Graid Technology Achieves Industry-First 100 Million IOPS on a Protected RAID 5 Volume for GPU-Initiated I/O
Storage resilience built to meet the performance demands of agentic AI
SUNNYVALE, CA / ACCESS Newswire / July 15, 2026 / Graid Technology today announced an industry-first result: 100 million random 512-byte IOPS on a protected storage volume built from 32 KIOXIA XD8 NVMe SSDs. The milestone highlights Graid Technology's ability to deliver extreme storage performance with data protection for next-generation GPU-driven computing environments.
Beyond the raw I/O result, Graid Technology demonstrated that its protected RAID 5 storage volume can support demanding AI training workloads without measurable performance compromise. In benchmark testing with a WholeGraph training workload, this configuration delivered training performance on par with the baseline environment, showing that high-speed data protection does not have to come at the expense of application throughput.
This result marks an important milestone for AI infrastructure, where modern workloads, including graph analytics, large-scale training pipelines, retrieval-intensive inference, and emerging agentic frameworks, increasingly depend on massive volumes of fine-grained, random I/O to keep GPUs fully utilized. As AI systems scale in model complexity, data size, and concurrency, storage can no longer be treated as a secondary consideration behind compute.
This level of performance shows that resilient NVMe storage can now operate at the scale current and future AI deployments require. This is especially significant for environments that require both extreme performance and enterprise-class data protection, where downtime, rebuild risk, or storage bottlenecks can directly impact productivity and infrastructure efficiency.
"With this milestone, Graid Technology continues to prove that customers do not need to choose between performance and protection," said Leander Yu, CEO of Graid Technology. "As AI infrastructure evolves toward more data-intensive and agent-driven workloads, storage architectures must keep pace with GPU performance. These results show that Graid Technology's GPU-accelerated, and now GPU-initiated, approach to storage performance and data protection can deliver the resiliency enterprises require, while sustaining the ultra-low-latency, high-IOPS access that next-generation AI workloads demand."
To learn more about how Graid Technology is inventing the future of storage for AI, enterprise, and high-performance computing, visit graidtech.com.
_________________________________
About Graid Technology
Graid Technology is the creator of SupremeRAID™, the world's first and only GPU-based RAID, and the global steward of Intel® Virtual RAID on CPU (Intel® VROC). Graid Technology delivers flexible RAID solutions that maximize NVMe performance while ensuring resilient, scalable data protection for modern data infrastructure. Graid Technology is headquartered in Silicon Valley with global operations and R&D in Taiwan. We are inventing the future of storage. Visit graidtech.com.
Media Contact:
Andrea Eaken, Sr. Director of Marketing, Americas & EMEA
[email protected]
.jpg)