XtalPi

XtalPi is a technology company that aims to transform drug discovery and development processes in terms of speed and scale by integrating quantum physics-based computational methods, artificial intelligence, and robotic automation. The company distributes resources tailored to the needs of its clients and partners through its operations in China and the United States. A recursive feedback loop established between dry (in silico) and wet laboratories enables mutual enhancement of modeling on cloud supercomputing and autonomous experimental infrastructure.

Founding

XtalPi was founded in 2015 in Boston, Massachusetts by three MIT postdoctoral physicists. In its early stages, the company offered services in crystal structure prediction and solid-state research. The success of its crystal structure prediction platform, validated in a blinded trial by Pfizer, opened the door to industrial collaborations in solid-form selection and polymorph screening and contributed to solid-form studies related to nirmatrelvir (Paxlovid) during the COVID-19 pandemic. Between 2016 and 2024, the company established an artificial intelligence R&D center, high-accuracy force field (XFF) and free energy perturbation solution (XFEP) platforms, wet laboratories for solid-state and synthesis, autonomous laboratory prototypes, an experimental computational R&D center in Shenzhen, and a pharmaceutical center in Shanghai; developed the antibody discovery platform XupremAb; deployed scalable intelligent robotic network laboratories; opened a demo laboratory in Boston; and introduced laboratory automation products tailored for the Chinese market. In 2024, its automation solutions were expanded to sectors such as chemistry, traditional Chinese medicine, and materials development, and the company began trading on the Hong Kong Stock Exchange.

The company has over 1,000 employees, more than 160 granted patents, over 80 academic-pharmaceutical-biotech partners, and laboratory space exceeding 10,000 square meters. Its investor base includes global capital groups.

Technology

XtalPi’s platform integrates robotic automation with a hybrid Artificial Intelligence + Quantum Mechanics (AI+QM) approach. Cloud-based parallel GPU computations enable predictions of binding free energy, crystal structures, and drug-like properties using free energy perturbation and force field-based methods. Data from synthesis, screening, and characterization experiments conducted in wet laboratories are used for model calibration and active learning cycles.

XMolGen supports compound generation and virtual screening by leveraging big data and generative models based on synthetically accessible building blocks. The system generates ready-to-screen candidate sets for virtual docking and FEP workflows; it provides assessments of synthetic accessibility and drug-like property predictions. XFEP rapidly performs relative and absolute free energy estimations on a scalable cloud infrastructure, enabling evaluation of thousands of interactions per week. PatSight extracts chemical structures, activity, and ADMET data from patents and publications using optical methods, accelerating SAR analyses through CSV/SDF outputs and SMILES representations. XFF delivers high-precision force field calculations. XtalGazer supports crystal structure prediction and polymorph screening. The company’s proprietary model trials, referred to as “proteinGPT,” are directed toward protein engineering and biological large language model applications.

The digital chemistry pipeline accelerates project initiation through intellectual property screening and structure-data integration. The candidate molecule space is expanded using generative chemistry, followed by rapid candidate filtering based on FEP. Active learning, robotically focused library synthesis, and fast DMTA (design-make-test-analyze) cycles are applied in hit discovery, hit optimization, and candidate selection phases. Within structural biology, binding modes and polymorph relationships are elucidated using cryo-EM and micro-ED; CSP and solid-form selection processes are carried out in drug development.

The company has reported case studies including the discovery of potent non-covalent hits against GPX4 in 28 days, identification of novel leads in highly flexible allosteric regions using XMolGen + XFEP, template-based core elaboration for clotrimazole and ketoprofen, and analysis of structural differences and dynamic relationships among nirmatrelvir solid forms. Findings from cryo-EM on the mechanism of SARM1 support the structural biology pipeline.

Organization and Culture

The chair of the board and co-founder is Shuhao Wen, the CEO and co-founder is Jian Ma, and the chief innovation officer and co-founder is Lipeng Lai. The corporate culture is defined by principles of market orientation, responsiveness to customer needs, experimental validation, and data-driven process improvement.

The company has been featured on lists such as Forbes China Enterprise Technology 50, Deloitte China Technology Fast 50, CB Insights Digital Health 150, MIT Technology Review China “50 Smartest Companies,” and 36Kr “AI Top 50.”

Partnerships

XtalPi conducts collaborative projects with major pharmaceutical companies and biotech startups in areas including polymorph screening, crystal form selection, computational screening, DEL, HTS, and structural characterization. It offers its software components via cloud-based or on-premise deployment options and has adapted its robotic wet laboratory solutions for the chemistry and materials development industries.

XtalPi’s platform, which connects its hybrid AI+QM computational engine with robotic experimental infrastructure, institutionalizes data-driven recursion between dry and wet laboratories. The integration of patent text mining, generative molecular design, free energy calculations, structural biology, and solid-state science components into a single pipeline creates an R-G axis extending from target discovery to solid-form selection. This integrated structure aims to enhance speed, scale, and candidate quality in drug R&D processes.