Replicate

Acquired by Cloudflare · 2025

Operates cloud platform to run 50,000+ open-source ML models via API with auto-scaling and pay-per-use billing. Raised $57.9M before acquisition; created Cog for packaging ML models; first platform to handle Stable Diffusion at scale.

Defined abstractions for generative AI serving; acquired by Cloudflare in late 2025 to democratize model deployment for all developers.

replicate.com Infrastructure & APIs

United Statesacquired

At a glance

Total raised

$57.8M

last round acquired

Headcount

11-50

approximate · public range

Headquarters

San Francisco

United States

Founded

2019

7 years active

Status

Acquired

by Cloudflare

Subsector

Infrastructure & APIs

For subscribers

Subscribe to see the full profile

Including Funding rounds, Bull / Bear thesis, Stock + earnings, Roster changes, Patents, News, and Open roles.

Founding

50% off

/month

94 of 100 spots left

Full

$10

/month

Available once all 100 Founding Member spots are claimed.

Get full access

Already subscribed? Sign in →

Funding history

Total raised

$57.8M

Rounds tracked

Latest stage

Acquired

Date	Stage	Amount	Valuation	Lead investors
Nov 17, 2025	Acquired	—	—	Cloudflare
Dec 5, 2023	Series B	$40M	$350M	Andreessen Horowitz
Feb 21, 2023	Series A	$12.5M	—	Andreessen Horowitz
Feb 21, 2023	Seed	$5.3M	—	Sequoia Capital

Founder backgrounds

Andreas Jansson

Founder & CTO

Holds a PhD in Music Informatics from City University of London and was previously a machine learning engineer at Spotify, where he served as Staff Machine Learning Engineer from 2014 to 2019. Prior to Spotify, he worked at The Echo Nest as an engineer. At Spotify, he experienced frustration with deploying ML models to production and built tools and templates that let researchers package and deploy models themselves—work that directly informed Replicate's mission. His leadership is marked by pragmatism: he has stated, "I could read all about new developments in AI, but there were no tools readily available to actually apply any of it," a belief that shaped both his career path and the platform's focus on making ML infrastructure accessible to developers.

Ben Firshman

Founder & CEO

Created Fig (which evolved into Docker Compose) and served as Director of Product at Docker, where he led open-source initiatives. Earlier, he built The Guardian's award-winning iPad app and contributed to early work on GOV.UK. At Docker and informed by frustrations with piecing together ML models from papers, he was motivated to build infrastructure to help researchers and developers share work more easily. Since founding Replicate in 2019, he has expanded the platform from a research community to hosting thousands of models, scaling alongside generative AI adoption. His leadership approach blends product oversight with hands-on engineering contributions across backend, DevOps, and frontend work.

AI bull / bear

Bull case

Replicate proved the generative AI infrastructure thesis at scale, becoming the go-to platform for developers needing instant access to 50,000+ open-source models without DevOps overhead. The company's early bet on Stable Diffusion—being first to handle it at scale—demonstrated exceptional technical execution and market timing. Cog, their open-source model packaging standard, created defensible network effects by establishing the de facto format for reproducible ML deployments. The Cloudflare acquisition validates both the technology and strategic positioning. Cloudflare's global edge network and developer reach means Replicate's infrastructure will now underpin AI deployment for millions of applications. The pay-per-use model with auto-scaling solved real friction in the ML deployment stack, and integration with Cloudflare's ecosystem dramatically expands TAM beyond early adopters to mainstream developers building AI features. The $57.9M raised before exit suggests strong unit economics and a relatively capital-efficient path to product-market fit in a notoriously capital-intensive category.

Bear case

Replicate operated in an increasingly commoditized infrastructure layer where hyperscalers (AWS Bedrock, Google Vertex AI) and well-funded competitors (Hugging Face, Modal, Together AI) offer similar model-serving capabilities with deeper pockets and existing customer relationships. The open-source nature of models means limited moat; any cloud provider can wrap APIs around popular models, and differentiation comes down to price and latency—a race to the bottom. The acquisition timing suggests Replicate may have struggled to scale independently. Despite raising $57.9M and achieving early traction, sustainable unit economics in GPU-heavy inference remain challenging. Cloudflare likely acquired the team and technology rather than a breakout standalone business. The fragmented open-source model ecosystem also presents risk: as foundation model providers (OpenAI, Anthropic) improve and consolidate share, demand for running thousands of smaller open models via API may decline, undermining Replicate's core value proposition of breadth over depth.

Generated by claude-sonnet-4-5-20250929 · May 1, 2026Refreshes quarterly · see disclosures

In the news

Replicate · last 30 days

No articles ingested yet for Replicate. Once the hourly news pipeline is live, every article the classifier tags as mentioning this company appears here with its one-line AI summary and sentiment.

Watch

Leadership

Ben Firshman

Founder & CEO

in 𝕏

Andreas Jansson

Founder & CTO

in 𝕏

Products

Replicate API Platform
Cloud platform enabling developers to run over 50,000 open-source machine learning models via API with automatic scaling and pay-per-use billing. Supports image generation, video synthesis, audio processing, and text models including Stable Diffusion, FLUX, Llama, and Whisper. Developers deploy models with a single line of code, scale from zero to millions of requests, and only pay for actual compute time used, eliminating infrastructure management complexity.
Cog
Open-source containerization tool that packages machine learning models into standard, production-ready Docker containers. Simplifies ML deployment by automatically handling CUDA configuration, dependency management, and GPU setup. Developers define model requirements in a cog.yaml file and prediction interface in Python, then deploy to Replicate's infrastructure or run anywhere Docker operates, bridging the gap between research and production ML systems.
Model Fine-tuning Service
Feature enabling users to customize pre-trained models with their own datasets without training from scratch. Developers upload training data (often as zip files of images), Replicate runs automated training jobs, and returns a personalized model variant tailored to specific needs. Particularly popular for creating branded content variants, custom image styles, and domain-specific model adaptations, making advanced ML accessible to non-experts.

Open roles

We don't have a live feed for this company's ATS. Their careers page has every open role.

View all careers ↗