About Foundry

Foundry provides a cloud-native platform offering on-demand access to NVIDIA GPUs across multiple clouds without long-term contracts. The platform integrates with Kubernetes, enabling AI developers to flexibly reserve compute or bid on spot capacity for training and inference workloads. This model allows users to match GPU supply precisely to demand while optimizing costs through usage-based billing and resalable reservations.

<problem>AI developers and researchers need high‑performance GPU resources for training, fine‑tuning, and inference, but acquiring and managing on‑premise NVIDIA GPU clusters is costly, inflexible, and often requires long‑term contracts that do not match unpredictable workload demand.</problem> <solution>Foundry delivers a cloud‑native platform that provides on‑demand access to NVIDIA GPUs without the need for multi‑year contracts. Users can reserve compute for as little as three hours or bid on spot capacity, allowing them to match supply with workload intensity and control costs. The platform integrates natively with Kubernetes and offers an API for programmatic scaling, so developers can orchestrate distributed training and inference pipelines without managing infrastructure. Pricing is algorithmically set to market rates, and reserved instances can be resold when idle, further improving utilization. By aggregating multi‑cloud GPU, CPU, and storage resources into a single “omnicloud,” Foundry simplifies procurement and provides transparent, usage‑based billing.</solution> <features> - On‑demand access to NVIDIA H100, A100, A40, and A5000 GPUs with no contract commitments - Flexible reservation model: short‑term (as low as 3 hours) reserved instances and spot bids - Resellable reserved instances to monetize unused capacity - Native Kubernetes integration and API for automated scaling and orchestration - High‑performance networking (up to 3200 Gbps InfiniBand) for distributed training - Batch SDK and batch inference API for asynchronous workloads on spot capacity - Transparent, algorithmic market‑rate pricing across multi‑cloud resources - Enterprise‑grade security (SOC 2 Type II, HIPAA options) and co‑located storage with no egress fees </features> <target_audience>Primary customers are AI engineers, researchers, and data scientists at AI‑native startups, research labs, universities, and enterprise teams that require scalable, cost‑effective GPU compute for model development and production inference.</target_audience> <revenue_model>Foundry generates revenue through usage‑based pricing, charging customers per GPU‑hour at market‑determined rates for both reserved and spot instances.</revenue_model> <traction>As of March 2024, Foundry reported eight‑figure annual revenue and a valuation of $350 million after emerging from stealth. Its customer base includes LG, KKR, Stanford, MIT, Carnegie Mellon University, Arc Institute, Harvard Data Science Institute, and other leading research institutions.</traction> <sources> - https://fortune.com/2024/03/21/foundry-stealth-launch-350-million-valuation - https://mlfoundry.com/blog/restoring-the-promise-of-the-public-cloud-for-ai - https://mlfoundry.com/blog/introducing-foundry - https://www.sequoiacap.com/article/partnering-with-foundry-ai-compute-on-demand - https://finance.yahoo.com/news/exclusive-foundry-comes-stealth-350-115151375.html - https://mlfoundry.com/ </sources>

What does Foundry do?

Where is Foundry located?

Foundry is based in Palo Alto, United States.

When was Foundry founded?

Foundry was founded in 2022.

How much funding has Foundry raised?

Foundry has raised 80000000.

Location

Palo Alto, United States

Founded

2022

Funding

80000000

Employees

40 employees

Major Investors

Sequoia Capital, Lightspeed Venture Partners, Redpoint, Microsoft Ventures (M12), Conviction, NEA

Find Investable Startups and Competitors

Search thousands of startups using natural language

AI voice (2021+)underground pipe robots energy flexibility software

Start Searching

Foundry

⚠️ AI-generated overview based on web search data – may contain errors, please verify information yourself! You can claim this account with your email domain to make edits.

Executive Summary

mlfoundry.com 3K+

Crunchbase

Founded 2022 – Palo Alto, United States

Funding

Estimated Funding

$50M+

Major Investors

Sequoia Capital, Lightspeed Venture Partners, Redpoint, Microsoft Ventures (M12), Conviction, NEA

Team (40+)

No team information available.

Company Description

Problem

AI developers and researchers need high‑performance GPU resources for training, fine‑tuning, and inference, but acquiring and managing on‑premise NVIDIA GPU clusters is costly, inflexible, and often requires long‑term contracts that do not match unpredictable workload demand.

Solution

Foundry delivers a cloud‑native platform that provides on‑demand access to NVIDIA GPUs without the need for multi‑year contracts. Users can reserve compute for as little as three hours or bid on spot capacity, allowing them to match supply with workload intensity and control costs. The platform integrates natively with Kubernetes and offers an API for programmatic scaling, so developers can orchestrate distributed training and inference pipelines without managing infrastructure. Pricing is algorithmically set to market rates, and reserved instances can be resold when idle, further improving utilization. By aggregating multi‑cloud GPU, CPU, and storage resources into a single “omnicloud,” Foundry simplifies procurement and provides transparent, usage‑based billing.

Features

On‑demand access to NVIDIA H100, A100, A40, and A5000 GPUs with no contract commitments

Flexible reservation model: short‑term (as low as 3 hours) reserved instances and spot bids

Resellable reserved instances to monetize unused capacity

Native Kubernetes integration and API for automated scaling and orchestration

High‑performance networking (up to 3200 Gbps InfiniBand) for distributed training

Batch SDK and batch inference API for asynchronous workloads on spot capacity

Transparent, algorithmic market‑rate pricing across multi‑cloud resources

Enterprise‑grade security (SOC 2 Type II, HIPAA options) and co‑located storage with no egress fees

Target Audience

Primary customers are AI engineers, researchers, and data scientists at AI‑native startups, research labs, universities, and enterprise teams that require scalable, cost‑effective GPU compute for model development and production inference.

Revenue Model

Foundry generates revenue through usage‑based pricing, charging customers per GPU‑hour at market‑determined rates for both reserved and spot instances.

Traction

As of March 2024, Foundry reported eight‑figure annual revenue and a valuation of $350 million after emerging from stealth. Its customer base includes LG, KKR, Stanford, MIT, Carnegie Mellon University, Arc Institute, Harvard Data Science Institute, and other leading research institutions.

Sources:

1 2 3 4 5 6