Overview
Poolside Infrastructure Company builds large language models tailored for software development tasks, including a software-engineering foundation model, an API, and an IDE-integrated AI assistant. The models are trained in-house and deployed via a full-stack approach, with the option to fine-tune on a client’s codebase and run on-premises for data security. A differentiator is RLCEF—reinforcement learning from code execution feedback using synthetic data to improve performance and safety. They partner with AWS to offer access on Bedrock and target users from individual developers to large enterprises in regulated industries, aiming to make AI-assisted software development faster and more secure.
Work here?Claim your company
Simplify's take
What believers are saying
- April and July 2026 releases expanded Laguna to 256K and 1M contexts.
- Nvidia invested $1 billion in August 2026, validating Poolside’s platform and team.
- AWS and CoreWeave partnerships widen distribution for enterprise and on-prem inference.
What critics are saying
- Nvidia’s $6 billion license concentrates Poolside’s value in one counterparty through 2027.
- OpenAI, Anthropic, and Cursor compress pricing and talent retention in enterprise coding AI.
- If Laguna benchmarks stall, Poolside becomes a licensing shell, not an enduring platform.
What makes Poolside Infrastructure Company unique
- Poolside’s Laguna models target agentic coding, not generic chat, since April 2026.
- Its Model Factory builds models end-to-end, enabling on-prem deployments for regulated buyers.
- Open-weight releases like Laguna XS.2 and S 2.1 create adoption and switching costs.
Help us improve and share your feedback! Did you find this helpful?