Exclusive | ByteDance's AI drug discovery unit spins off for financing, AI4S enters industrialization phase

Text by | Zhou Xinyu

Edited by | Zhang Yuxin

Exclusive information obtained by "Intelligent Emergence" reveals that ByteDance's AI pharmaceutical business line has initiated the process of spin - off and independent financing.

It is reported that after the spin - off, ByteDance will still hold a controlling stake in the new company. The core team, core algorithms, technology platform, and existing pipeline assets of AI pharmaceuticals will be transferred to the new entity as a whole. Meanwhile, this business will continue to receive computing power support from Volcengine.

The new company will be led by ByteDance's AI pharmaceutical team. The ByteDance AI pharmaceutical team was established in 2021 and is led by Liu Kai. It is reported that the core members are about 50, composed of AI4S algorithm talents and senior pharmaceutical experts. Since its establishment, this team has undertaken the core functions from basic model research to industrialization.

Previously, the relevant team within ByteDance responsible for the protein structure prediction model has also been merged into the AI pharmaceutical team led by Liu Kai. The relevant algorithm model teams have completed the integration and will continue to advance the basic model research in this field, with a small number of personnel leaving.

The business progress of ByteDance's AI pharmaceuticals is an important basis for this spin - off and financing.

ByteDance has multiple technological achievements in the field of AI pharmaceuticals. In 2025, ByteDance's AI4S team released the molecular structure prediction models Protenix and Seedfold, and iterated Protenix - v1/v2 in 2026, building high - precision open - source structure prediction capabilities for biological complex systems such as proteins and ligands.

In terms of protein design and prediction, the team has launched protein binder design tools such as PXDesign.

Meanwhile, ByteDance has also launched the AI pharmaceutical platform Anew Labs for real - world drug R & D.

The official website of Anew Labs shows that the team has published research such as AnewSampling, AnewOmni, AnewFEP, AnewSynth, scNext, covering areas such as protein - ligand dynamic structure prediction, all - atom molecular generation, free energy calculation, synthesis feasibility prediction, and virtual cells, and has launched early - stage drug pipelines such as IL17AA/AF/FF and IL4R.

In April 2026, Anew Labs first disclosed the IL - 17 small - molecule project at the annual meeting of the American Society for Immunology, achieving for the first time globally the blockade of three dimers (AA/AF/FF) of the IL - 17 family with small molecules. Since IL - 17 is an important pathway for autoimmune diseases such as psoriasis and ankylosing spondylitis, and the simultaneous inhibition of A/F (two key inflammatory factors) has been verified to have clinical value by antibody drugs.

This indicates that ByteDance's AI pharmaceutical capabilities have advanced from models and algorithms to the verification stage of specific targets, specific molecules, and internal pipelines.

With technological progress and continuous exploration in AI pharmaceuticals, ByteDance believes that the opportunity from scientific research to industry is maturing. Therefore, it has integrated internal teams in various directions and is determined to test the waters of industrialization.

Of course, the industrialization of AI4S faces great challenges.

The verification cycle of the AI4S business is long, and the processes are more complex. Taking pharmaceuticals as an example, it covers multiple complex processes such as model R & D, wet experiments, and clinical verification. Therefore, it has a greater demand for professional talents, and the required organizational management form is also different from that of Internet businesses.

People familiar with the matter revealed that this business spin - off is to establish an independent organizational structure that better suits the characteristics of this business. ByteDance hopes to attract top - notch talents through this adjustment, thereby promoting the combination of basic model capabilities in this field, algorithms, and the pharmaceutical industry.

Meanwhile, the pharmaceutical industry itself is also under efficiency pressure.

Over the past two decades, the R & D investment of global pharmaceutical companies has been continuously increasing. IQVIA, one of the world's largest medical health data and clinical research service companies, predicts that global drug expenditure will reach approximately $2.3 trillion by 2028.

The market size is large enough, but the core pain points of high R & D costs, long cycles, and high failure rates of new drugs have not fundamentally changed. The industry is eager to introduce AI technology to break through these limitations.

Currently, AI4S research is accelerating, as evidenced by its significantly enhanced ability to solve complex problems.

Take the iteration of the AlphaFold (a protein structure prediction model developed by Google DeepMind) series as an example: from the initial verification of feasibility, to AlphaFold 2 achieving atomic - level precision prediction of 200 million proteins, and then to AlphaFold 3 breaking through the limitation of single proteins and accurately predicting complex interaction systems - this proves that AI has penetrated into important links of drug design.

If protein structure prediction is still a basic research problem, then the multi - modal molecular generation models that have emerged in recent years directly address the core problem of the pharmaceutical industry - drug design, which may also indicate that AI pharmaceuticals are gradually moving from research to industrial application.

ByteDance has been deploying in AI4S for many years. Around 2020, ByteDance began to systematically enter fields such as AI pharmaceuticals, molecular simulation, and computational biology. Since then, it has teams covering areas such as first - principles calculations, quantum chemistry, molecular dynamics, material simulation, and energy and drug molecule generation.

After the establishment of the large - model research team Seed, AI4S has also become part of ByteDance's cutting - edge technology layout.

A person close to this spin - off said that this is ByteDance's first attempt at the industrialization of AI4S, and the company attaches great importance to it internally. "Biotechnology has its own industrial logic. By spinning off, it has the flexibility of independent decision - making, and we hope to establish a successful industrial path for AI4S in China."