Coiled
Coiled provides ephemeral cloud compute for Python data engineers and scientists, eliminating Kubernetes and Docker for running data science, ML, and AI workloads at scale. Customers include NASA, Moderna, NVIDIA, ARM, D.E. Shaw, and the US Air Force.
- Company typePrivate
- Founded2020
- HeadquartersNew York, United States
- Headcount11–50
- GTM typeB2B
- OfferingSoftware
What Coiled does
Coiled (legally Coiled Computing) is a New York-based cloud compute platform founded in 2020 that provides ephemeral, zero-management cloud infrastructure for Python data engineers, data scientists, and ML/AI practitioners. The platform abstracts away Kubernetes, Docker, and cloud configuration by allowing users to call up cloud VMs directly from Python via `import coiled`, automatically replicate local Python environments (Package Sync) onto those machines, run distributed workloads on managed Dask clusters, execute serverless functions via the `@coiled.function` decorator, or submit batch jobs with hardware specified in script comments. Infrastructure spins up on demand and tears down immediately after work completes, running inside the customer's own AWS, GCP, or Azure account.
The product portfolio centers on three core primitives — Batch Jobs, Serverless Functions, and Managed Dask Clusters — supplemented by Jupyter Notebooks on Cloud, native integrations with workflow orchestrators (Prefect, Dagster, Airflow) and ML tooling (PyTorch, Hugging Face, MLflow, Weights & Biases, Neptune, Optuna, XGBoost), and GPU support for A10G, A100, and H100 instances. The company's go-to-market is product-led growth combined with community-led distribution through the Dask and PyData open-source ecosystems, supported by a Slack community, technical blog, benchmark studies, and a generous freemium tier ($25/month in customer-cloud credits). Monetization comes from a usage-based fee of $0.05 per CPU-hour beyond the free tier, with cloud provider costs passed through directly to AWS/GCP/Azure. Enterprise and team features (SSO, cost controls, advanced user management) represent the upside monetization lever, supported by a recent Head of Sales hire.
The company has raised approximately $26 million across a $5M seed (October 2020, Costanoa Ventures) and a $21M Series A (May 2021, Bessemer Venture Partners), and serves a diverse customer base spanning enterprise (Moderna, NVIDIA, ARM, D.E. Shaw, Nextroll), government (NASA, US Air Force), and high-growth startups (KoBold Metals, Kestrel, Floodbase, Accure, Guac) across biotech, aerospace, mining, financial services, defense, environmental, and energy verticals.
Coiled firmographics
Firmographics- Name
- Coiled
- Legal name
- Coiled Computing
- Website
- https://coiled.io
- Company type
- Private
- Founded year
- 2020
- Operating status
- Operating
- Headcount range
- 11–50 employees
- Short description
- Coiled provides ephemeral cloud compute for Python data engineers and scientists, eliminating Kubernetes and Docker for running data science, ML, and AI workloads at scale. Customers include NASA, Moderna, NVIDIA, ARM, D.E. Shaw, and the US Air Force.
- Ownership category
- akta.pro rank
Coiled industry classification
Industry- Product category
- Cloud Computing Platform for Data Science and ML
- NAICS
- Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (51821), Computer Facilities Management Services (541513)
- SIC
- Services-Computer Programming, Data Processing, Etc. (7370), Services-Computer Processing & Data Preparation (7374)
- akta.pro primary industry
- Compute Resource Management & Scheduling (Cluster/Workload Managers) (HDABADAK)
- akta.pro secondary industries
- Bare Metal Provisioning & Lifecycle Automation (HDABADAJ), MLOps/LLMOps & Model Lifecycle Management Services (BPAEAHAH)
Keywords
Where Coiled is headquartered
LocationHeadquarters
- HQ city
- New York
- HQ country
- United States
- HQ region
- North America
Markets served
Coiled business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations
Revenue model
- Usage-based Cloud Compute: Customers pay for compute resources consumed. Coiled charges $0.05 per CPU-hour for usage beyond the free tier. Cloud provider costs (typically $0.02-0.05 per CPU-hour for compute) are paid by customers directly to AWS/GCP/Azure. Free tier includes $25/month for users running in their own cloud account (enough for ~500 CPU-hours).
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Freemium | Monthly | Free tier for individual users |
| Usage-based | Pay-as-you-go | Pay-as-you-go beyond free tier |
Go-to-market motion2 records
Distribution channels3 records
Marketing channels5 records
Coiled product offering
Product offeringCore offering
Coiled provides a cloud compute platform that lets Python data engineers and scientists run distributed data processing, machine learning, and AI workloads on cloud VMs (AWS, GCP, Azure) without managing Docker or Kubernetes. The platform centers on three core products: Batch Jobs for executing scripts on ephemeral cloud VMs, Serverless Functions via the @coiled.function decorator, and Managed Dask Clusters for distributed DataFrame, Array, and Futures computing, all unified by automatic environment synchronization that replicates local Python setups onto cloud machines.
Product overview
Coiled is a lightweight cloud compute platform purpose-built for Python data engineers and scientists. Its product portfolio centers on ephemeral, zero-maintenance cloud infrastructure across AWS, GCP, and Azure. The core platform consists of Batch Jobs (for running scripts on cloud VMs), Serverless Functions (Python-first serverless via @coiled.function decorator), and Managed Dask Clusters (distributed DataFrame/Array/Futures computing). These three tools share a common architecture: users write standard Python code locally, Coiled spins up ephemeral cloud VMs on demand in the user's own cloud account, executes the work, and automatically tears down infrastructure when done. A key platform feature is Package Sync, which replicates the local Python environment (pip, conda, uv, local packages) onto cloud machines without Docker. Integrations with Prefect, Dagster, Airflow, MLflow, and GitHub Actions extend orchestration and MLOps capabilities. Jupyter Notebooks on Cloud provide GPU-enabled notebook environments on cloud VMs with live file synchronization.
Differentiator
Problem solved
Functional benefit
Products and services
- Batch Jobs Run any code on ephemeral cloud VMs with zero infrastructure management. Users specify hardware requirements in script comments or environment variables, and Coiled spins up machines on demand, runs the job, and immediately tears down resources. No Docker or Kubernetes required. Targets data engineers and scientists who need to execute scripts at scale.
- Serverless Functions (Cloud Functions) A Python-first serverless compute service that wraps Python functions with the @coiled.function() decorator, automatically executing them on cloud VMs with automatic scaling, environment sync, and ephemeral resource cleanup. Supports GPU instances, ARM architecture, configurable VM types and regions, and a .map() for parallel execution across thousands of machines.
- Dask Clusters Managed Dask clusters on the cloud, enabling distributed computing (DataFrames, Arrays, Futures) with zero infrastructure management. Scales automatically from a single machine to thousands of workers on AWS, GCP, or Azure. Integrates with existing pandas and Dask code with minimal changes and includes observability dashboards.
- Jupyter Notebooks on Cloud Launch GPU-enabled Jupyter Notebooks on ephemeral cloud VMs. Automatically syncs local files and software packages (including GPU-configured PyTorch) and keeps notebooks running on cloud hardware. Changes sync back to local machine in real time.
Quantifiable outcome
- 360x faster processing with $0.90 cost for 1TB dataset
- +4 more outcomes
Companies that use Coiled
Customer profileNamed customers16 records
Segments4 records
Ideal customer profiles4 records
Coiled technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
Integration12 records
AI capability7 records
Feature7 records
Coiled partnerships and signals
Strategic signalPartnerships
Seven partnerships are on record, tiered core.
- DaskcoreDeep integration with Dask as the primary distributed computing framework. Coiled manages Dask clusters, handles worker provisioning, scaling, and cleanup. Users can take existing pandas/Dask code and scale to cloud with minimal changes.
- PrefectcoreNative integration with Prefect push work pools for serverless workflow orchestration. Users can create Coiled push work pools with single command and deploy Prefect flows to serverless infrastructure without managing workers.
- PyTorchcoreCoiled supports PyTorch training on GPU instances with automatic CUDA configuration. Works with Accelerate for multi-node distributed training. Environment syncs local PyTorch setup to cloud GPUs automatically.
- Hugging FacecoreSupport for Hugging Face transformers and training workflows. Environment synchronization handles Hugging Face token authentication via environment variables. Enables LLM fine-tuning on cloud GPUs.
- DagstercoreGuac case study demonstrates integration with Dagster for ETL orchestration. Coiled handles compute-intensive jobs while Dagster manages workflow scheduling. Architecture demonstrates seamless combination of both platforms.
- XGBoostcoreSupport for XGBoost training with Dask for distributed gradient boosting. Used in Guac's demand forecasting ML pipelines for ensemble modeling.
- XarraycoreNative support for Xarray workflows on terabyte-scale geospatial data. Processes Zarr datasets, handles chunked array operations. Enables planetary-scale climate and geospatial analysis.
Scale indicators6 records
Recent moves6 records
Expansion highlights6 records
Coiled competitors and assessment
Company assessmentDirect peers
- Anyscale: Anyscale offers managed Ray clusters for distributed Python compute and ML/AI workloads. It is Coiled's most direct head-to-head competitor, serving the same data engineer / data scientist persona with managed cloud compute for Python, just on Ray instead of Dask.
- Modal Labs: Modal provides serverless cloud compute for Python developers via a function decorator pattern and has a Python-developer-experience-led PLG motion. It overlaps with Coiled's @coiled.function serverless offering and targets the same 'Heroku for Python' positioning.
- Saturn Cloud: Saturn Cloud offers managed Dask clusters and Jupyter notebooks on cloud infrastructure for data science teams. It is a direct Dask-focused competitor to Coiled's managed Dask Clusters product with similar target personas and use cases.
Broad incumbents
- Domino Data Lab: Domino is an enterprise MLOps platform that includes managed compute for data science workloads. It competes with Coiled at the enterprise tier with broader capabilities (governance, model registry, deployment) but less focused on Python-native ephemeral compute.
- Databricks: Databricks offers a unified data and AI platform including managed Spark clusters and increasingly Python/ML compute via Databricks Runtime. It is a much broader incumbent that overlaps with Coiled's data engineering and ML training use cases at the enterprise level.
- AWS SageMaker: Amazon SageMaker provides managed ML training, processing, and notebook environments on AWS infrastructure. It competes with Coiled for ML training workloads, especially GPU compute, and benefits from native AWS integration and bundled pricing.
- Google Vertex AI: Google Cloud Vertex AI is a managed ML platform offering training, inference, and notebooks on GCP. It is a broad incumbent competing with Coiled for enterprise ML training workloads on the Google cloud, with bundled pricing advantages.
Others
- Prefect: Prefect is a workflow orchestration platform with native Coiled integration (push work pools). It is a partner rather than competitor, but its move into cloud-hosted execution environments creates potential overlap with Coiled's compute layer.
Emerging players
- Fly.io: Fly.io runs application workloads close to users via lightweight VMs. While not directly competing on Dask-style distributed compute, it represents an alternative developer-friendly cloud compute platform that could expand into Python data workloads.
- Railway: Railway is a developer-focused cloud platform offering simple deployment and compute with a PLG motion. It overlaps with Coiled in serving Python developers seeking low-friction cloud infrastructure, though focused on application hosting rather than distributed data compute.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat5 records
Key risks7 records
Key highlights7 records
Customer concentration
Coiled social profiles
Digital presenceCoiled financial estimates
Financial estimateRevenue estimate
Valuation estimate
Coiled leadership team
Management profileNumber of profiles
Profiles2 records
Coiled funding detail
Funding detailFunding overview
Funding rounds2 records
Investors5 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Coiled M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Coiled
What does Coiled do?
Coiled provides a cloud compute platform that lets Python data engineers and scientists run distributed data processing, machine learning, and AI workloads on cloud VMs (AWS, GCP, Azure) without managing Docker or Kubernetes. The platform centers on three core products: Batch Jobs for executing scripts on ephemeral cloud VMs, Serverless Functions via the @coiled.function decorator, and Managed Dask Clusters for distributed DataFrame, Array, and Futures computing, all unified by automatic environment synchronization that replicates local Python setups onto cloud machines.
Is Coiled a public or private company?
Coiled is a private company. It is classified as venture growth investor backed and is currently operating.
When was Coiled founded?
Coiled was founded in 2020. It employs 11 to 50 people.
Where is Coiled based?
Coiled is headquartered in New York, United States, in the North America region.
How does Coiled make money?
One revenue line is on record: usage-based Cloud Compute.
Who are Coiled's main competitors?
Direct peers on record are Anyscale, Modal Labs and Saturn Cloud. Broad incumbents are Domino Data Lab, Databricks, AWS SageMaker and Google Vertex AI. Prefect is listed as an others. Emerging players are Fly.io and Railway.
Does Coiled have an API?
Yes. Coiled provides a Python SDK (`import coiled`) that enables developers to programmatically create cloud clusters, deploy serverless functions, and submit batch jobs. Key API entry points include `coiled.Cluster()` to spin up ephemeral Dask clusters, `@coiled.function()` decorator for serverless function deployment, and `coiled.run` CLI for executing scripts on cloud VMs. The coiled Python package also exposes a REST API and interactive CLI. Documentation is available at docs.coiled.io. Developer documentation is at docs.coiled.io/index.html.
What industry is Coiled in?
Coiled's product category is Cloud Computing Platform for Data Science and ML. Its primary akta.pro industry code is HDABADAK, Compute Resource Management & Scheduling (Cluster/Workload Managers), with a secondary code of HDABADAJ, Bare Metal Provisioning & Lifecycle Automation. Its NAICS code is 51821 and its SIC code is 7370.