Poolside company profile: valuation, funding & revenue

This Poolside company profile covers the company's valuation, revenue, funding rounds, investors and history. Poolside works in artificial intelligence, with a focus on foundation models for software development. It was founded in 2023 and is headquartered in Paris, France. Its latest valuation on record is US$12 billion pre-money (reported) (August 2026). Every figure links to a public source, and anything the company hasn't disclosed is marked as such.

Revenue

Funding history

Poolside timeline

Key dates from the sources on this page, oldest first. Metric snapshots are listed after the funding events.

• 2023 — Poolside is founded in Paris by Jason Warner (CEO) and Eiso Kant.
• Oct 2024 — Series B: US$500M, valuing the company at US$3B. Investors: Bain Capital Ventures (lead); Nvidia, eBay and others.
• October 2025 — Reported talks: Reported in October 2025 to be raising about US$2 billion at a US$12 billion pre-money valuation, with Nvidia investing up to US$1 billion; not confirmed as closed.

Investors

Overview

Poolside is building foundation models specialised for software development, with the long-term aim of reaching artificial general intelligence through coding. Rather than build a general chatbot and then adapt it to code, Poolside trains its own models from the ground up with a focus on programming, using a technique it calls reinforcement learning from code execution feedback: its models write code, run it, see whether it works and learn from the results at enormous scale. It sells its models and coding assistant to large enterprises and government agencies that want powerful AI for software development deployed securely, often in their own environments.

Jason Warner, the CEO, and Eiso Kant founded Poolside in 2023. Warner was previously chief technology officer of GitHub, where he oversaw the launch of GitHub Copilot, and Kant founded source{d}, an early company applying machine learning to code. Poolside has teams in Paris and the United States, and its early backers include French billionaire Xavier Niel, reflecting its European roots.

Strategy

Poolside's thesis is that software is the best domain for advancing AI capabilities, because code can be automatically tested: a program either runs and passes its tests or it doesn't. This creates a near-limitless source of feedback for training. Poolside has generated large volumes of synthetic programming tasks and run them in execution environments to teach its models through trial and error. The company believes that models trained this way will become better at reasoning more broadly, not just at coding.

Products and customers

Poolside's offering includes its models, a coding assistant that integrates with developers' editors, and agent capabilities for longer tasks. Its go-to-market focus is large enterprises and the public sector, particularly organisations in regulated industries such as defense, finance and government that can't send their code to external cloud services. It deploys models within customers' own cloud accounts or data centres, and has partnered with Amazon Web Services and other providers to make this easier.

Project Horizon

To train ever-larger models, Poolside announced Project Horizon, a plan to build a massive AI data centre campus in West Texas with capacity of up to 2 gigawatts, developed in phases with partners including CoreWeave. This move into infrastructure reflects how frontier AI developers increasingly need control over large amounts of computing power, and ties into its funding discussions with Nvidia.

Funding

Poolside raised a US$126 million seed round in 2023, one of the largest seed rounds in European history. In October 2024 it raised a US$500 million Series B at a US$3 billion valuation, led by Bain Capital Ventures with participation from Nvidia, eBay and others. In October 2025, Bloomberg reported that Poolside was raising about US$2 billion at a US$12 billion pre-money valuation, with Nvidia investing between US$500 million and US$1 billion, partly linked to purchases of Nvidia chips. We haven't seen confirmation that this round closed, so the data above uses the October 2024 figure.

Competition

Poolside competes with frontier AI labs that offer coding models, including OpenAI, Anthropic and Google; with coding tool makers such as GitHub Copilot, Cognition and Cursor; and with European AI developers such as Mistral AI, which also offers coding models for private deployment. Its differentiation is its focus on secure enterprise and government deployment and its research approach to training through code execution.

Poolside at a glance

Poolside is an AI company founded in 2023 by former GitHub CTO Jason Warner and Eiso Kant, with operations in Paris and the United States. It builds foundation models for software development trained with reinforcement learning from code execution. It raised US$500 million at US$3 billion in October 2024, led by Bain Capital Ventures, and was reported in 2025 to be raising at a much higher valuation with Nvidia.

Why secure deployment matters

Many large organisations, particularly banks, defense contractors and government agencies, have strict rules about where their source code can go. Code can contain trade secrets, security vulnerabilities or classified information. Cloud-based coding assistants that send code to external servers may not be allowed. Poolside's focus on deploying models inside customers' own environments addresses this barrier, giving it access to a segment of the market that some competitors find harder to serve.

Risks

• Competition with frontier labs: companies with far greater resources are also racing to build the best coding models.
• Capital intensity: training frontier models and building data centres requires enormous ongoing investment.
• Commercial traction: enterprise sales cycles are long, and revenue figures haven't been disclosed.

Questions for enterprise buyers

• Can the model be deployed fully within your own infrastructure?
• How does its coding quality compare on your own codebase and languages?
• What support, fine-tuning and security certifications are available?

Reinforcement learning from code execution

Most language models learn first by predicting the next word in huge amounts of text, then are refined with human feedback. Poolside adds a large stage where the model attempts programming tasks in sandboxed environments, runs its code against tests and receives automatic rewards for correct solutions. Because computers can check code far faster than humans can review it, this approach can generate billions of training signals. The challenge is creating enough varied, realistic tasks, and making sure models learn general skills rather than tricks for passing specific tests. Several frontier labs now use similar techniques, which means Poolside's early bet has become an industry trend.

European roots

Poolside is one of the most prominent AI companies with deep roots in France, alongside Mistral AI. Its Paris presence gives it access to French engineering talent and to European customers that prefer suppliers with a local footprint, while its US operations keep it close to American enterprise and government buyers.

Latest developments (October 2026)

Newcomer reported in August 2026, citing a letter to investors, that Poolside had agreed a non-exclusive licensing deal with Nvidia worth about US$6 billion plus a US$1 billion investment at a US$12 billion pre-money valuation, according to Newcomer. Neither company had publicly confirmed the terms in the sources we reviewed. Poolside builds AI models for software development. Its 2024 valuation of US$3 billion is the latest confirmed figure.

What to watch

Watch for confirmation of a new funding round, Project Horizon progress, major enterprise and government contracts, and new model releases. Our guide to the most valuable AI startups compares Poolside with peers.

Takeaways

Regulated enterprises can access advanced coding AI without sending code outside their walls. Founders can see how a sharp research thesis attracts large funding. Investors should separate confirmed rounds from reported talks and weigh heavy capital needs.

Sources

Every figure on this page comes from the public sources below. Where sources disagreed, we used the company's own announcement or a primary news report. Private-company data changes; check the date beside each figure.

1. Newcomer
2. Bloomberg
3. TechCrunch