As artificial intelligence’s competitive landscape evolves, the second half of the large model race is zeroing in on real-world applications with new depth. The shift is evident across sectors like healthcare, tourism, big data, and urban governance, where specialized AI models are inching closer to commercialization. The industry’s ecosystem is now unmistakably entering a new chapter.
One player stands out. Before the recent AI surge, a small startup named BioMap had already zeroed in on life sciences. Founded in 2020 by Baidu CEO Robin Li and former Baidu Ventures CEO Wei Liu, BioMap launched with a specific aim: to build foundational AI models in life sciences from the ground up.
By 2021, BioMap had developed “xTrimo,” an AI model with over a trillion parameters, touted as the world’s first large-scale multimodal model for life sciences.
BioMap’s approach diverges from the common practice of repurposing language models for biological data. Instead, it targets molecular sequences—amino acids, nucleotides—using transformer architecture to uncover hidden patterns. This enables BioMap to model not just individual proteins but their interactions within cells and broader cellular systems, tackling complex challenges in drug discovery and life sciences research.
In October, BioMap launched its latest model, xTrimo V3, which expands on V2’s achievements. The new model has 2.1 trillion parameters, covering not only proteins but also DNA, RNA, and four other life sciences modalities. With enhanced architecture and new data, xTrimo V3 represents a significant leap forward in BioMap’s foundational model capabilities.
In a media briefing after the conference, BioMap’s CEO Wei Liu, alongside president Yongfu Deng and vice president of technology Xiaoming Zhang, spoke with media outlets, including 36Kr, to outline the startup’s four-year journey and its strategic approach to foundational AI models in life sciences.
Tackling big questions in life sciences
This year’s Nobel Prize in Chemistry highlighted the “AI for Science” (AI4S) paradigm, showing how AI-driven research is transforming fields like drug discovery by reducing development cycles and costs. The AI-driven pharmaceutical market, now exceeding RMB 10 billion (USD 1.4 billion), underscores the momentum underway.
BioMap stands out in this crowded space. While many companies focus on enhancing existing drug development workflows, BioMap tackles problems that current data and simulation tools can’t yet solve. “Our goal isn’t a series of one-off technical milestones or academic papers,” Liu told 36Kr. “We’re applying AI to advanced, vertical scenarios in life sciences, creating systematic solutions that often combine multiple AI models.” In doing so, BioMap is tapping into a deep market appetite; pharmaceutical giants spend billions annually on experimental models, creating new ground for AI to replace or extend conventional tools.
Liu envisions AI models capturing 20–30% of drug development budgets in advanced research, bridging gaps in precision and cost-effectiveness that traditional methods miss. As life sciences research shifts toward rational design over natural compound screening, BioMap aligns with an industry on the edge of change.
Beyond pharmaceuticals: A full-scope life sciences model
Building foundational AI models for life sciences is just one part of BioMap’s trajectory. To address the demands of commercialization, BioMap initially concentrated on technical breakthroughs. But since late last year, the company has sped up its commercial efforts, underscored by a billion-dollar partnership with Sanofi. This collaboration is unique, focusing not on a specific drug but on BioMap’s foundational model—a rare approach that highlights the increasing value of model-based AI in pharma.
According to Liu, BioMap’s foundational investments have translated into market recognition. The company now serves over 300 clients worldwide, with USD 2 billion in contract value. Its clients include multinational pharmaceutical companies, contract research organizations (CROs), startups, and research institutions.
Beyond drug development, BioMap’s models extend across life sciences. The company supports applications from general protein interactions to cell and gene therapies, synthetic biology, and more. BioMap has recently completed several enzyme generation projects for international clients, and a new partnership with DBN Group will bring agritech and biotech applications to China, focusing on biosynthesis, gene editing, and proteomics.
“Biomanufacturing has been slower to adopt new technology compared to biopharma,” Liu said, “but this lag represents an opportunity for us to introduce transformative technology.” In synthetic biology and biomanufacturing, Liu believes China’s potential could even outpace the US.
Pharmaceutical R&D remains BioMap’s primary focus. With global investment in drug development exceeding USD 1 trillion annually, BioMap’s AI models stand to boost efficiency, paving the way for more targeted treatments in personalized and rare disease medicine.
“Our view is that life sciences AI models are a long-term play,” Liu said. “We prioritize future-oriented goals, targeting research themes that will drive the life sciences industry forward by three to five years. By addressing high-impact, forward-looking challenges, we aim to create social value while also sharing in the economic benefits.”
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This article was written by Wang Fangyu for 36Kr.