Yanrong Technology and Intel Unveil F9000X Turbo All-Flash Storage Appliance for AI Workloads
On September 22, 2026, at the Intel Connection 2026 event in Suzhou, Yanrong Technology and Intel jointly unveiled the F9000X Turbo, a new all-flash storage appliance designed for large-scale AI workloads. The system is powered by sixth-generation Intel Xeon processors, PCIe 5.0 NVMe flash media, and supports 400Gbps InfiniBand/RoCE networking.
According to Yanrong, a three-node F9000X Turbo cluster delivers 540 GiB/s aggregate bandwidth and 9 million IOPS, supporting high-throughput, high-concurrency data access and large-scale checkpoint reads/writes in AI workloads. For a 2PB all-flash configuration using 15.36TB high-density NVMe SSDs, the company said the number of devices could be reduced from 16 to 8, cutting node count by 50% and helping lower overall TCO compared with the previous-generation approach.
The appliance also posted multiple top results in the MLPerf Storage v3.0 benchmark, including a 544 GiB/s aggregate read bandwidth on the 3D U-Net workload with a three-node cluster. Yanrong said this marks the third consecutive year its all-flash AI storage has set world records in MLPerf Storage tests, following 2024 and 2025. The company said it will continue working with Intel and other partners on storage innovation for large-model training and inference scenarios.
