Dongshan Precision: Dual Engines of 800G/1.6T Optical Modules and AI PCBs

Overview

This report presents minutes from management discussions with Dongshan Precision following the Asia Communications Technology Summit, published by Goldman Sachs. Although Goldman Sachs does not currently provide formal coverage or ratings for Dongshan Precision, the report conveys clear positive fundamental signals: the company is leveraging a dual-engine strategy of 'optical modules + AI PCBs' to position itself within AI infrastructure, and management expresses strong confidence in meeting clients’ rapidly growing demand for AI computing support. Key focal points include the pace of migration toward next-generation 1.6T/3.2T optical modules and how AI PCB capacity expansion optimizes the product portfolio.

Core views

The optical module business is advancing toward higher-speed generations. Management confirmed that the company’s product line already spans the full range from 10G to 1.6T, with next-generation 3.2T optical modules currently under development. Future strategic focus will center on 800G, 1.6T, and higher-speed products, scaling and adjusting capacity continuously to align with AI-driven demand surges. Technologically, the company masters both mainstream approaches—EML and Silicon Photonics (SiPh)—and is committed to enhancing in-house optical chip performance to strengthen supply chain autonomy and product competitiveness. The AI PCB business is entering a phase of dual optimization in both capacity and product structure. The company emphasizes its end-to-end delivery capability—from AI PCBs to high-speed optical modules—enabling it to offer low-latency, high-throughput integrated solutions for data centers. In response to robust demand from generative AI, management has outlined a clear PCB capacity expansion plan and will continue optimizing its product mix to increase market share and value capture in data centers and AI servers. At the industry level, the rigid demand for high-speed data transmission driven by generative AI is reshaping the entire supply chain. Goldman Sachs believes that as compute density per rack increases, all categories of AI infrastructure components will experience strong growth, significantly expanding the total addressable market (TAM). With its diversified portfolio spanning PCBs (FPC/RPCB/Rigid-Flex), precision components, and optical modules—and especially its vertical integration enabled by the privately held subsidiary Source Photonics—Dongshan Precision is well-positioned to benefit from this wave of AI infrastructure investment.

Analysis framework

The report employs a typical 'conference minutes + cross-industry validation' analytical framework. First, it directly quotes management statements made at the summit to obtain timely, first-hand insights into capacity planning, technology roadmaps, and customer demand. Second, it contextualizes these company-specific developments within broader macro industry trends, linking Dongshan Precision’s progress in optical modules and PCBs to the capital expenditure cycle of data centers driven by generative AI, thereby assessing whether its growth narrative aligns with sector-wide beta. Finally, it evaluates the company’s competitive moat and risk resilience in a fast-evolving environment by examining its vertical integration (e.g., in-house optical chips) and multi-path technology strategy.

Methodology notes

• Industry/Supply Chain Analysis FrameworkUpstream-Midstream-Downstream Transmission

  Transmission mechanism of AI compute demand to hardware layers

  The report treats the generative AI boom as an upstream catalyst and analyzes how it propagates downstream along the path of 'data transmission demand → volume and price uplift for optical modules/PCBs.' Understanding this transmission chain helps investors discern whether the company’s growth stems from short-term order fluctuations or long-term structural industry trends.

• Competitive Strategy FrameworkMoat / competitive advantage

  Competitive barriers built through vertical integration and multi-technology-path strategy

  The report particularly highlights the significance of the company’s vertical integration ('optical chips + optical modules') and its concurrent mastery of both EML and Silicon Photonics technology paths. In the rapidly evolving AI hardware landscape, such integration and technological redundancy are critical moat indicators that mitigate single-technology risks, ensure supply security, and enhance customer stickiness.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Dongshan Precision (002384.SZ)

  Beneficiary: AI infrastructure expansion drives demand for optical modules and AI PCBs

  Strengths

  Vertical integration from optical chips to modules; dual expertise in EML and Silicon Photonics; full-speed product coverage from 10G to 1.6T; end-to-end AI PCB delivery capability

Key data

• Optical Module Speed Range10G - 1.6TNext-generation 3.2T products are under development; current capacity expansion focuses on 800G/1.6T
• Optical Module Technology PathsEML + Silicon Photonics (SiPh)Mastery of both mainstream technology approaches reduces reliance on a single path
• PCB Product TypesFPC, RPCB, Rigid-FlexApplications span consumer electronics, automotive, data centers, and AI servers
• Optical Module SubsidiarySource PhotonicsPrivately held subsidiary enabling vertical integration from optical chips to modules

Impact & implications

For investors tracking the AI hardware supply chain, these minutes confirm Dongshan Precision’s strategic positioning in the AI infrastructure upgrade cycle. The company is not only scaling existing capacity to meet current demand but also proactively capturing future opportunities through 1.6T/3.2T R&D and AI PCB portfolio optimization. Its end-to-end solution capability differentiates it from pure-play assemblers or single-component suppliers, potentially enabling higher market share and pricing power among data center clients. This also signals that the momentum behind AI compute deployment has meaningfully extended beyond core GPU/ASIC chips to supporting components like optical interconnects and carrier boards.

What to watch

• Ramp-up progress and customer validation status for 800G/1.6T optical module capacity
• Development milestones and mass production timeline for 3.2T optical modules
• Pace of new AI PCB capacity release and effectiveness of product mix optimization
• Performance improvements and adoption rate of in-house optical chips