Factory

Updated October 26, 2026

Introducing Flexible Deployment Options at Factory

By Factory Team - September 18, 2026 - 1 minute read

Company

Product

Share

Halftone illustration of a server rack beside a panel bearing the Factory logo

Today, we’re introducing Factory Private, a new deployment option that lets organizations run some or all of Factory in their own VPCs, including sandboxes through OpenCompute while Factory operates the control plane. Factory now offers two deployment models: Factory Managed and Factory Private. Together, these models give organizations more flexibility and control over how Factory is operated, where Factory’s autonomous software engineering agents, known as Droids, run, and where data resides.

At Factory, we believe the most consequential software work involves an organization’s critical data and systems. Where that work runs depends on each organization’s security, compliance, and operating requirements. That is why we built a deployment stack that lets teams use the same Factory platform across all environments, without changing how they build software.

Two deployment models for different requirements

Factory Managed - provides a Factory-hosted control plane while Droids run across developer machines, CI/CD infrastructure, or Factory-managed Droid Computers. Teams can also bring their own keys and machines.

Factory Private - places some or all of Factory’s infrastructure in the customer’s VPC, for organizations with a specific regulatory or policy reason. By default, Factory runs the control plane and Droid sandboxes run in the customer’s VPC through OpenCompute, so code execution stays in their environment. The sandbox infrastructure is governed by the customer’s existing IAM, network, and security controls.

Factory Private enabled the NVIDIA Nemotron research team to use Factory’s agentic software development capabilities in their own infrastructure for post-training Nemotron 3 Ultra, including evaluation and optimization.

“Detailed task trajectories and verified outcomes helped us understand how the model used tools and recovered from errors, guiding our next post-training cycle. We used Factory to improve Nemotron 3 Ultra’s capabilities while keeping our code, models, and other critical IP within our own infrastructure” - Jian Zhang, Director of post-training, Nemotron Research team.

Across these deployment models, Factory brings sovereign, model-independent intelligence to the full software development lifecycle. Droids can understand requirements and issues, plan work, write and modify code, run tests, and prepare changes for review. Factory Private keeps code execution, and the code and internal systems Droids work with, inside the customer’s VPC by default. Because Factory is model-agnostic, teams can use frontier, open, or custom models through their own providers and gateways, adapting their model strategy without changing how their teams build software.

To find the right deployment model for your organization, contact our sales team. If you’re a federal organization, reach out to our federal team to discuss your requirements. Learn more about Factory deployment options and deployment patterns in the Factory docs.

Factory Private is currently in Private Preview. To request access, contact Factory or contact your Factory account team.

Ready to build the software of the future?

Start building

Arrow Right Icon