San Jose, USA – March 19, 2026 – H3 Platform, in collaboration with Micron, today demonstrated a production-ready PCIe Gen6 GPU-accelerated storage architecture at NVIDIA GTC 2026, marking a significant step forward in enabling deployable AI-native infrastructure.

Building on the breakthrough performance milestone first achieved at SC’25, the system—powered by the Falcon 6048 PCIe Gen6 platform—delivers up to 230 million IOPS through GPU-initiated storage access. The demonstration integrates Micron’s next-generation PCIe Gen6 NVMe SSDs with H3 Platform’s PCIe Gen6 switching architecture and advanced software stack, forming a fully optimized, end-to-end AI storage solution.

At GTC 2026, the showcase demonstrates a production-ready PCIe Gen6 GPU-accelerated storage architecture, marking a significant step beyond experimental validation. The joint solution enables customers to confidently deploy next-generation AI storage infrastructure in real-world environments.

The architecture is built on a GPU-centric data path that enables direct communication between NVMe storage and GPUs, eliminating traditional CPU bottlenecks and significantly improving data throughput and latency consistency. By combining Micron’s high-performance Gen6 SSDs with H3 Platform’s PCIe Gen6 switching fabric, the system delivers scalable, high-concurrency data access optimized for modern AI workloads.

Falcon 6048 serves as a platform for next-generation AI infrastructure, supporting demanding use cases such as large-scale inference pipelines, vector data processing, and graph-based analytics. The collaboration with Micron further demonstrates how tightly integrated storage and compute architectures can unlock new levels of performance and efficiency for AI-native data centers.

“We are proud to collaborate with Micron and our ecosystem partners in advancing a production-ready PCIe Gen6 GPU-accelerated storage architecture,” said Brian Pan, CEO of H3 Platform. “Falcon 6048 represents a critical milestone in the evolution of AI infrastructure—where storage is no longer a passive component, but an active participant in the compute pipeline. By enabling direct GPU-to-storage data paths at scale, we are helping customers move from proof-of-concept to production deployment, accelerating the adoption of AI-native data center architectures.”

H3 Platform will continue to work closely with Micron and ecosystem partners to drive innovation in production-ready GPU-accelerated storage and composable infrastructure, enabling scalable, high-performance solutions for the next generation of AI workloads.

---

About H3 Platform

Founded in 2016 and headquartered in San Jose, California, H3 Platform is a leading provider of Composable Infrastructure and PCIe/CXL-based memory and storage solutions. The company’s technologies enable data centers to achieve exceptional performance, scalability, and flexibility across AI and HPC applications.

For more information, visit www.h3platform.com.

---

Media Contact
For more information, please contact:
H3 Platform Media Relations
Email: william.tenn@h3platform.com | Tel: +886-2-2698-3800 #121