Foundry rebrands as Mithril, announcing 550% platform growth and new inference service that cuts large-scale AI data processing costs by more than 50%

PALO ALTO, Calif., Aug. 4, 2025 /PRNewswire/ -- Mithril announced today its rebrand from Foundry, reflecting its mission to simplify access to distributed AI compute. The Mithril brand embodies the platform's mission: transforming powerful but traditionally rigid AI infrastructure into a fluid, universally accessible resource for AI practitioners.

Since launching in limited preview in August 2024, Mithril's platform usage has grown by over 550%. Today, the platform provides critical AI compute to customers including Cursor, LG AI Research, and Arc Institute. In parallel to the rebrand, Mithril is introducing its highly demanded batch inference service, which improves price-performance by 2–5x for large-scale AI workloads such as document analysis, content generation, and multimodal annotations.

"Traditional AI cloud procurement models mean that AI-native companies need to constantly shop around for the best compute prices and often need to buy more than what they actually need," said Jared Quincy Davis, founder and CEO of Mithril. "By building on our omnicloud platform, Mithril customers have been able to massively simplify getting access to the compute they need. Rather than spending effort negotiating with every cloud, they can provision compute for a few hours or multiple months through a single platform with transparent, market-based pricing."

While the GPU market, driven by the rapid adoption of AI, is set to reach nearly $230 billion by 2030, the cost per unit of performance for GPU compute is declining rapidly. Mithril's omnicloud platform is designed to harness this new paradigm, virtualizing capacity across multiple providers and distributing it through a fluid market with transparent pricing. By aggregating capacity in this way, Mithril is able to provide customers with the best economics for AI compute without bespoke negotiations or long-term contracts.

Building on this foundation, Mithril today unveiled its new batch inference API, purpose-built to orchestrate large-scale inference workloads across its omnicloud platform. The service enables customers to run compute-intensive tasks—such as document summarization, video transcription, multimodal annotation, and synthetic data generation—at a fraction of the traditional cost. By leveraging Mithril's distributed compute marketplace, the batch inference API delivers 2–5x better price-performance than conventional inference services, while eliminating the need for customers to manage complex infrastructure or negotiate access to scarce GPU capacity. With simple, token-based pricing and support for both off-the-shelf and custom models, teams can submit jobs of virtually any scale and rely on Mithril to handle orchestration, scaling, and infrastructure reliability.

With the addition of batch inference, Mithril extends its core mission: to make compute as universally available and accessible as electricity. Whether provisioning GPUs for foundational research or scaling inference workloads across enterprise datasets, Mithril's omnicloud platform provides AI teams with the flexibility, efficiency, and simplicity they need to push the boundaries of what's possible.

"With Mithril, we trained the first textless conversational audio base model for under $200,000, instead of the $10 million or more we would have needed to raise to reserve multi-year capacity," says Devansh Pandey, co-founder of Standard Intelligence, a startup building foundational models towards AGI. "Burst compute, which Mithril excels at, gives even small startups a leg up against incumbents."

"The tremendous demand for AI compute capacity and the large fraction of idle time makes sharing a perfect solution," said Paul Milgrom, a Stanford economics professor, Nobel laureate, and Mithril advisor who also has helped Google with its market-leading ad platform. "Mithril's innovative market is the right approach to making that happen."

About Mithril
Mithril is the omnicloud platform for AI compute, designed to give every AI practitioner the freedom to access and scale the resources they need, when they need them. By aggregating distributed capacity from leading cloud providers and data centers into a single omnicloud marketplace, Mithril maximizes price-performance while eliminating the complexity of managing infrastructure. Trusted by AI leaders like Cursor, LG AI Research, and Arc Institute, Mithril enables teams to focus on advancing AI rather than negotiating for compute. To learn more visit: mithril.ai

View original content to download multimedia:https://www.prnewswire.com/news-releases/mithril-arrives-expanding-omnicloud-ai-compute-platform-trusted-by-cursor-lg-ai-research-arc-institute-and-others-302520349.html

SOURCE Mithril