This company offers cloud-based bare metal instances for compute-intensive workloads, providing the performance of physical machines with cloud elasticity. Their services include container services, distributed storage, and elastic GPU computing infrastructure.
Funding
Funding not disclosed
Founders
Product
Problem
Compute-intensive workloads, such as those found in AI/ML, often require specialized hardware and infrastructure that can be expensive and difficult to manage on-premises. Traditional cloud providers may not offer the optimal balance of performance, flexibility, and cost-effectiveness for these demanding applications. Scaling resources up or down can also be slow and inflexible, hindering innovation and time-to-market.
Solution
RunSun Cloud provides on-demand access to high-performance computing infrastructure, including bare-metal servers with the latest NVIDIA GPUs. Their Kubernetes-native cloud platform allows users to quickly deploy and manage GPU-accelerated containers, while distributed storage and high-speed networking ensure optimal performance. RunSun Cloud's infrastructure is designed to deliver the performance of physical machines with the elasticity and scalability of the cloud, enabling users to scale resources up or down in seconds. This allows teams to accelerate model training, reduce latency, and achieve significant cost savings compared to traditional cloud providers.
Target Audience
RunSun Cloud targets machine learning teams, AI researchers, and other users with compute-intensive workloads who require high-performance GPU resources and flexible infrastructure.
Features
- On-demand access to NVIDIA GPUs, including H100, A100, and GH200
- Bare-metal cloud instances for maximum performance and configurability
- Container service for easy deployment and management of GPU-accelerated containers
- Distributed, fault-tolerant storage with triple replication
- High-speed networking with InfiniBand and private VPCs
- Multi-GPU instances with 1x, 2x, 4x, or 8x GPUs
- Pre-configured software for machine learning, including PyTorch, TensorFlow, CUDA, and Jupyter
- Support for large-scale GPU clusters, ranging from 64 to 60,000 GPUs