Skip to main content

Modal Labs

Category: AI Infrastructure

A serverless cloud platform designed to run data and AI applications at scale by abstracting away infrastructure management through a Python-native interface. Modal Labs was founded in 2021. The company is led by Erik Bernhardsson. Based in New York, USA. Team size: 50-100. Total funding raised: $110.0M. Latest round: Series B. Key investors include ["Lux Capital","Redpoint Ventures","Amplify Partners","Definition Capital","Creandum"].

Founded
2021
Headquarters
New York, USA
Team size
50-100
Total funding
$110.0M

Value proposition

Eliminates the 'DevOps tax' for AI teams by providing sub-second cold starts and instant scaling from zero to thousands of containers without managing Kubernetes or Docker clusters.

Products and solutions

["Serverless GPU/CPU Execution Engine","Modal Functions (Distributed Task Queue)","Modal Volumes (High-performance persistent storage for model weights)","Modal Web Endpoints (Instant API deployment for ML models)","Distributed Data Structures (Shared Dicts and Queues)","Scheduled Jobs and Cron Workloads"]

Unique value

Unlike traditional serverless providers, Modal uses a custom-built container runtime and filesystem optimized specifically for the heavy dependencies and large binary blobs typical of modern AI/ML workloads.

Target customer

Machine learning engineers, data scientists, AI startups, and enterprise data teams requiring scalable GPU/CPU compute.

Industries served

["Artificial Intelligence & Generative AI","Biotechnology & Drug Discovery","Financial Services (Quantitative Analysis)","Data Engineering","Autonomous Systems"]

Technology advantage

Achieves sub-second cold starts for complex environments by utilizing a proprietary lazy-loading filesystem and a highly optimized container orchestration layer that bypasses the overhead of standard Kubernetes.

How they differentiate

Modal differentiates through a custom-built container runtime and lazy-loading filesystem that enables sub-second cold starts for heavy AI workloads. Unlike competitors focused solely on model inference, Modal provides a programmable, Python-native interface for arbitrary distributed code execution, abstracting away all Kubernetes and Docker management.

Main competitors

["Baseten","Replicate","RunPod","Lambda Labs"]

Key partnerships

["NVIDIA (GPU infrastructure access)","Major Cloud Providers (AWS/GCP/CoreWeave for underlying compute)","Hugging Face (Integration for model deployment)","Venture Partners: Lux Capital, Redpoint Ventures, and Amplify Partners"]

Notable customers

["Ramp","Substack","Scale AI","Meta","Cohere","SphinxBio"]

Major milestones

["Reached Unicorn status with a $1.1B valuation in late 2025","Announced General Availability and Series A in October 2023","Reportedly in talks for a Series C/Growth round at a $2.5B valuation in early 2026","Voted to the Enterprise Tech 30 list as a top promising early-stage company"]

Growth metrics

Achieved an 8-figure revenue run rate and 50x growth in usage within a single year; currently powers cloud infrastructure for over 10,000 teams.

Market positioning

High-performance serverless infrastructure provider for AI/ML engineering teams, positioned as the 'DevOps-less' alternative to raw cloud providers and specialized MLOps platforms.

Geographic focus

Global (Headquartered in New York, USA)

Patents and IP

Proprietary custom container runtime and specialized distributed systems architecture (specific public patent filings not disclosed).

About Erik Bernhardsson

Erik was previously the CTO at Better.com, where he scaled the engineering team from 5 to over 300 members. Prior to that, he was a Principal Engineer at Spotify for over six years, where he built the core music recommendation system and created renowned open-source projects including Luigi (workflow manager) and Annoy (approximate nearest neighbors).

Official website: