Chinese AI Healthcare Accelerates: Big Tech Spins Off Drug R&D Units, AI Agents Enter Hospitals

Jun 15, 2026 18:43 CST Updated 18:43

Tencent Healthcare

Developer of Digital Healthcare Products

BioMap

Developer of Innovative Drug R&D Platform

Tencent

Internet Comprehensive Service Provider

On June 5, 2026, Tencent Healthcare released the “Tong Xiao’an” outpatient intelligent agent, co-developed with the Affiliated Hospital of Nantong University; just five days later, ByteDance’s AI drug discovery business initiated a spin-off for independent financing.

At one end, there is an upstream breakthrough aimed at tackling the “double-ten dilemma” in new drug R&D—major pharmaceutical companies are spinning off units for financing, while platform-based enterprises have secured large-scale international business development (BD) deals. At the other end, downstream penetration is deepening as AI agents enter top-tier (Grade 3A) hospitals. Within a single week, China’s AI healthcare sector has simultaneously pressed the “accelerator button” on both upstream and downstream fronts, launching substantive breakthroughs into deeper waters through scenario-specific and track-specific approaches.

Spin-off Financing by Major Tech Companies
International Orders Secured

On the evening of June 10, 2026, news that ByteDance’s AI drug discovery business line was undergoing a spin-off and initiating independent financing sent ripples through the long-quiet primary market for AI-driven pharmaceuticals. According to multiple media reports, ByteDance will retain controlling interest in the newly formed entity, which will be led by Liu Kai, the former head of ByteDance’s AI drug discovery team. The core team comprises approximately 50 members, including AI4S (Artificial Intelligence for Science) algorithm specialists and seasoned pharmaceutical experts. Core algorithms, protein structure prediction models, and existing pipeline assets will be transferred in their entirety to the new entity, which will continue to access computing power support from Volcano Engine. This spin-off marks ByteDance’s first foray into the industrialization of AI4S initiatives. The move is driven by the biomedical industry’s characteristics—long development cycles, high capital intensity, and stringent regulation—with independent operations enabling more agile decision-making, facilitating engagement with industry capital, and attracting top-tier talent.

Established in 2021, the team has accumulated five years of expertise, covering the full spectrum of functions from foundational model research to early-stage pipeline validation. In 2023, the corporate entity Anew Therapeutics was incorporated in Singapore, with team members distributed across Shanghai, San Francisco, and Singapore. In terms of key technologies, the team has continuously iterated on its protein structure prediction models—launching the Protenix and Seedfold models in 2025, and upgrading them to version 2 (v2) in 2026. In March of the same year, in collaboration with Tsinghua University, the team released AnewOmni, a full-atom molecular generation model. It is reported that an inhibitor targeting autoimmune diseases such as psoriasis has demonstrated differentiated advantages in preclinical studies and is currently in the lead optimization phase.

ByteDance is not an isolated case. BioMap, a subsidiary of Baidu, filed a confidential listing application with the Hong Kong Stock Exchange in March 2026; Alibaba Health has established a dual-wheel strategy of “Full-Stack Pharmaceuticals + Medical AI”; JD Health is building full-scenario applications leveraging its “Jingyi Qianxun” large language model; Tencent has disclosed patents for GLP-1 weight-loss drugs designed by AI; and Huawei Technologies Co., Ltd. formed its healthcare and medical business unit in 2025.

Meanwhile, XtalPi Holdings, a leading AI-driven drug discovery company listed on the Hong Kong Stock Exchange, announced a major strategic partnership on the evening of June 9. The company has entered into an AI drug discovery collaboration with a globally renowned biopharmaceutical firm that possesses multiple commercialized products, focusing on a specific GPCR target to jointly develop a best-in-class oral small-molecule drug. This agreement builds upon a successful pilot phase, in which XtalPi integrated quantum physics with AI algorithms to achieve breakthrough hit rates, thereby validating its platform’s capability to address complex metabolic targets. Under the terms of the deal, the partner will pay an upfront fee and cover all early-stage R&D costs, while XtalPi is eligible for milestone payments across preclinical, clinical, and commercialization stages, as well as future sales royalties, bringing the potential total transaction value to over $400 million. This model, which ties near-term R&D revenue to long-term pipeline value, not only reduces the cost and risk for XtalPi in tackling high-barrier targets but also secures significant upside potential.

This collaboration marks another instance of a Chinese AI-driven pharmaceutical platform securing a high-value business development (BD) deal with a multinational pharmaceutical company, following similar achievements by leading firms such as Insilico Medicine. The closure of this series of international orders has, to some extent, bolstered market confidence in the commercial viability of AI-driven early-stage drug discovery. Wen Shuhao, Chairman of the Board of Directors of XtalPi, stated that this partnership represents another real-world validation of XtalPi’s “AI + Robotics” foundational R&D system on the global stage of top-tier biopharmaceutical innovation.

According to statistics from industry research firms, as of the first half of 2026, more than 170 AI-designed or optimized drug pipelines have entered clinical trials globally. Among these, over ten are advancing into Phase III trials, making 2026 the year with the most intensive release of clinical data in the history of AI-driven drug development. The capital market has responded clearly: Huashen Zhiyao completed a $787 million financing round; Insilico Medicine reached an $888 million collaboration agreement with Servier for anti-tumor drugs; and several platform-based companies have also secured new rounds of investment.

However, the other side of the coin is that AI currently primarily compresses the preclinical discovery cycle (shortening it from an average of 4–5 years to approximately 18 months), but remains unable to bypass the rigid time barriers inherent in clinical trials itself. Some early-stage AI drug discovery startups have been exposed as "swimming naked" during this round of market consolidation, due to excessive cash burn and a lack of substantive progress in their pipelines. Industry insiders note that major pharmaceutical companies spinning off units for independent financing, and specialized platforms securing large international deals, indicate that capital preference is increasingly concentrating on top-tier players with closed-loop capabilities and proven business development (BD) execution.

AI Agent Enters the Consultation Room
Accelerated Validation of Business Model

Unlike the high risks and long cycles associated with new drug development, AI applications in healthcare services—such as patient triage, pre-consultation inquiries, report interpretation, and follow-up management—feature shorter feedback loops and more direct user experience perceptions. At the Smart Healthcare Special Session of the Tencent Cloud AI Industry Application Conference held on June 5, 2026, Jiang Hailin, Director of the Information Center at the Affiliated Hospital of Nantong University, and Wu Zhigang, General Manager of the Tencent Healthcare User Platform, jointly disclosed operational data for “Tong Xiao’an,” an intelligent outpatient care agent co-developed by both parties.

After being integrated into the hospital’s WeChat official account, the system serves over 3,000 patients daily, with an intelligent triage accuracy rate of 98% and a patient issue resolution rate of approximately 95%. This AI agent enables AI applications to evolve from isolated, single-point interventions to hospital-wide, end-to-end collaborative synergy, embedding itself into every service touchpoint across pre-consultation, during-consultation, and post-consultation phases.

The six core AI capabilities of the Tencent Healthcare Open Platform—intelligent triage, intelligent Q&A, intelligent pre-consultation, report interpretation, medication planning, and patient recall—precisely address the pain points in hospital patient services, allowing for on-demand activation and flexible configuration of service scenarios and data permissions. The platform offers multiple integration methods, with individual modules deployable in as little as 1–2 days and the full suite of capabilities implemented within just 1–2 weeks. Moving forward, both parties plan to extend the open platform’s capabilities into follow-up visit management and continuous health services, “enabling Tong Xiao’an to evolve from an outpatient consultation agent into a comprehensive health steward.”

In the broader healthcare services market, the commercialization of AI applications is accelerating. SenseTime Healthcare has completed a new round of financing exceeding RMB 500 million, pushing its post-money valuation above USD 1 billion and entering the ranks of unicorns. The round was jointly invested by Raffles Healthcare Growth Fund, Singapore’s Temasek Capital, Hong Kong’s High-Talent Fund, Huagai Capital, Guoke Capital, Far East Horizon Capital, and Shanghai Lingang Fund, with existing shareholders such as Lenovo Capital following up.

Furthermore, United Imaging Intelligence completed a RMB 1 billion Series A financing round in June 2025, reaching a valuation of RMB 10 billion; Deepwise Medical covers the entire workflow of screening, diagnosis, treatment, management, research, and education with its “digital intelligence foundation,” holding 15 NMPA Class III certificates and nearing 200 million annual AI calls; Infervision focuses on AI medical imaging products, having accumulated nearly RMB 900 million in financing. In the A-share market, Winning Health explicitly stated that it will establish an AI Medical Division in 2026 to fully advance the AI integration of its WiNEX products; Yinkang Life released an intelligent agent for full-cycle tumor management in December 2025.

Intensive Approval of Class III Medical Device Certificates in the Field of Medical Imaging AI: Desheng Bio’s AI AutoVision® Chromosome Karyotype Auxiliary Diagnostic Software Receives the First Class III Certificate Derived from a Global Foundation Large Model for Medical Imaging; Dian Diagnostics’ Subsidiary Yice Technology and Libo Ping An Have Each Obtained Class III Certificates for Cervical Cytopathology AI.

However, beneath the boom lie hidden concerns. Industry experts argue that while AI agents enhance efficiency, issues such as data exfiltration risks and ambiguous liability definitions remain to be clarified. Currently, their role is largely complementary; a long road still lies ahead before they become indispensable “colleagues” for physicians.

A research report from CITIC Securities indicates that 2026 will be a year of greater certainty for the commercialization of AI in healthcare. Meanwhile, Everbright Securities points out that the core of future competition lies not in who has higher model parameters, but in who possesses exclusive, high-quality data and can achieve continuous iteration through business scenarios. Demand-driven scenarios determine willingness to pay, and the market values those who can truly bridge the "last mile" from technology to product.

Whether large tech giants are spinning off their internally incubated businesses into independent capitalization tracks, or specialized platforms are securing landmark international orders, various signs indicate that China’s AI healthcare sector is bidding farewell to pure conceptual narratives and entering the deep waters of scenario-specific validation and track-specific industrialization. However, longstanding challenges such as prolonged clinical translation cycles, data silos, and unclear business models continue to loom over every participant.

Investment in AI-driven drug discovery has shifted from backing conceptual hype to prioritizing engineering capabilities. Moving from isolated improvements to end-to-end synergy, AI’s role in the pharmaceutical industry is transitioning from a “concept premium” phase to a period of financial validation.

Layout: Li Yongshi

Proofreader: Chen Shuwen

Reviewed by: Ma Fei