HPC systems, used for scientific simulations, climate modeling, and drug discoy.

ery, rely on our transceivers for parallel processing across thousands of CPUs

and GPUs. 200G, 400G, and 800G optical modules are now standard in leading

HPC deployments, offering extreme bandwidth and ultra-low latency critical for

Technologieslike NVIDIA SHARP leverage these

osks sensietomicros:

high-speed interconnects to improve performance for MPi and deep learning op

erations by offloading collective operations to the network

Cloud service providers utilize our transceivers to support rapidly expanding com

puting resources and to enable spine-leaf architectures, ensuring scalability and

efficiency across vast networks. The high port density allowed by our QSFP-DD ane

OSFP form factors is essential for scaling out massive clusters and managing the

growth of bandwidth demand. Our 800G OSFP 2xFR4 transceivers, for instance, are

ideal for cross-data center interconnections and HPC scenarios, supporting stable

transmission up to 2km.

For enterprise data centers, our transceivers enable network updates and support

growing demands for higher bandwidth, offering solutions for Al inference workloads

and general data center traffic, RoCE-enabled solutions, compatible with traditiona

Ethernet infrastructure, provide a cost-effective way for enterprises to enhance per

formance without a complete network overhaul.