Baseten, an AI inference company based in San Francisco, has acquired Blaxel, an infrastructure firm specializing in secure and persistent execution environments for AI agents. The acquisition expands Baseten’s ability to offer perpetual sandbox execution within the same production platform used for inference, training, and post-training. Blaxel’s team and technology will join Baseten to strengthen tools for agentic workloads and long-running autonomous tasks.
Shifting Needs in AI Infrastructure
Modern AI applications are moving beyond models that generate answers toward agents that directly take actions. This shift creates new infrastructure demands, particularly for isolated execution environments that can persist across long-running tasks without adding unnecessary overhead. Baseten views secure, stateful compute as an essential layer for production AI systems.
Blaxel’s Sandbox Technology
Blaxel has developed infrastructure specifically for this emerging class of agentic applications. Its platform includes secure microVM-based sandboxes that spin up and resume up to five times faster than competing solutions. The technology natively supports agent code execution, tool use, MCP servers, and other workloads that require secure and stateful compute.
Accelerating Agentic Workloads
Baseten said that combining Blaxel’s execution infrastructure with its inference platform will create a more complete runtime for production AI applications. Developers will be able to run models and the environments where those models act on infrastructure designed to work together. This approach cuts the latency and complexity that can arise when state and data move between separate systems.
Executive Perspectives
Amir Haghighat, co-founder and chief technology officer of Baseten, said agents need models as well as secure, fast, scalable, and reliable environments for production. He added that bringing Blaxel into Baseten will accelerate the company’s vision for agent infrastructure at scale. Paul Sinaï, co-founder and chief executive officer of Blaxel, said the combination can deliver a model and execution environment in a single system.
Integration Roadmap
In the near term, Blaxel’s technology will expand Baseten’s foundation for secure code execution, tool use, and session persistence. Over time, these capabilities will be integrated more deeply with Baseten’s inference, storage, observability, and enterprise infrastructure. The goal is to reduce latency and complexity by eliminating the need to move state and data between separate systems.
Recent Growth and Market Position
The acquisition follows a period of significant expansion for Baseten, including a Series F fundraise at a $13 billion valuation. Baseten also recently acquired Parsed, a reinforcement learning startup focused on post-training and continual learning. These moves reflect Baseten’s broader strategy to provide a complete runtime for production AI applications.
About Baseten
Baseten builds systems software that runs the entire workload for AI applications, from GPUs and autoscaling to observability, billing, and developer tools. Its customers include leading AI companies such as Abridge, Clay, Cursor, Lovable, Mercor, and OpenEvidence. Founded in 2019 and based in San Francisco, Baseten has raised more than $2 billion from investors including Altimeter Capital, Battery Ventures, Greylock, IVP, and NVIDIA.
With the acquisition of Blaxel, Baseten is positioning itself to support the next wave of AI agents that must reliably execute actions in production. The combination of inference and secure execution infrastructure aims to give developers a faster and more cohesive platform for building agentic applications. As AI systems continue to take on more complex tasks, unified infrastructure may become a key competitive advantage.