Tagged articles
optical modules
19 articles · Page 1 of 1
Google's OIF workshop report analyzes how AI model training across tens of thousands of xPUs drives 448Gbps/lane interconnects, comparing NVIDIA NVLink and Google ICI architectures, evaluating PAM4 through PAM12 modulation trade-offs, and proposing CPC co-packaged copper and OSFP3.2T modules as key hardware solutions for scale-up and scale-out networks.
448GbpsAI clustersPAM6
0 likes · 7 min read

The article explains how Google’s TPU v7 supercomputer uses a simple yet powerful networking scheme—1.5 optical modules per TPU for intra‑rack communication and an additional 2.6 modules per TPU for inter‑rack high‑speed links—enabling massive AI model training with balanced cost and performance.
AI supercomputingGoogleTPU
0 likes · 13 min read

DeepSeek V3 cuts training to 2.788 M H800 GPU‑hours with FP8 mixed‑precision and a fully optimized framework, slashes token costs by 96% versus ChatGPT O1, and its efficient inference and model‑compression techniques are reshaping AI‑agent development, spurring demand for low‑latency, high‑bandwidth optical modules and edge‑computing infrastructure.
AICommunication IndustryDeepSeek
0 likes · 5 min read

The rapid growth of AI, HPC, and cloud workloads is driving exponential demand for 400G, 800G, and even 1.6T optical modules, prompting a shift in packaging technologies, modulation schemes, and parallel‑lane architectures to meet higher bandwidth and lower‑cost requirements in modern data centers.
AI computeHigh-Speed Networkingdata center
0 likes · 8 min read

The rapid growth of AI large‑model training and inference is pushing data‑center networks toward non‑blocking Fat‑Tree topologies and ultra‑high‑speed 400G/800G optical modules to meet massive bandwidth, low‑latency, and multi‑task concurrency requirements.
400G800GAI
0 likes · 8 min read

This article analyzes the factors influencing optical‑module requirements for AI GPU clusters, compares four typical network configurations for A100 and H100 SuperPODs, and provides step‑by‑step calculations that reveal the projected market demand for 200G, 400G, and 800G modules in 2023‑2024.
AIGPUSuperPod
0 likes · 14 min read

This article analyzes the evolving data‑center network architectures for large AI clusters, detailing leaf‑spine and Fat‑Tree designs, NVLink interconnects, and calculating the precise optical‑module requirements for NVIDIA A100, H100, and GH200 deployments, while also comparing industry examples from Meta, AWS, and Google.
AI clustersFat-TreeNVLink
0 likes · 12 min read

The NVIDIA L40S GPU, built on the Ada Lovelace architecture with 48 GB GDDR6 memory and 846 GB/s bandwidth, delivers over 1.45 PFLOPS tensor performance and superior FP16/FP32 efficiency for generative AI training and inference, while its lower power and GDDR6 design may influence demand for mid‑range optical modules in data centers.
GPUGenerative AIL40S
0 likes · 8 min read

This article explains how optical modules convert electrical signals to light and back at the physical layer of fiber‑optic networks, detailing their internal components, key performance indicators such as transmit power and extinction ratio, various form‑factor and wavelength classifications, naming conventions, common failure causes, and practical protection measures.
fiber opticsmodule failure preventionnetwork hardware
0 likes · 15 min read

The article explores how modern cloud computing relies on high‑speed optical modules for massive data transmission, detailing the shift from proprietary, closed‑box solutions to open, scalable designs, the rapid development of 40G‑400G optics, quality‑control challenges, and future demands from AI, 5G, and edge computing.
cloud computingnetwork infrastructureoptical modules
0 likes · 10 min read

This article explains the operating principles, structural components, key performance metrics, various classifications, naming conventions, common failure causes, and protective measures for optical modules used in fiber‑optic communication systems.
electro-optical conversionfiber communicationhardware protection
0 likes · 15 min read

The article explains the leaf‑spine network architecture introduced by Facebook, details its three layers (cabinet, Leaf, Spine), quantifies the required servers, switches and optical modules per cabinet, and analyzes the associated CAPEX for ultra‑large cloud data centers.
CapExLeaf-Spineoptical modules
0 likes · 11 min read

This article provides a comprehensive overview of optical module technology, explaining basic concepts, packaging types, classification categories, and essential performance metrics, and concludes with a download link for additional resources.
Technology Fundamentalsfiber opticsoptical modules
0 likes · 2 min read

The article analyzes how the rapid rollout of 5G and the expanding capacity of modern data centers are creating a long‑term boom for optical modules, detailing market forecasts, network architecture shifts, supply‑chain consolidation, domestic chip advancements, and the competitive positioning of key industry players.
5GChipdata center
0 likes · 11 min read

The article reviews the shift from 10G to 25G and higher‑speed optical modules, comparing 25G native chips with 10G over‑clock solutions, discussing supply‑chain dynamics, network upgrade paths, 400G packaging standards, and the growing demand for high‑speed transceivers in modern data‑center and telecom infrastructures.
25G400Gnetwork infrastructure
0 likes · 8 min read

The article details the design of high‑density silicon‑photonic 400G/100G optical modules using 25.6T ASIC chips, compares integrated and separated packaging schemes, presents test results, and explains why silicon photonics is crucial for future high‑performance data‑center networking.
400GASIChigh density
0 likes · 8 min read

This comprehensive analysis examines the optical module industry's supply chain, technology trends, market segmentation across telecom, access, and data center applications, the impact of 5G and cloud growth, cost pressures, R&D investment, and the strategic shift toward high‑end and domestically sourced components.
5GPhotonic IntegrationR&D
0 likes · 28 min read

The article analyzes the rapid growth of 400G optical modules, comparing OSFP and QSFP‑DD form factors, detailing Cisco's new Nexus 400G switch lineup and Broadcom's BCM87400 PAM‑4 PHY, and explains how these technologies are set to meet expanding data‑center bandwidth demands while highlighting market adoption trends and potential cost trade‑offs.
400GBroadcomCisco
0 likes · 16 min read

This article examines how data center operators and the supply chain are advancing toward 400 Gbps Ethernet, detailing the technical innovations, market forces, and future challenges that shape high‑speed networking, optical modules, and ASIC development for ultra‑large scale data centers.
400GASICEthernet
0 likes · 12 min read
