Foundry provides an orchestration platform that enables AI developers to access NVIDIA GPU clusters on-demand, facilitating training, fine-tuning, and inference without long-term contracts. The platform addresses the challenge of unpredictable compute needs by offering flexible pricing options, including reserved and spot instances, ensuring reliable performance for critical workloads.
Funding
$80M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
MVN+1Founders
Product
Problem
AI developers often face challenges in accessing and managing NVIDIA GPU clusters for training, fine-tuning, and inference, particularly with unpredictable compute demands and the limitations of long-term contracts. This can lead to inefficient resource utilization and increased costs due to idle capacity or difficulty in scaling resources on demand.
Solution
Foundry provides a flexible compute platform that allows AI developers to access NVIDIA GPUs on-demand, eliminating the need for long-term contracts and simplifying resource management. The platform offers both reserved and spot instances, enabling users to optimize cost and performance based on workload requirements. Reserved instances guarantee capacity for critical workloads and can be resold when not in use, while spot instances provide cost-efficient compute for flexible training and inference tasks. By offering programmatic scaling via API and integration with Kubernetes, Foundry streamlines workload orchestration and allows developers to focus on AI development rather than infrastructure management.
Target Audience
The primary target audience includes AI engineers, researchers, and scientists who require flexible and scalable access to GPU compute resources for training, fine-tuning, and inference tasks.
Features
- On-demand access to NVIDIA H100s, A100s, A40s, and A5000 GPUs without contracts
- Flexible pricing options with reserved and spot instances to optimize cost and performance
- High-performance networking with up to 3200Gbps InfiniBand for distributed training
- Native Kubernetes integration for simplified workload orchestration and horizontal scaling
- API for programmatic scaling and instance management
- Co-located storage with no ingress or egress fees
- Support for custom scripts on startup and SSH access
- SOC2 Type II certification and HIPAA compliance options for enterprise-grade security