PCBs: Powering AI Servers 2025

With the rapid advancement of artificial intelligence technology, the AI server market is experiencing unprecedented growth. Within this hardware ecosystem, printed circuit boards (PCBs) play a critical role as the structural foundation for electronic components and the provider of electrical connections. The stringent demands of AI servers for high-performance computing, high-speed data transmission, and efficient thermal management are reshaping the technical standards and market landscape of the PCB industry. From traditional multilayer boards to high-end high-density interconnect (HDI) boards, and emerging integrated circuit substrates, PCB technology is emerging as a key factor that either constrains or propels AI computing capabilities. This article will delve into the technological evolution, market dynamics, and future trends of PCBs within the AI server market.

Revolutionary Demands and Challenges of AI Servers on PCB Technology

The core feature of AI servers lies in their ability to process massive amounts of data through parallel computing, imposing more stringent technical demands on PCBs than traditional servers.

AI servers typically incorporate multiple accelerator cards such as GPUs and TPUs. These chips feature an enormous number of pins and extremely high signal transmission rates. Therefore, motherboards and accelerator cards require ultra-high-layer PCBs with 20 or even 30+ layers, along with HDI technology featuring line widths/spacing below 100μm. This high-density design ensures tens of thousands or even hundreds of thousands of connection points fit within limited space while minimizing signal transmission paths to boost data processing speed.

Data transfer rates between processors, memory, and accelerator cards in AI servers have surpassed 100Gbps and are advancing toward 200Gbps and beyond. This places extreme demands on PCB signal integrity. Through strictly controlled impedance matching, minimized insertion loss, and the use of low-loss materials (such as M6 and M7 grade high-speed materials), PCBs must guarantee high-speed signals remain distortion-free and free from crosstalk during transmission. Leading AI server PCBs now universally support PCIe 5.0 and 6.0 standards while reserving interfaces for future co-packaged optics (CPO) technology.

AI servers consume thousands of watts under heavy loads, generating immense heat. PCBs must employ high-thermal-conductivity materials (e.g., metal substrates, ceramic fillers) while incorporating active cooling structures like thermal vias and embedded heat pipes into the design. Advanced PCBs even integrate thermal management channels directly within multilayer boards to achieve chip-level precision cooling.

AI accelerator chips can reach peak power consumption of hundreds of watts, demanding massive instantaneous currents. The PCB's power distribution network (PDN) requires thick copper designs (e.g., 3oz or higher copper thickness) to reduce DC resistance and ensure voltage stability. Simultaneously, dedicated power and ground planes within multilayer boards minimize noise interference, delivering “clean” power to the chips.

These technical challenges are driving the PCB industry's transition from traditional manufacturing to high-end precision engineering. The associated technological barriers have also fostered a highly specialized market landscape.

Competitive Landscape and Supply Chain Analysis

The AI server PCB market exhibits high concentration and technology-driven characteristics, with global and regional manufacturers fiercely competing in this high-growth sector.

Currently, the high-end AI server PCB market is dominated by a handful of manufacturers possessing cutting-edge technologies. Taiwan's Zhen Ding Technology and Unimicron Technology, Japan's Kishin and Meiko Electronics, and Austria's AT&S hold leading positions in high-layer count, high-density server PCBs. These manufacturers possess deep expertise in HDI, substrate technology, and high-speed material applications, forming close collaborative development relationships with chipmakers like NVIDIA, AMD, and Intel. Their technological barriers manifest in:

• 1) profound understanding of high-speed material properties and process adaptation capabilities;
• 2) alignment and lamination techniques for ultra-high multilayer boards;
• 3) precision control in micro-via drilling and electroplating processes.

Driven by surging demand for AI servers in China, domestic PCB firms like Shenzhen South Circuit, Huatong Circuit, and Jingwang Electronics are accelerating technological upgrades to enter the AI server supply chain. Their experience in communication equipment PCB production provides a solid foundation for this market expansion. Government initiatives like the “East Data West Computing” project and policies promoting self-reliance in critical technologies further create market opportunities for local PCB manufacturers. Currently, some leading Chinese manufacturers can mass-produce 16-20 layer high-speed server PCBs and are tackling technical challenges for higher layer counts.

PCB performance largely depends on upstream materials. High-frequency, high-speed copper-clad laminates from suppliers like Shengyi Technology, Rogers, and Panasonic form the foundation for premium AI server PCBs. Simultaneously, precision processing and inspection equipment—including laser drilling machines, direct imaging systems, and automated optical inspection devices—remains dominated by a handful of international manufacturers. Supply chain stability and technological synergy have become critical factors in PCB manufacturers' delivery capabilities.

The cost of AI server PCBs significantly exceeds that of traditional server PCBs, primarily due to:

• 1) High-end laminate material costs (accounting for over 40% of raw material expenses);
• 2) Increased processing costs due to more complex manufacturing processes;
• 3) Higher risks of yield loss. As technology matures and economies of scale take effect, prices for high-end PCBs show a gradual downward trend. However, technological innovations—such as materials and processes supporting higher speeds—maintain product premiums. Currently, the unit price for high-end AI server motherboard PCBs ranges from $500 to $2,000, representing 5 to 10 times that of standard server PCBs.

Future Innovation Directions

The continuous evolution of AI computing demands will drive PCB technology toward higher performance, greater integration, and new form factors.

As AI chip computing power increases significantly each year, the data transmission bandwidth requirements between chips double every two years. PCB technology is evolving in several directions: First, it continues to enhance electrical performance to support single-channel speeds of 112Gbps or even 224Gbps; Second, developing novel substrate technologies like silicon and glass substrates to achieve finer line widths and lower transmission losses; Third, advancing the commercialization of co-packaged optics (CPO) technology, integrating optical engines within PCBs or substrates to fundamentally overcome electrical interconnect bandwidth limitations.

AI computing is shifting from single large chips toward multi-chip heterogeneous integration. In 2.5D and 3D packaging technologies, silicon interposers or redistribution layers (RDL) handle inter-chip interconnects, but ultimately still require PCB connections to the external environment. In the future, PCBs may need to directly integrate finer interconnect structures or even form integrated solutions with packaging substrates. Embedded chip technology will also embed some chips directly within the PCB, further shortening interconnect distances and enhancing overall performance.

Future AI servers will see continued increases in power density, making liquid cooling the mainstream thermal solution. PCBs must incorporate interfaces and compatible structures for cold plates and microchannel liquid cooling systems. Simultaneously, smart PCBs integrating temperature sensors and thermal expansion control structures can monitor and adjust thermal states in real time, preventing failures caused by overheating or thermal stress. In materials, high-thermal-conductivity materials like graphene and carbon nanotubes may break traditional thermal management limits.

With exponential growth in AI server deployments, PCB environmental impact is gaining attention. The industry is developing halogen-free flame retardants, recyclable substrate materials, and low-energy manufacturing processes. Environmental regulations for electronics in the EU and China will drive the PCB industry toward green manufacturing—presenting both challenges and opportunities for technological innovation.

Industry analysts project the AI server PCB market will achieve a CAGR exceeding 25% from 2023 to 2028, significantly outpacing traditional server PCB markets. Investment hotspots are concentrated in:

• 1) Expanding high-end HDI and IC substrate production capacity;
• 2) R&D and manufacturing of high-speed materials;
• 3) Advanced inspection and processing equipment;
• 4) Simulation design software and testing/validation services. Regional production capacity distribution will also adjust with shifts in AI server manufacturing locations, with PCB capacity in Southeast Asia and the Americas potentially increasing due to supply chain diversification strategies.

The PCB market for AI servers represents not only the high-end segment of electronics manufacturing but also serves as a crucial window into the evolution of AI hardware. In this technology-driven, rapidly growing market, innovation and collaboration will be the dominant themes, propelling the entire value chain—from material suppliers to end users—toward the next generation of computing.

One-Stop HDI PCB Manufacturer and Its PCB Via Filing Capabilities

If you're looking for turnkey HDI electronics manufacturing services (EMS) from hardware development to PCBA fabrication and box-build assembly, you can work with the one-stop HDI PCBA manufacturer PCBONLINE.

Founded in 1999, PCBONLINE has R&D capabilities for HDI projects and EMS manufacturing capabilities, including via filling for stacked vias. It provides 4-to-64-layer HDI PCB fabrication, assembly, and PCBA box-build assembly. You can order various HDI PCBs from PCBONLINE, such as FR4, polyimide (flexible PCB), polyimide + FR4 (rigid-flex PCB), and PTFE/Rogers (high-frequency PCB).

Here'e the PCB via filing capabilities at PCBONLINEL:

• Micriavia filling with copper: laser via size 0.1-0.125mm, priority 0.1mm
• Finished hole size for via-in-pad filling with resin: 0.1-0.9mm (drill size 0.15-1.0mm), 0.3-0.55mm normal (drill size 0.4-0.65mm)
• Max aspect ratio for via-in-pad filling with resin PCB - 12: 1
• Min resin plugged PCB thickness: 0.2mm
• Max via-filling ith resin PCB thickness: 3.2mm
• Making different hole sizes with via filling in one board: Yes
• Via filling with copper/silver: Yes

If you need HDI PCBAs or any other PCBAs requiring via filling, please send your email to PCBONLINE at info@pcbonline.com. We will provide one-on-one engineering support to you.

Conclusion

Via filling is used for creating stacked vias in HDI PCB fabrication, BGA/CSP/QFN IC packaging, and filling PCB via-in-pad with resin during multilayer PCB fabrication. If you need one-stop electronics manufacturing for your HDI PCBA project, contact the one-stop advanced PCB manufacturer PCBONLINE for high-quality PCBA and box-build solutions tailored to your project's needs.