushing it to the IND or PCC stage with clear clinical signals, or by having a clear BD pathway for large pharmaceutical companies;

The third is differentiated data sovereignty. "Only with proprietary data can there be better models, better molecules, and better experiments, which can create long-term barriers;
The fourth aspect is team cohesion, where the ideal level is for Co founders to understand both AI and medicine, as well as mutual understanding, which is extremely scarce globally;
The fifth is the awareness of input-output ratio, which is not about burning money to chase model parameters, ranking and brushing PR, but truly results oriented.

Even though the logic of industry barriers is very clear and the pace of industry landing continues to accelerate, the AI pharmaceutical track has not yet ushered in true ultimate verification. Zhang Jiang candidly stated that "there is currently no drug in the world that is fully designed by AI that has been approved. Some of the drugs involved in AI have already reached phase three, and everyone is waiting for the first approval to prove it. If proven true, the valuation of the entire track will be reconstructed; If there are problems with key projects, shuffling will accelerate. ” 

- freehand summary -

AI pharmaceuticals have come this far, with no shortage of heat or stories, but what is lacking is a moment of clinical significance for 'ultimate proof'.

ByteDance chooses to independently finance at this node and follow the rules of pharmaceuticals. Other large manufacturers pour in in their own ways, and start-ups dig deep barriers at each subdivision level. These concurrent forces together form the current appearance of the track: full of expectations and uncertainty.

AI pharmaceuticals have moved from 'can molecules be designed' to 'whether the designed molecules are useful in humans'. This is a huge progress, but the most difficult part is still ahead. ”Zhangjiang summarized.

With a batch of core AI new drug projects rushing towards the third phase of clinical trials, the ultimate validation that the industry has been waiting for many years is now within reach.
Under the current trend, the answer to whether the myth of AI pharmaceuticals can be realized and whether the industry value can be truly realized will soon be revealed in clinical results.