OUR MISSION
Our linear co-packaged optics significantly outpace existing solutions in speed, density and integration, while also reducing power consumption, allowing scale-up and scale-out architectures for training and inference continue their hyper-linear growth as the industry advances from generative AI to agentic AI to physical AI.
PRODUCTS
"CPUs, GPUs, and TPUs used in HPCs and AI Clusters are starved for bandwidth. A 10x increase is needed now, and another 10x will be needed for next-generation architectures based on disaggregated compute and storage."
LightCounting
Cras leo lorem, gravida vel tristique nec, tempor a urna.
Cras leo lorem, gravida vel tristique nec, tempor a urna.
Cras leo lorem, gravida vel tristique nec, tempor a urna.
Cras leo lorem, gravida vel tristique nec, tempor a urna.
MULTIPURPOSE
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Cras leo lorem, gravida vel tristique nec, tempor a urna. Suspendisse scelerisque finibus sapien, quis posuere lacus tincidunt ut.
INTRODUCING
Optimized for low-latency, distance-independent operation, the XT1600 is perfect for AI scale-out and datacenter networks. Its topology flexibility and industry-leading density, combined with low power consumption, make it the ideal solution for the most depending applications.
INTRODUCING
Nitro extends the Nubis linear E2E approach to copper. Excellent linearity, with no user adjustment required, preserves the prevalent usage model of passive cable in scale-up networks while noise optimization provides superior link performance.
our innovations and achievements
Copyright Nubis Inc. © 2025 All Rights Reserved