ByteDance's drug discovery unit, Anew Labs, has begun fielding teams at major international science and AI conferences in Boston, Rio de Janeiro, and Barcelona, presenting generative AI-designed therapies and computational chemistry frameworks that signal the Chinese tech giant's serious ambitions in pharmaceutical AI.
Most people still think of ByteDance as the company behind TikTok. That framing is increasingly incomplete. The company has been expanding aggressively into AI infrastructure, large language models, and now pharmaceutical drug discovery, and the pace at which it is doing all three simultaneously is worth paying attention to. Anew Labs, ByteDance's dedicated drug discovery unit, also operating under the names Anew Therapeutics and ByteDance AI Drug Discovery, is not a skunkworks experiment or a PR exercise. It has 36 core members, a scientific advisory board stacked with people who spent careers at Innovent Biologics, Amgen, and Takeda, and it is now presenting original research at the conferences where serious drug discovery work gets evaluated by peers.
In mid-April, Chris Li, Anew Labs' head of biology, presented one of the unit's four pipeline drug candidates at Immunology2026 in Boston, organised by the American Association of Immunologists. The treatment in question was designed by generative AI to target certain autoimmune diseases. She Yuli, Anew's head of data, attended the International Conference on Learning Representations in Rio de Janeiro in late April, one of the most selective venues in deep learning research. Yu Haoyu, head of computational chemistry, is headed to Barcelona for the 2026 Free Energy Workshop starting May 4, where he plans to present AnewSampling, a framework the team has built to reproduce molecular dynamics at the all-atom level. Three different conferences, three different scientific disciplines, all in the span of a few weeks. This is a coordinated push, not coincidence.
The scientific advisory board deserves more than a name-check. Liu Yongjun served as president of Innovent Biologics, one of China's most commercially successful biopharmaceutical companies, with a track record of bringing monoclonal antibodies through clinical development and regulatory approval. Ji Ma was a principal scientist at Amgen, a company that invented the field of recombinant protein therapeutics. Hua Zou is scientific director of protein chemistry at Takeda California, one of the largest pharmaceutical companies in the world. These are not celebrity advisors lending their names to a credibility exercise. They are domain experts whose involvement signals that Anew Labs is pursuing a credible scientific program rather than a demo platform.
ByteDance's broader AI investment context matters here. The company released Protenix-v1 earlier this year, an open-source biomolecular structure prediction model that matches AlphaFold3-level performance in predicting protein, DNA, RNA, and ligand structures. It made the code and model parameters publicly available under an Apache 2.0 licence. That release was part of ByteDance's positioning as a serious AI research organisation rather than purely a consumer technology company. Protenix-v1 is directly relevant to drug discovery: predicting how proteins fold and how small molecules bind to them is foundational to designing therapies that actually work. Anew Labs has the internal compute, the model development capability, and now the domain talent to turn that infrastructure into drug candidates.
The four pipeline candidates Anew Labs currently has in development are focused on oral small molecules for immunology targets, specifically protein-protein interaction targets that have been validated by antibody research but are difficult to drug with conventional small molecule approaches. This is not an easy area to work in. PPI targets have historically resisted small molecule drug discovery because the interaction surfaces are large and flat, lacking the well-defined pockets that traditional medicinal chemistry looks for. AI-driven approaches, particularly generative models that can explore chemical space more broadly than human chemists and molecular dynamics simulations that can model binding at atomic resolution, offer a genuine route through that difficulty. AnewSampling, the framework Yu Haoyu is presenting in Barcelona, is directly aimed at improving the accuracy of those simulations.
The geopolitical dimension to this is worth acknowledging plainly. ByteDance is a Chinese company operating under intensifying regulatory pressure in the United States. TikTok's future in the US market remains contested. Against that backdrop, ByteDance is building a drug discovery operation with offices in Shanghai, Singapore, and San Jose, presenting research at American and European science conferences, and recruiting talent from Western pharma companies. Whether US regulators treat AI-designed drug candidates developed by a Chinese tech company as a national security consideration has not been tested yet, but it is a question that will eventually need an answer, particularly if Anew Labs advances candidates into clinical trials in Western markets.
The pharmaceutical industry's AI transformation is already underway. Companies like Recursion, Insilico Medicine, and Relay Therapeutics have been building AI drug discovery platforms for years, and the first fully AI-nominated drug candidates are approaching late-stage trials. ByteDance entering this space with significant compute resources, proprietary foundation models, and a team assembled from the best addresses in biotech is not a marginal development. It is a well-resourced new entrant arriving at exactly the moment when the technology is mature enough to produce real candidates. The conference presentations are the first public signal that Anew Labs is ready to be evaluated on its science. The peer community will decide whether it measures up.
Also read: Google told employees it is proud of its Pentagon AI contract and the era of internal veto power over defense deals is effectively over • The Oscars will not recognize AI actors or AI-written scripts and that decision will shape how investors value synthetic media startups • Apple Got Caught Using Anthropic's Claude and the Internet Had a Field Day
