news

• AI clusters are driving data centers into a new era of high-speed data transmission. As GPUs, AI ASICs, and switch chips continue to advance rapidly, the volume of data exchanged within systems and between racks is increasing dramatically. For large-scale AI training and inference environments, the key challenge is no longer simply increasing compute power, but enabling data to move reliably between nodes, switches, optical modules, and servers with lower latency, lower loss, and higher energy efficiency.

• As per-lane speeds move from 112G to 224G, and further toward 448G, every segment of the high-speed channel becomes increasingly sensitive. PCB traces, connectors, cages, cables, and thermal structures directly affect signal integrity, power consumption, and platform scalability. As a result, the internal interconnect architecture of a switch is no longer merely a mechanical design consideration; it has become a core factor in determining whether AI high-speed communication can continue to scale.

• Against this backdrop, VLC (Vertical Line Card) is emerging as a critical architectural shift for next-generation high-speed switches. By adopting a vertical configuration, VLC rearranges the relative positions of the Switch ASIC, I/O connectors, and optical modules, creating a shorter and more direct signal path to maintain stable transmission under higher bandwidth and power conditions.

• ---

The Key Values of VLC

• 1. Shorter High-Speed Paths for Enhanced Signal Integrity
• In large-scale AI clusters, switches must continuously handle high-bandwidth, high-density east-west traffic. As per-lane speeds move beyond 224G, signals become far more sensitive to path length and impedance continuity.
• •Reduced Channel Loss: By shortening the electrical path between the ASIC and optical modules, VLC minimizes losses introduced by PCB traces, vias, and transition interfaces.
• •Improved Margins: This significantly improves insertion loss, return loss, and crosstalk, allowing switches to reliably support 224G, 448G, and future higher speeds while mitigating signal degradation.

• 2.Lower Power Consumption and Higher Transmission Efficiency
• The energy pressure in AI data centers does not come only from compute chips; a significant portion stems from the networking infrastructure. Longer paths increase the system's dependence on retimers, DSPs, or stronger equalization, raising power and thermal loads.
• •Diminished Retimer Dependence: The shortened path reduces the need for heavy signal compensation.
• •Power-Efficient Transmission: This architectural optimization enhances overall switch power efficiency and alleviates thermal pressure, making VLC a vital design direction for reducing the energy cost of data movement.

• 3.More Room for Thermal Design in High-Density Systems
• AI switches concentrate multiple massive heat sources—such as switch ASICs, high-speed optical modules, and power assemblies—within highly confined spaces.
• •Optimized Cooling Paths: Through its vertical architecture, VLC grants system designers more freedom to integrate advanced liquid cooling modules, optimize airflow guidance, and isolate heat between optical modules and ASICs.

•Enhanced Serviceability: It improves access for module maintenance and replacement, ensuring long-term stable operation under high power density.

• 4. Greater Mechanical Flexibility and Platform Scalability
• VLC breaks through the physical layout limitations of traditional horizontal board architectures.
• •High-Density I/O Layout: Switches can integrate higher-density I/O configurations within limited front-panel and system space.
• •System-Level Integration: It helps platforms maintain robust scalability and manufacturability under higher radix and greater thermal demands, allowing data center network capacity to scale seamlessly alongside compute expansion.
• 

• ---

• Optical-Copper Parallelism: A New Design Logic for AI Communication

The evolution of AI high-speed communication is not about one transmission medium replacing another. Instead, it is about enabling optical and copper interconnects to work together synergistically across different application points.

What is "Optical-Copper Parallelism"?

Optical transmission is well-suited for longer distances and inter-rack connections, while copper interconnects continue to excel in short-distance, low-latency, and cost-efficient board-level environments.

In a VLC architecture, the distance between optical modules and the ASIC is minimized, creating a highly direct electrical path. This allows copper interconnects to maintain exceptional signal integrity over shorter distances, while optical modules handle high-speed external transmission. VLC does not eliminate copper; rather, it empowers copper to play a more efficient, critical role inside AI high-speed switches alongside optical paths.

• Core Pillars of Next-Generation System Design:
•      ● Optical Transmission: Supports high bandwidth, longer distances, and external connectivity.
•      ● Copper Interconnects: Handles short-distance, low-latency, high-density board-level connections.
•      ● Thermal Structures: Ensures stable operation under extreme power density (Air/Liquid Cooling).
•      ● Mechanical Design: Seamlessly integrates modules, cables, boards, and cooling infrastructure.

     ● Customization: Optimizes individual platforms according to space, power, and transmission demands.

• True competitiveness in future AI high-speed communication will stem from the comprehensive integration of optical, copper, thermal, mechanical, and advanced manufacturing capabilities.

• ---

• NEXTRON Supports Optical-Copper Parallelism with System-Level Integration

Within the VLC architecture ecosystem, NEXTRON brings strong system-level integration capabilities to next-generation high-speed communication platforms. From a mechanical design perspective, we help ensure durability under repeated mating cycles while supporting broad module compatibility. For high-density module configurations, we help maintain stable high-speed signal transmission and minimize signal loss. To address demanding thermal requirements, NEXTRON also provides structural design and integration services for both air cooling and liquid cooling systems.

• Leveraging years of expertise in connector technology, NEXTRON offers highly flexible, customized interconnect and mechanical solutions. From connector components, high-speed cables, and module structural parts to full-system integration, we support customers based on their specific routing conditions, installation requirements, and mass production needs. NEXTRON empowers customers to accelerate the deployment of next-generation high-speed communication platforms and lead the next era of AI transmission.