
The most efficient ML inference platform
Our investment in Deep Infra

Felicis backed DeepInfra from the very beginning, joining the company's seed round. Felicis co-led the $18 million Series A in 2025, and again participated in the $107 million Series B in 2026.
Founders
Headquarters
Stage
Domain
Status
DeepInfra is building the inference layer AI companies run on — an AI inference cloud that lets developers deploy over 100 machine learning models spanning text, speech, image, video, and embeddings through simple, cost-effective APIs. The company runs its own inference-optimized infrastructure across secure US-based data centers, with a zero-retention policy and SOC 2 and ISO 27001 certification, giving customers a fast, secure alternative to building and managing GPU infrastructure themselves. For customers who want dedicated hardware, DeepInfra also offers DeepCluster, its own NVIDIA GPU clusters. Its founding team scaled the backend of imo, a messaging app with over 200 million monthly users, before turning that same expertise toward solving AI inference at scale. Customers include Hugging Face, and Salesforce.
from the Felicis community.
from the Felicis community.