Specifications
Product
Falcon 6048
Type
PCIe Gen6 GPU-accelerated memory & storage server
NVMe drive count
up to 44 × E1.S NVMe SSDs
GPUs
up to 2 GPUs
Network cards
up to 4 NICs
Throughput
over 200 million IOPS
Key feature
Direct GPU-accelerated access across NVMe devices
Applications
Generative AI, vector databases, GNN workloads
Co-developed with
AIC
Demonstrated at
NVIDIA GTC 2026
Vendor
H3 Platform
Benefit
Eliminates I/O bottlenecks for AI-native data centers
Overview
H3 Platform and AIC developed the Falcon 6048, a PCIe Gen6 GPU-accelerated memory and storage server shown at NVIDIA GTC 2026. It supports up to 44 E1.S NVMe SSDs, 2 GPUs and up to 4 NICs while delivering over 200 million IOPS.
By enabling direct GPU-accelerated storage across the NVMe devices, Falcon 6048 removes the host-CPU I/O path that bottlenecks generative AI, vector-database and graph-neural-network workloads.
Key Benefits
200 M+ IOPS with direct GPU storage access, 44 E1.S NVMe bays in a PCIe Gen6 platform, and 2 GPU + 4 NIC expansion for AI-native data centers.
Applications
Generative AI inference and RAG pipelines, vector databases, graph neural networks, and AI-native data-center storage tiers.
Request a Quote — H3 PLATFORM FALCON 6048 — PCIE GEN6 GPU-ACCELERATED STORAGE SERVER
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →