Delos Data raises over $100 million to cut AI data center bottlenecks

Delos Data, a startup founded by Intel veterans, has raised more than $100 million to develop networking chips and software aimed at reducing data-transfer bottlenecks inside increasingly complex AI data centers. The company argues that GPUs, CPUs and other accelerators can become costly idle assets when they wait for data to move among computing, memory and storage systems. Its technology is designed to support data centers using mixed hardware from Nvidia, AMD, Cerebras and other suppliers, rather than relying on a single, homogeneous infrastructure stack. Delos introduced its Nonstop AI Reference Architecture and Data Interface, which it says can deliver 10 times lower latency and 10 times greater efficiency, alongside previously announced cluster and server products. The funding will support expanded hardware and software engineering, product development and sales as AI inference and agentic workloads increase demand for faster, more flexible data movement.