Foundry, an emerging cloud provider founded by alumni from Google DeepMind's core Deep Learning team, is debuting the Foundry Cloud Platform, the real-time market and orchestration engine for GPU compute that democratizes access to the necessary infrastructure to build and deploy AI. Built from the ground up to solve for inaccessibility to GPU compute—typically reserved for well-funded labs—the Foundry Cloud Platform serves to reduce the operational complexity of AI development and enhance compute cost efficiency by up to 6x, according to the company.
While the demand for more and more AI-powered technology is based on its promise of ease and efficiency, supporting these endeavors requires significant firepower. Outpacing the capacity of traditional public clouds, mega tech corporations and AI startups alike are investing significant capital in securing the necessary hardware to support AI initiatives. Between long-term, pricey contracts and a lack of resources to support massive GPU requirements, many enterprises are fundamentally incapable of joining the race to implement AI, according to Foundry.
Born from this sense of AI inaccessibility, Foundry Cloud Platform enables AI teams at every scale to access GPU compute in an efficient, approachable manner. Tapping into the underutilized GPUs deemed incapable of supporting the unique compute requirements of AI, Foundry “addresses this market failure by aggregating and redistributing idle compute capacity to enable faster breakthroughs while improving return on GPU investments,” according to Jared Quincy Davis, founder and CEO of Foundry.
“Foundry is opening the door for more organizations to build with AI by offering unmatched simplicity and cost efficiency,” continued Davis. “Much of the cutting-edge development that could previously only be conducted by labs like OpenAI and DeepMind will now be obtainable by others as Foundry makes GPU compute more accessible and affordable.”
Employing auction theory—or the method of setting dynamic prices based on supply and demand—Foundry ensures that its customers are getting the best price for their workloads, according to Davis. Additionally, Foundry Cloud Platform offers the following efficiency boons:
- Resellable reserved instances for as little as 3 hours, replacing fixed, long-term contracts with the ability to reserve interconnected clusters whose idle capacity is sellable to drive greater cost efficiency
- Unreserved and relisted compute available as spot instances, a pool of buffer capacity, or “healing buffer,” that are purchasable at prices up to 90% lower compared to reserved compute
"With Foundry Cloud Platform, we made actionable discoveries in hours, not weeks," said Matt Wheeler, research engineer at Infinite Monkey, an AI startup developing architectures for AGI. "When we believe we could benefit from additional compute, we just turn it on. When we need to pause to study our results and design the next experiment, we turn it off. Because we aren't locked into a long-term contract, we have the flexibility to experiment with a variety of GPUs and empirically determine how to get the best price-performance for our workload."
Foundry Cloud Platform is also offering Kubernetes workload orchestration, serving to abstract the scheduling required to spread workloads across several nodes. By utilizing Kubernetes clusters through Foundry Cloud Platform, AI teams are empowered to optimize price-performance and mitigate inference latency during traffic spikes through rapid, horizontal capacity scaling, according to the company.
“The release of Foundry Cloud Platform is just the beginning,” said Davis. “Over the coming months, we will be releasing additional features to make the platform even more accessible and useful to a wider range of AI developers. Our vision is to orchestrate the world’s compute capacity to render it universally accessible, spurring innovation in AI.
Foundry has also achieved SOC 2 Type II certification, ensuring that its platform adheres to the highest standards of security and compliance for customer data.
To learn more about Foundry Cloud Platform, please visit https://mlfoundry.com/.