Poolside AI

What is Poolside AI?

Poolside AI is one of the most heavily funded AI coding platforms ever built, having raised $2.6B across multiple rounds including a reported $1B investment from Nvidia at a ~$12B valuation in late 2025. Founded by Jason Warner (former GitHub CTO) and Baptiste Rozière, Poolside targets large enterprises that need AI coding assistance but can't send their codebase to third-party cloud providers.

Key Features

Self-Hosted Deployment — Poolside models run within the customer's own infrastructure (AWS, Azure, GCP, or on-premise). Code, context, and generated outputs never leave the customer's environment — addressing the primary barrier to enterprise AI coding adoption.

Codebase Fine-Tuning — Poolside fine-tunes its models on each organization's specific codebase, internal libraries, and coding conventions. The result is an AI assistant that understands your company's specific patterns, not just generic code from the internet.

AWS Bedrock Availability — Poolside is available as a managed model on AWS Bedrock, allowing AWS-centric enterprises to access it through familiar infrastructure without standing up their own model serving.

Enterprise Integration — SSO, audit logging, usage governance, and integration with enterprise development workflows (GitHub Enterprise, Jira, etc.).

Muse Code Editor — Poolside's proprietary coding interface (alongside IDE integrations) designed for AI-native workflows.

Who Uses Poolside

Large enterprises, banks, defense contractors, and regulated industries where IP protection and data residency requirements make cloud-based AI coding tools non-viable. Customers typically have 5,000+ developers.

The Market Position

Poolside occupies a specific but important niche: enterprises that need the productivity benefits of AI coding assistance but have hard requirements around data sovereignty. This is a smaller market than GitHub Copilot's mass-market approach, but a higher-value one.

Bottom Line

Poolside AI is not for most organizations — its minimum deal size, deployment complexity, and enterprise focus put it out of reach for typical software teams. But for large enterprises with strict data residency requirements and significant developer headcount, it may be the only viable option for AI coding assistance.