Case Study: fal accelerates open-source AI model training and inference with Lambda

How fal scaled training and inference with Lambda

fal, a pioneering company in generative AI, faced significant challenges in accessing on-demand NVIDIA GPU resources for training and experimenting with its open-source AuraFlow model series. The initial process of negotiating short-term cloud contracts was slow and hindered their ability to iterate quickly. They needed a more agile infrastructure from a vendor like Lambda that could scale instantly without long-term commitments.

Lambda provided its 1-Click Cluster (1CC) solution, offering fal self-serve, on-demand access to GPU clusters. This allowed fal to accelerate development, releasing new model versions in weeks instead of months. The agility gained by using Lambda enabled fal to experiment freely, solidify its standing as a leader in the open-source community, and open new revenue streams through custom AI model development for customers.

Fal Ai

Batuhan Taskaya

Founding Engineer