Run:ai’s Atlas platform virtualizes and orchestrates AI workloads with a focus on maximizing GPU utilization and abstracting hardware from developers and IT teams. Developers can interact with pooled resources through standard tools like Jupyter notebooks, while IT gains better visibility into resource usage. The company emphasizes low-latency, high-throughput inference scheduling, claiming near-100% GPU utilization for inference servers. Run:ai says its annual recurring revenue grew 9x in the last year and its staff more than tripled; Insight Partners noted the company has almost quadrupled its customer base. The company plans to use new funding to grow its team and said it will consider strategic acquisitions to accelerate time to market, while preferring in-house development. Run:AI builds a Kubernetes-based platform that virtualizes and orchestrates AI compute and GPU workloads across cloud and on-premises environments. Its scheduler pools GPU resources and allows teams to share capacity while giving IT better visibility into utilization. Customers across automotive, finance, defense, manufacturing and healthcare report average GPU utilization increases from 25% to 75%. The company plans to use the new funding to triple its development team and accelerate an aggressive roadmap around virtualizing AI workloads. Run:AI also intends to expand sales and marketing into additional industries and markets. Financially, Run:AI raised a $30 million Series B led by Insight Partners, bringing total funding to $43 million. Run:AI offers a high-performance compute virtualization and acceleration layer for deep learning, designed to speed neural network training and enable development of much larger AI models. The platform uses advanced mathematics to split large models into multiple smaller models that run in parallel, which helps bypass GPU RAM limits. Its technology bridges the gap between data science and computing infrastructure by abstracting and optimizing hardware resources for deep learning workloads. Run:AI was founded in 2018 and operated in stealth until this raise. The company has signed several early customers internationally and has established a US office.

Total raised

$118M

Funding rounds

4

Latest round

Series C

Latest activity

Mar 2022