ByteDance Enters AI Drug Discovery Race as South Korea Ramps Up Investment

TikTok's owner is moving into pharmaceutical labs. Anew Labs, a company spun off from ByteDance, unveiled at two major scientific conferences the first therapies designed entirely by artificial intelligence, and governments across Asia, led by South Korea, are responding by accelerating their own AI-in-pharma programs so as not to fall behind.

Anew Labs steps into the spotlight

Anew Labs, founded in 2021, spent years operating as an internal ByteDance research project focused on AI for science. In April 2026, the company showed the world the fruits of that work for the first time: a small-molecule IL-17 inhibitor, generatively designed to block all three dimeric variants of the inflammatory protein at once. According to the company, this is the first time in the world that a small molecule, rather than an antibody, has achieved this kind of blockade.

The presentation took place at the annual meeting of the American Association of Immunologists in Boston. A month later, Anew Labs also appeared at the BIO International Convention in San Diego, and the company's head of computational chemistry spoke at the Free Energy Workshop in Barcelona. The firm has four drug candidates in its portfolio and says its generative model, AnewOmni, trained on more than 5 million biomolecular complexes, achieves a success rate of 23-75 percent when designing molecules that target the KRAS G12D and PCSK9 proteins.

Spinning off from ByteDance

On June 10, 2026, ByteDance formally began spinning off its AI4S pharmaceutical unit into a separate, independently funded company, transferring the team, algorithms, technology platform and existing projects to it. ByteDance, however, retains a controlling stake. The team, led by Kai Liu, numbers roughly 36-50 core members, including AI-for-science specialists and pharmaceutical industry veterans, and its scientific advisory board has been joined by, among others, former Innovent Biologics chairman Liu Yongjun and former Amgen chief scientist Ji Ma.

The company operates simultaneously in Shanghai, Singapore and San Jose, California, a footprint meant to ease access to Western clinical partners and regulators. The computing infrastructure ByteDance built to serve TikTok's 1.5 billion users has been redirected to an entirely different task: designing molecules, where binding affinity to a protein target replaces user engagement as the metric that matters.

South Korea's response

The Chinese tech giant's entry into pharmaceuticals has sharpened a debate already underway in South Korea, where experts have long pointed to the country's lag in AI-driven drug discovery research. Between 2015 and 2024, South Korea produced a total of 1,016 scientific papers on the subject, putting it only ninth in the world. The Korea Health Industry Development Institute has launched a program that will channel 49.5 billion won by 2029 into preclinical AI model development and 40.4 billion won by 2027 into next-generation antibodies, so-called biobetters.

On July 1, 2026, at the sixth meeting of the science and technology cabinet, the government approved 52 projects submitted by 25 ministries out of 121 applications filed by 28 agencies. The winning projects will receive a combined 3,000 of the 10,000 GPUs secured by the government, with distribution starting in April 2026, phased in according to project timelines and reviewed monthly for resource usage. The largest allocations went to the Ministry of Trade, Industry and Energy's autonomous vehicle project (304 GPUs) and the Ministry of Science and ICT's biomedical program (208 GPUs).

What it means for the industry

A typical new drug development cycle takes more than a decade and costs about 2 trillion won, and the South Korean government aims to cut that in half. The country is home to roughly 850 pharmaceutical and biotech companies, which are expected to gain access to shared foundation models funded through these programs. Western giants are making similar bets: Isomorphic Labs, an Alphabet company, has signed deals worth 2.4 trillion won with Eli Lilly and 1.7 trillion won with Novartis, while Recursion Pharmaceuticals acquired Exscientia for roughly 1 trillion won.

Artificial intelligence is transforming into an independent research agent with autonomy at every stage of the drug development cycle - Pyo Joon-hee, director at the Korea Pharmaceutical and Bio-Pharma Manufacturers Association

Not everyone shares the unqualified enthusiasm. Kim Yong-woo, head of the Korea Health Industry Development Institute, notes that AI on its own only speeds up the identification of drug candidates, while clinical trials and regulatory approvals, which typically take 10-15 years after a molecule's discovery, remain a bottleneck no language model can bypass.

The wider race

ByteDance is not the only Chinese tech player testing pharmaceuticals as a new outlet for its AI infrastructure, Baidu, Tencent, Alibaba and JD are weighing similar moves, though each has chosen a different model of involvement. For Polish biotech companies and investors tracking the sector, the signal echoes one the semiconductor industry sent earlier: the capital and computing infrastructure of major internet platforms is starting to seep into fields long dominated by traditional pharmaceutical companies, and Asian governments are treating this as a cue to accelerate their own programs rather than wait on results from Western giants like Isomorphic Labs.

So far, none of Anew Labs' candidates has entered human clinical trials, and the company has not disclosed a timeline for regulatory submissions in the US or Europe. The next test will be whether the AI-designed IL-17 inhibitor and the other three programs in its portfolio clear preclinical validation at a pace that justifies the buzz surrounding their debut at scientific conferences in 2026.