Deciphex improves AI training for pathology with Oracle AI Infrastructure
Medical technology startup transforms digital pathology workflows by training AI models on high performance OCI AI Infrastructure.
“Our work with OCI is a game-changer, because it provides the reliability, performance, and scalability we need to train AI models at scale, helping clinicians make faster, more accurate diagnoses and accelerating pharma drug discovery. With OCI’s robust infrastructure, we’re now training models with 2X the data while halving our costs. This allowed us to push the boundaries of digital pathology and continue to innovate for better patient outcomes.”
Deciphex, an Irish medical technology company, set out to address the global shortage of diagnostic clinicians amid rising patient demand. The company’s goal was to expand access to high-quality pathology services and reduce diagnostic delays using AI-powered workflows and one of the world’s largest annotated digital pathology repositories. However, with its previous cloud provider, Deciphex faced challenges in scaling GPU resources efficiently. After migrating to Oracle AI Infrastructure with GPU-enabled bare metal instances, high-throughput VMs, and RDMA-enabled networking, the company doubled its price-performance ratio while reducing total cost of ownership. Looking ahead, the company plans to expand its footprint on OCI to accelerate large-scale foundation model development. By combining AI and digital pathology at scale, Deciphex aims to reduce backlogs, support faster diagnoses, and accelerate clinical research pipelines.