PALO ALTO — Foundry has introduced the Foundry Cloud Platform, aiming to redefine how AI practitioners access and manage compute resources. This innovative platform offers state-of-the-art GPU compute with unmatched flexibility and reliability, enabling AI developers to train, fine-tune, and deploy models without the long procurement processes or rigid contracts that currently dominate the landscape.
The platform allows users to reserve GPUs elastically for as little as three hours while ensuring reliability through automated failover systems. Foundry’s orchestration platform mitigates hardware failures by proactively replacing failed nodes, offering customers seamless compute availability.
Tackling AI Infrastructure Challenges
AI practitioners often face hurdles like GPU failures and opaque procurement processes, diverting focus from core research. Foundry aims to simplify this by creating abstraction layers that eliminate these pain points. Unlike traditional public cloud providers, Foundry’s infrastructure is designed to meet the specific demands of modern AI workflows, ensuring smooth operations and optimal efficiency.
Cost-Effective and Elastic Computing
In addition to standard reserved compute, Foundry offers access to preemptible spot instances at prices up to 20 times better than traditional GPU clouds. These instances are ideal for workloads like live inference, batch processing, and hyperparameter optimization. Foundry also introduces features like hosted Kubernetes clusters and persistent storage auto-mounting to make these instances easier to manage.
The platform’s design fulfills the original promise of cloud computing: delivering cost-effective scalability and rapid task completion. Foundry enables dynamic, bursty AI workflows to leverage elastic compute efficiently, bypassing the rigid long-term contracts that hinder flexibility in traditional cloud solutions.
Advancing Cloud Elasticity for AI
Foundry addresses a critical gap in the AI cloud landscape, delivering elasticity that scales to meet the unpredictable demands of modern AI workloads. With significant technical breakthroughs, the platform ensures that companies can access the compute they need, when they need it, without overpaying or compromising efficiency.
Deliberate Rollout
Foundry is currently in the process of a phased rollout, with access being granted selectively to ensure a high-quality experience for early users. The company is scaling its capacity to meet the growing demand for its platform and plans to expand availability in the coming months.
For more information or to request access, visit the Foundry Cloud Platform website.
