华福证券发布研报称，随着AI大模型训练进入"万卡集群"时代，传统基于电交换的网络架构正面临网络 功耗高、带宽要求高的瓶颈。OCS交换机的"全光交换"机制从根本上保证网络的极致带宽和前瞻兼容 性。目前商用OCS交换机主流为MEMS技术方案，占比超过70%。核心环节上，国内涌现出一批优秀的 OCS核心零部件供应商，业务涵盖MEMS振镜、透镜、光器件等核心环节。 MEMS方案是目前市场的主流，占比超过70%，尤其在谷歌等企业中被广泛采用。谷歌自2022年起在其 传统数据中心中导入该技术，并于2023年应用于TPU V4，计划在2025年的TPU V5P中进一步推广。根 据QY Research，2024年MEMS技术方案的OCS交换机市场规模为192.36百万美元，约占整个全光交换 (OCS)交换机市场52.49%的份额，预计2031年将达到836.79百万美元，届时全球占比将减少至41.38%。 国产厂商积极参与OCS交换机研发与制造 整机方面：华为全联接大会2024上，华为重磅推出了其最新的全光交换机——Huawei OptiXtrans DC808；凌云光参与OCS整机研发，压电陶瓷路线代表；另外中际旭创（3 ...

Core Viewpoint - Optical Circuit Switches (OCS) differ from traditional switches by enabling direct optical switching without optical-electrical conversion, leading to reduced hardware costs and power consumption [1][2][5]. Group 1: Overview of Optical Switches - OCS operates by forwarding optical signals through wavelength division multiplexing and cross-connection, with optical-electrical conversion occurring at the server side [1][2]. - The main performance indicators of optical switches include switching speed, insertion loss, return loss, durability, and reliability [2][3]. Group 2: Current Development Status - OCS technology has gained attention due to Google's initiatives, with increasing applications in data centers [5][6]. - The rise of east-west traffic in large data centers has created bandwidth bottlenecks in traditional three-layer architectures, necessitating the adoption of the Spine-and-Leaf architecture for better scalability and reduced latency [6][8]. Group 3: Market Demand and Shipment Volume - Global shipments of optical switches reached 10,000 units in 2023, with an expected increase to approximately 13,000 units in 2024, driven by growing bandwidth demands in machine learning and cloud computing [16][19]. Group 4: R&D Directions - Despite the promising outlook for optical switch technology, challenges remain, including high initial capital expenditures and signal insertion loss [14][19]. - Continuous technological advancements are anticipated to enhance performance and reduce manufacturing and deployment costs, solidifying the foundation for large-scale applications [19].