Cedana
Cedana is a B2B company specializing in infrastructure solutions that enhance cloud compute efficiency by enabling real-time management and migration of compute workloads across various vendors, significantly reducing costs and latency.
Services
Cedana enables real-time save, move, and restore for compute workloads, aiming to transform cloud compute into a real-time, arbitraged commodity. Key services include expanding GPU capacity, increasing reliability, reducing latency by up to 10x, and optimizing CPU/GPU utilization by up to 5x, leading to significant cost savings. The solution applies to AI training and inference, HPC, data analysis, dev tools, and infrastructure, and offers features like pause/migrate/resume across GPUs and real-time migration between VMs.
Mission and Vision
Cedana's vision is to democratize cloud compute globally by removing barriers between builders and compute resources while increasing resource efficiency. The company aims to move compute in real-time across the globe, creating a global compute exchange. This approach addresses bottlenecks in access to compute for research and progress, allowing for more efficient use of computational resources.
Key Features of Cedana's Solutions
Cedana offers several key features designed to optimize compute management: automatic job movement across instances and vendors without code modification, preference settings for cost and performance, assignment of reliability and performance requirements from individual jobs to entire tasks, and targeting price/performance criteria based on workload. The company also ensures failure resilience, allowing work to continue with no wasted or idle hours.
Y-Combinator Involvement
Cedana is part of the Y-Combinator batch S23. Through this involvement, the company gains access to a network of industry leaders and experts, fostering innovation and growth within the tech ecosystem. Being a Y-Combinator company provides Cedana with opportunities for mentorship and funding, aiding in the realization of its objectives.