Netris has raised $15 million in a Series A round led by Andreessen Horowitz, with a16z partner Guido Appenzellerjoining the company’s board. The network automation startup provides software that helps neocloud operators cut the time required to bring AI inference and training infrastructure to market, addressing a bottleneck that can leave expensive GPU clusters sitting idle for months before generating revenue.
The company’s platform automates the setup, configuration, and ongoing operations of data centre networking, while providing hardware-level abstraction and multi-tenancy isolation, allowing neoclouds to serve multiple customers from shared infrastructure. Unlike traditional software-defined networking approaches, Netris runs its automation entirely in hardware, a distinction its CEO Alex Saroyan argued is essential for AI workloads, where traffic volumes are too high for software-based solutions to keep pace.
The platform is vendor-agnostic, compatible with both Nvidia and AMD server environments. Nvidia itself endorsed Netris two years ago after a product demonstration, recommending the company to several of its own customers. Netris is now live across more than 35 GPU clusters globally, totalling approximately one million GPUs, with operators including Lightning AI, Foxconn, Hewlett Packard Enterprise, TensorWave, and Telus.
Saroyan noted that Netris does not use AI in its own platform, relying instead on deterministic algorithms developed over eight years, on the basis that configuring thousands of switches requires precision and repeatability rather than the unpredictability that can accompany AI-driven systems. The new funding will support engineering and sales hiring, expanded hardware vendor support, and further algorithm development.
