AI News AI资讯 3d ago Updated 3d ago 更新于 3天前 49

ShanghaiTech PhD Team Accelerates Mass Production of Domestic Ultra-High-Speed Photodetection Chips, Securing Million-Level Angel Investment 36氪首发丨上科大博士团队加速国产超高速光电探测芯片量产,获千万级天使轮融资

Shanghai Kuailin Optoelectronics secured tens of millions in angel funding led by Xiaomiao Langcheng to accelerate mass production of domestic ultra-high-speed photodetection chips. The team, originating from ShanghaiTech University, has achieved world-class performance with UTC-PD chips exceeding 200 GHz bandwidth and 0.8 A/W responsivity, addressing critical bottlenecks in AI optical interconnects. The company focuses on InP-based photodetectors for 800G, 1.6T, and 3.2T datacom applications, a 上海快粼光电完成数千万元天使轮融资,加速国产超高速InP光电探测芯片量产,填补国内高端收端光芯片空白。 团队源自上海科技大学,基于UTC-PD结构实现超200GHz带宽与0.8 A/W响应度,技术指标达国际第一梯队。 产品精准卡位AI算力集群互联瓶颈,适配800G/1.6T/3.2T高速光模块及LPO/CPO等下一代数据中心架构。 核心挑战在于实验室技术向量产线的工艺Know-how移植,公司计划2027年实现核心产品规模化交付。 投资方看好其在单波200G/400G PD芯片及6G太赫兹通信领域的长期潜力,视为算力基础设施关键一环。

65
Hot 热度
70
Quality 质量
75
Impact 影响力

Analysis 深度分析

TL;DR

  • Shanghai Kuailin Optoelectronics secured tens of millions in angel funding led by Xiaomiao Langcheng to accelerate mass production of domestic ultra-high-speed photodetection chips.
  • The team, originating from ShanghaiTech University, has achieved world-class performance with UTC-PD chips exceeding 200 GHz bandwidth and 0.8 A/W responsivity, addressing critical bottlenecks in AI optical interconnects.
  • The company focuses on InP-based photodetectors for 800G, 1.6T, and 3.2T datacom applications, aiming to solve the "interconnection bottleneck" in AI clusters and reduce reliance on imported high-end optical chips.
  • Strategic emphasis is placed on bridging the gap between laboratory performance and manufacturing yield, targeting scalable delivery by 2027 while maintaining competitiveness in 6G terahertz communication technologies.

Why It Matters

This development addresses a critical supply chain vulnerability in China's AI infrastructure, where high-end optical receiver chips have long been dominated by foreign entities. As AI data centers transition from compute-bound to interconnect-bound architectures, the demand for low-power, high-bandwidth optical components is exploding. Kuailin’s ability to localize this technology ensures greater supply chain security and supports the rapid scaling of next-generation data center networks (LPO/CPO) essential for advanced AI workloads.

Technical Details

  • Core Technology: Utilizes Modified Uni-Travelling-Carrier Photodiode (MUTC/UTC-PD) structures that simplify carrier transport from two types to one (electrons), significantly enhancing bandwidth without compromising responsivity.
  • Performance Metrics: R&D samples demonstrate bandwidth exceeding 200 GHz and responsivity of 0.8 A/W, placing them in the first tier globally.
  • Product Roadmap: Focuses on single-wave 100G, 200G, and next-generation 400G photodetectors compatible with 800G, 1.6T, and 3.2T optical modules.
  • Manufacturing Strategy: Establishes a full-stack development system covering device structure design, III-V material epitaxy, and packaging integration, with a specific focus on transferring process know-how to foundry lines for high-yield mass production.
  • Application Scope: Targets high-speed optical interconnects for AI clusters, Data Center Interconnects (DCI), and future 6G terahertz communications.

Industry Insight

  • Supply Chain Localization: The success of domestic teams like Kuailin in achieving world-class metrics signals a shift from imitation to innovation in China's photonics sector, reducing geopolitical risks associated with semiconductor exports.
  • Interconnect as the New Frontier: Investors and manufacturers must prioritize optical interconnect solutions over pure compute upgrades, as bandwidth density and power efficiency become the primary constraints for scaling AI clusters.
  • Yield Over Peak Performance: For hard tech startups, the critical barrier to commercialization is not just lab-scale performance but the ability to maintain high yields during mass production. Companies that solve the "lab-to-fab" translation problem will capture significant market share in the upcoming 1.6T/3.2T module cycles.

TL;DR

  • 上海快粼光电完成数千万元天使轮融资,加速国产超高速InP光电探测芯片量产,填补国内高端收端光芯片空白。
  • 团队源自上海科技大学,基于UTC-PD结构实现超200GHz带宽与0.8 A/W响应度,技术指标达国际第一梯队。
  • 产品精准卡位AI算力集群互联瓶颈,适配800G/1.6T/3.2T高速光模块及LPO/CPO等下一代数据中心架构。
  • 核心挑战在于实验室技术向量产线的工艺Know-how移植,公司计划2027年实现核心产品规模化交付。
  • 投资方看好其在单波200G/400G PD芯片及6G太赫兹通信领域的长期潜力,视为算力基础设施关键一环。

为什么值得看

本文揭示了AI算力扩张背景下,光互联从“算力瓶颈”转向“互联瓶颈”的关键节点,突出了国产高端光芯片自主可控的战略紧迫性。对于关注半导体、光通信及AI基础设施的从业者而言,它提供了关于UTC-PD技术突破、量产难点及商业化路径的一手行业洞察。

技术解析

  • 核心技术架构:采用III-V族材料(InP)外延生长,基于改进型UTC-PD(Uni-Traveling-Carrier Photodiode)结构。通过优化器件设计,将载流子从“两种”变为“一种”,利用电子快速输运优势,解决了传统结构中带宽与响应度难以兼得的矛盾。
  • 关键性能指标:研发样品实现超过200 GHz的带宽和0.8 A/W的响应度,这两项核心指标在国际范围内处于第一梯队,显著提升了光接收链路的传输速率和灵敏度。
  • 应用场景适配:产品主要面向数通领域的高速光接收需求,支持单波100G、200G及下一代400G光探测芯片,广泛适用于800G、1.6T、3.2T高速光模块,以及LPO(线性驱动可插拔光学)、CPO(共封装光学)等先进封装架构。
  • 量产与工艺挑战:技术难点已从单纯的器件设计转向“技术”与“产线”的适配。关键在于将外延、制程、测试等环节的工艺Know-how完整移植到代工厂量产线,确保批量出货时的良率控制和产品一致性,而非仅追求实验室单颗性能。

行业启示

  • 光互联成为AI基建新焦点:随着AI集群规模扩大,光模块迭代加速(向1.6T/3.2T演进),高速光探测芯片作为接收端核心器件,其国产化率和性能直接决定算力网络的扩展上限,是供应链安全的“卡脖子”关键环节。
  • 硬科技创业需跨越“量产鸿沟”:高校背景团队虽具备顶尖学术成果,但商业化的最大壁垒在于工程化能力。成功的关键在于建立从设计到封装集成的完整体系,并深度整合代工资源,解决良率与可靠性问题,才能拿到进入主流供应链的“门票”。
  • 技术路线的长期价值布局:虽然当前聚焦于光通信这一刚性爆发市场,但UTC-PD技术在6G太赫兹通信、光电量测等领域具有通用性和延展性。企业需在满足短期现金流的同时,保持对前沿通信技术的长期研发投入,以捕捉下一代通信标准的红利。

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Chip 芯片 Funding 融资 Research 科学研究