AI infrastructure company Cornelis has raised $205 million in new funding to accelerate its efforts to reduce reliance on Nvidia’s dominance in AI hardware. The investment will support the development and deployment of its innovative network technology aimed at improving GPU utilization in large-scale AI workloads.

Central to Cornelis’ strategy is the launch of Active Compute Fabric, a new networking solution designed to address a critical bottleneck in AI training: idle GPU time caused by delays in data delivery. The company states that a significant portion of GPU compute cycles are wasted waiting for data to arrive, limiting overall system efficiency.

Active Compute Fabric aims to minimize this latency by optimizing data movement across AI clusters, ensuring that GPUs remain fed with the information they need to compute continuously. By improving data throughput and synchronization, the technology could help reduce the total cost of ownership for AI infrastructure while increasing performance per watt.

For content creators and AI developers, this advancement could mean faster model training times, lower cloud computing costs, and more accessible access to high-performance AI tools—especially as demand for generative AI, video rendering, and real-time analytics continues to grow. Cornelis’ approach shifts focus from raw chip power to smarter system-level design, offering an alternative path to scaling AI workloads.

The $205M funding round underscores growing investor confidence in alternatives to traditional GPU-centric AI architectures. While Nvidia remains the dominant player in AI accelerators, companies like Cornelis are targeting systemic inefficiencies that hardware alone cannot solve. As AI models grow larger and more complex, innovations in data fabric and network architecture may play an increasingly vital role in sustaining progress.