COHERENT Q2 FY 2026 AI DATACENTER DEMAND LIFTS

Actual Shipments of Optical Modules in 2026

By 2026, the shipment volume of 800G optical modules is expected to exceed 40 million units, with demand showing a pattern dominated by North America and followed by China. Coupled with the explosive demand for AI inference and the expansion of emerging application scenarios, the high prosperity of the optical module industry will continue in 2026. Procurement teams relying on outdated 12-week forecasting models are hitting a wall. Spot-buying mixed batches introduces PAM4 firmware mismatches, causing uncorrectable FEC errors and RDMA latency spikes exceeding 50ms under. 10GBASE-T optical modules (copper-based) are projected to dominate Ethernet networks until 2026, with a 35% market share, due to their cost-effectiveness. This brochure summarizes our coverage of AI Clusters, Data Centers and Optical Networks with in-depth analysis of the market for optical transceivers, including the optical and integrated circuits (IC) used in these modules.

Custom AI Server Chips

Big Tech is shifting from Nvidia GPUs to custom AI chips to reduce inference costs and improve efficiency at scale. Broadcom and Marvell are leading the custom silicon boom, designing chips for Google, Meta, and OpenAI. Tucked away on Microsoft's Redmond campus is a lab full of machines probing the basic building block of the digital age: Silicon. (NASDAQ: AMD) is preparing to launch the next generation of its AI accelerators in the second half of 2026, introducing the Instinct MI450 alongside the Helios rack-scale platform (MI455X), both part of the MI400 series. Frontier AI attracts hundreds of billions in global investment, with governments and hyperscalers racing to lead in domains like drug discovery and autonomous infrastructure. Recent reports from The Information reveal that Apple (AAPL) is taking a significant step in the AI race by developing its first server processor specifically tailored for artificial intelligence applications.

AI Server Hardware Computing

AI servers accelerate model training and real-time inference, delivering powerful computing with CPUs, GPUs, and specialized AI accelerators. Their scalable and efficient architecture enables businesses to run AI workloads faster and more effectively. That's the job of an AI server—a custom-built system that keeps AI applications fast, scalable, and efficient. AIME is specialized in high-performance computing solutions tailored for artificial intelligence.

How to use sensors in an AI server

Sensors in AI agents act as the primary interface between the agent and its environment, enabling the system to gather real-world data for decision-making. These devices convert physical phenomena—like light, sound, temperature, or motion—into digital signals that AI algorithms. Virtual sensors can be used in any system where real-time monitoring and control are required, and where the use of physical sensors might be impractical or costly. Leveraging AI techniques can improve the accuracy and reliability of virtual sensors. Modern AI models are data-hungry, computation-heavy beasts that need specialized hardware just to function, let alone perform at their best. Today, intelligent sensor systems perform many different tasks, including speech recognition, intelligent heating control, or autonomous driving functions. What is sensor data?This article explains how a modern IIoT Gateway eliminates that complexity and creates a robust, scalable data pipeline from the machine level all the way to your ML models — without writing a single line of code.

FTTH uses a 400G AI server

Based on the 3D-mesh architecture of AI DCs, ISP optical transport and premium private line solution adds 400G ultra-high-speed planes in hotspot areas to guarantee high bandwidth and SLAs for AI computing power. These components are not mere upgrades but foundational necessities to support the data-heavy operations of AI. AI infrastructure and applications will bring new opportunities to ISPs and operators, including new traffic brought by AI DCI and AI application device-cloud synergy, as well as value-added sales of network latency brought by real-time interactive applications. The definitive guide to selecting, deploying, and maximizing 400G optical transceivers for network architects, procurement managers, and operations teams building the infrastructure that powers today's AI, cloud, and carrier networks. This article explains how 400G/800G Ethernet fabrics enable scalable, low-latency, high-bandwidth AI data center networks, addressing GPU traffic, congestion control and modern architecture needs. AI can enable more efficient network design and management, reducing costs, while improving service and flexibility – providing certain preconditions are met. How is AI changing FTTH network design? The global FTTH network design market is expected to grow from $1.

COHERENT Q2 FY 2026 AI DATACENTER DEMAND LIFTS

Actual Shipments of Optical Modules in 2026

Custom AI Server Chips

AI Server Hardware Computing

How to use sensors in an AI server

FTTH uses a 400G AI server

Get In Touch

Connect With Us

Email

Spain Office (HQ)

EU Technical Center

Headquarters (Spain)