CoreWeave acquires AI agent-training startup OpenPipe
CoreWeave, the fast-growing provider of cloud servers for AI model training, announced on Wednesday that it has reached an agreement to acquire OpenPipe, a two-year-old Y Combinator–backed startup specializing in reinforcement learning tools for enterprise AI development.
The acquisition underscores CoreWeave’s strategy to move further up the AI technology stack, offering not just infrastructure but also advanced software capabilities tailored to enterprise needs.
Reinforcement Learning Meets Scalable AI Cloud
“Reinforcement learning is emerging as a pivotal force to strengthen model performance on agentic and reasoning tasks,” said Brian Venturo, co-founder of CoreWeave, in a statement to TechCrunch. “By combining OpenPipe’s advanced self-learning tools with CoreWeave’s high-performance AI cloud, we’re expanding our platform to give developers at AI labs and beyond an important advantage in building scalable intelligent systems.”
OpenPipe is best known for ART (Agent Reinforcement Trainer), an open-source toolkit that helps companies train AI agents with reinforcement learning. This approach allows AI systems to improve continuously by being rewarded for correct outputs—making them better suited for reasoning tasks and custom enterprise applications.
Building on Momentum
The terms of the deal were not disclosed. However, OpenPipe’s trajectory has been strong: the Seattle-based startup raised a $6.7 million seed round in March 2024, backed by Costanoa Ventures, Y Combinator, Google DeepMind’s Logan Kilpatrick, GitHub co-founder Tom Preston-Werner, and GitHub Copilot co-creator Alex Graveley.
This move follows CoreWeave’s March 2025 acquisition of Weights & Biases, an AI developer platform widely used for model tracking and experiment management. With OpenPipe now in its fold, CoreWeave is positioning itself as not only the cloud provider for leading AI labs such as OpenAI, but also as a one-stop platform for smaller enterprises seeking to build custom AI systems.
Expanding AI Agent Development
Reinforcement learning has increasingly become a focus area for AI startups and research labs. By tailoring AI models to specific tasks and enterprise needs, reinforcement-trained agents can deliver more reliable and context-aware outputs. However, this training is compute-intensive—a natural fit for CoreWeave’s cloud infrastructure business.
By bringing OpenPipe’s team and technology in-house, CoreWeave will integrate reinforcement learning workflows directly into its offerings. Existing OpenPipe customers will transition to CoreWeave, giving the cloud provider a new customer base and an expanded suite of services.
The Bigger Picture
The deal highlights a trend of consolidation in the AI ecosystem, as infrastructure providers seek to layer on higher-level capabilities to attract developers and enterprises. For CoreWeave, combining scalable cloud computing with reinforcement learning tools could further cement its position as a key enabler of the next generation of intelligent systems.
