SK Biopharmaceuticals has secured large-scale AI computing infrastructure by introducing 128 of NVIDIA’s latest artificial intelligence (AI) accelerators, the B200. The company plans to develop AI models specialized for drug discovery in-house and accelerate the adoption of AI across its research operations.
SK Biopharmaceuticals announced on September 2 that it had introduced 128 B200 GPUs based on NVIDIA’s Blackwell architecture. The 128 B200 GPUs deliver 576 petaflops (PFLOPS) of computing performance per second based on FP8, equivalent to approximately 0.58 exaflops.
The infrastructure was secured through a cloud subscription model via “Haein,” a B200-based data center cluster built by SK Telecom. The company plans to use high-performance GPUs to develop and operate AI models tailored to research objectives and data characteristics more quickly.
SK Biopharmaceuticals plans to focus the computing resources it has secured on advancing “TR-AX” (Translational Research-AI Transformation). TR-AX integrates AI into translational research, which connects preclinical research findings with clinical outcomes, with a focus on enhancing the value of drug candidates in the preclinical and early clinical stages.
In particular, the company plans to apply foundation models specialized for the biotech and healthcare sectors to actual research, enhancing its analytical capabilities for diseases, targets and compounds. It will expand the use of AI throughout the drug development process, from candidate discovery to research decision-making support.
In addition to research and development, the company will build a dedicated large language model (LLM) for internal operations. With sufficient training and inference resources, it aims to shorten research cycles involving experiments, analysis and retraining while improving productivity and decision-making efficiency.
SK Biopharmaceuticals also plans to open its AI infrastructure to joint research with external companies, research institutions and universities rather than using it solely for its own research. Through this initiative, the company plans to expand its infrastructure into an AI research platform supporting open innovation and industry-academia-research collaboration.
The company will also strengthen security and governance for research data. By managing sensitive biotech and healthcare data within a controllable infrastructure, it aims to enhance security and establish a foundation capable of responding to future regulatory and compliance requirements.
SK Biopharmaceuticals is also considering ways to commercialize its AI infrastructure and technologies in the future. The company plans to expand joint research and strategic partnerships and, over the long term, identify new business models such as AI-based research services.
SK Biopharmaceuticals CEO Lee Dong-hoon said, “Large-scale in-house AI infrastructure is a critical foundation that will determine our competitiveness in future drug development. Based on the B200 infrastructure, we will further strengthen TR-AX and maximize the use of AI throughout the entire drug development process to increase the speed and efficiency of our R&D.”
Meanwhile, SK Biopharmaceuticals is also pursuing expansion into new modalities. Key areas include radiopharmaceutical therapy (RPT) and targeted protein degradation (TPD). RPT, in particular, has significant strategic value because the company can leverage infrastructure and raw-material sourcing capabilities across the SK Group.