SAN MATEO, Calif. — August 27, 2025 — Leads & Copy — Alluxio, the AI and data-acceleration platform, reported strong results for the second quarter of its 2026 fiscal year. The company launched Alluxio Enterprise AI 3.7 during the quarter, which delivers sub-millisecond TTFB (time to first byte) latency for AI workloads accessing data on cloud storage.
Alluxio reported new customer wins across industries and AI use cases, including model training, deployment, and feature store query acceleration. MLPerf Storage v2.0 benchmark results underscored Alluxio’s leadership in AI infrastructure performance, achieving GPU utilization and I/O acceleration across training and checkpointing workloads.
CEO Haoyuan Li said Alluxio Enterprise AI 3.7 eliminates cloud storage performance bottlenecks in AI infrastructure. The company’s distributed caching technology delivers value to customers building and serving AI models at scale, Li said.
Alluxio Enterprise AI 3.7 features include ultra-low latency caching for cloud storage, enhanced cache preloading, and role-based access control for S3 access. In the first half of 2025, Alluxio AI saw customer growth exceed 50% compared to the previous period.
Alluxio’s distributed caching architecture maximizes GPU efficiency and AI workload performance. In the MLPerf Storage v2.0 benchmarks, ResNet50 achieved 24.14 GiB/s supporting 128 accelerators with 99.57% GPU utilization, Llama3-8B achieved 4.29 GiB/s read and 4.54 GiB/s write, and Llama3-70B achieved 33.29 GiB/s read and 36.67 GiB/s write.
Alluxio Enterprise AI version 3.7 is now available.
Beth Winkowski, Winkowski Public Relations, LLC for Alluxio, 978-649-7189, beth@alluxio.com
Source: Alluxio
