Foundry, a rising cloud provider founded by alumni from Google DeepMind’s core Deep Learning team, has unveiled the Foundry Cloud Platform. This cutting-edge platform serves as a real-time market and orchestration engine for GPU compute, designed to streamline AI infrastructure access, reduce operational complexity, and enhance cost efficiency by up to 6x. With this launch, Foundry aims to democratize AI development and accelerate global innovation.
Addressing the GPU Compute Challenge
The surge in AI development has turned GPU servers into highly sought-after commodities, straining traditional public clouds and prompting massive investments by tech giants and startups to secure necessary hardware. However, the current GPU market is plagued by inefficiencies, including long-term contracts and underutilized capacity, which hinder broader access to crucial compute resources.
- Inefficiency in the GPU Market: “The GPU compute market is one of the most inefficient commodity markets in history, directly limiting critical AI innovations,” says Jared Quincy Davis, founder and CEO of Foundry. “Foundry Cloud Platform addresses this by aggregating and redistributing idle compute capacity, enabling faster breakthroughs and better GPU investment returns.”
Key Features of Foundry Cloud Platform
- Aggregated Compute Pool: Offers a dynamically-priced pool of GPU capacity optimized for various AI workloads.
- Resellable Reserved Instances: Provides self-serve access to short-term GPU reservations, allowing customers to reserve clusters for as little as three hours and resell idle capacity for additional credits.
- Spot Instances: Unreserved and relisted compute is available for bidding, ideal for interrupt-tolerant tasks such as model inference and hyperparameter tuning.
- Dynamic Pricing and Capacity Management: Utilizes auction theory to adjust market-driven prices based on real-time supply and demand, increasing overall GPU capacity to stabilize prices.
- Kubernetes Workload Orchestration: Integrates with Kubernetes to automate the scheduling of reserved and spot instances, optimizing performance and minimizing latency during traffic spikes.
Success Stories and Impact
- Infinite Monkey: An AI startup focused on AGI development, which leverages Foundry Cloud Platform to access state-of-the-art GPUs on-demand. “We made actionable discoveries in hours, not weeks,” says Matt Wheeler, Research Engineer at Infinite Monkey, praising the platform’s flexibility and cost-effectiveness.
- Arc Institute: A nonprofit researching complex diseases, including cancer and neurodegeneration. “Foundry delivers exactly the compute we need, when we need it, without procurement friction,” notes Patrick Hsu, Co-Founder and Core Investigator at Arc Institute.
Commitment to Security
Foundry has achieved SOC 2 Type II certification, ensuring high standards for security and compliance to protect customer data.
