ScaleFlux is focusing on the data pipeline bottleneck in AI data centers, where memory and storage speeds fail to match the rapid performance growth of GPUs and TPUs. CEO Hao Zhong noted that processor performance has increased nearly 1,000 times over the past eight years, creating a significant efficiency gap.
To address this, ScaleFlux has introduced new controller chips designed for the second and third stages of the data pipeline. These include controllers specifically built for PCIe Gen 6 SSDs and Compute Express Link memory, aimed at improving data access for inference and agentic workloads.
As AI applications shift from training to inference, the demand for higher capacity and bandwidth in external DRAM and SSDs has intensified. ScaleFlux positions its hardware as a solution to ensure that expensive compute resources do not remain idle while waiting for data.
