Foundry, a Black-founded and led cloud computing service provider focused on AI workloads, has emerged out of stealth. According to Fortune, the company has raised $80 million in total seed and Series A funding and is now valued at $350 million. Foundry has more than eight figures in revenue with customers such as LG, KKR, Stanford, MIT, and Carnegie Mellon University.
“We’re announcing an exciting milestone at Foundry. We have closed $80 million in seed and Series A funding! With a mission of orchestrating the world’s compute capacity, making it universally accessible and useful, exciting things are on the horizon,” Jared Quincy Davis, Foundry founder and CEO wrote on LinkedIn.
Founded in 2022 by Jared Quincy Davis, Foundry is on a mission to ensure humanity maximizes the utility of computing power in the $200 billion AI market. The company seeks to address the “root-node problem” of infrastructure at the core of AI, rendering it more accessible at scale. It is building a new breed of public cloud built end-to-end for machine learning workloads which makes accessing AI compute resources as reliable and simple as ‘flipping a light switch’.
OpenAI’s ChatGPT launch and the corresponding explosion in the demand for AI compute with shortages highlighted the inefficiencies of the market.
The design of the current public cloud infrastructure was built to make serving and scaling web applications easy and doesn’t suit the emerging AI/Machine Learning (ML) workloads. CTOs and teams have had to become virtual supply chain managers to use today’s AI infrastructure which draws their attention away from actually building products. Foundry is therefore making data more accessible and useful by helping clients meet research deadlines and manage training runs faster and at less cost.
Davis believes that solving the challenges of modern AI infrastructure requires a fundamental reimagining from the ground up. He has assembled a cross-disciplinary team with experts in AI/ML, distributed systems design, hardware, traditional software, product, finance, and business. The team has experiences from DeepMind’s Core Deep Learning team, Microsoft, OpenAI, and Meta’s infrastructure teams, Stanford’s Future Data Systems Group, and X (formerly Twitter). Foundry intends to build the strongest technical architecture that is highly responsive to the salient practical and commercial needs of practitioners.
Related stories
While pursuing a Ph.D. in machine learning at Stanford, Davis had difficulty getting the Graphics Processing Unit (GPU) time he needed to train his models. He was simultaneously working as a research scientist on DeepMind’s Core Deep Learning team and fully understood the gaps between distributed systems for traditional cloud environments and what was needed for AI. A polymath with interests in AI, distributed systems, crypto and economics, Davis had a theory on how to train models far more quickly but had no practical way of doing so. He was inspired by AlphaGo, a computer program that plays the board game Go to create Foundry.
“Part of the reason I started working in deep learning was I saw AlphaGo from DeepMind and I thought that technique, if extended to real world problems, can be pretty transformative. But there are some limitations in that initial approach that made it not quite viable for problems at scale. So, I actively started working on filing those gaps,” Davis said on how a games computer program inspired him to create Foundry.
By solving the problem he once faced as a student and configuring models to run on distributed hardware, Davis has opened up a massive supply of underutilized enterprise computing. His radical new approach to cloud computing workload orchestration comes from the realization that the AI compute ecosystem is restrained not by under-supply, but by under-utilization.