Goldman Sachs significantly raises PCB market size expectations: both volume and price increase + specification upgrades, with a compound growth rate expected to exceed 140% in the next two years!

Goldman Sachs has aggressively revised its potential market size (TAM) for AI PCB/CCL. According to Wind Trading Desk, Goldman Sachs analysts Chao Wang and Allen Chang's team released a research report on the PCB (Printed Circuit Board) and CCL (Copper Clad Laminate) industry in Taiwan on January 6, 2026. Goldman Sachs raised its 2027 TAM for AI PCB from the previous expectation of $17.4 billion to $26.6 billion, and for CCL from $8 billion to $18.3 billion. This adjustment implies that the compound annual growth rate (CAGR) from 2025 to 2027 will reach an astonishing 140% and 178%, respectively. This growth is not a linear extrapolation but is based on precise calculations of architectural changes: "One key reason our global team significantly raised the AI PCB/CCL TAM is that we accounted for the new PCB/CCL content added inside the VR200/300 racks. This new content will replace bridge cables (shifting from cables to PCBs) to improve the overall cost structure and enhance computing performance." For investors closely monitoring the AI hardware supply chain, this report sends a clear signal: Driven by next-generation architectures like NVIDIA's GB300, the PCB industry is undergoing a qualitative transformation from mere 'shipment growth' to a 'surge in value.' The expansion of AI infrastructure is far from over, and as technological architectures evolve, the market space (TAM) for core components is experiencing a radical revaluation. Architectural Change: Cables Disappear, PCBs Take the Stage Why such an aggressive expectation? The core logic lies in the structural changes brought about by NVIDIA's next-generation architecture (VR200/300 rack). Traditional cables are being phased out, replaced by more complex PCB/CCL components. This shift from "cables" to "boards" directly improves the cost structure of computing power expansion while significantly increasing the value of PCBs. The report specifically mentions that new mid-boards and backplane demand will explode in the second half of 2026 and the second half of 2027, with the PCB/CCL value corresponding to a single GPU experiencing exponential growth. The so-called "specification upgrade" is not just empty talk; it directly translates into real monetary technical barriers and profit moats. As mainstream PCB specifications upgrade from 24-28 layers in 2025 to over 40 layers in 2027, HDI technology specifications will also advance from 4+N+4 to 6+N+6, significantly increasing production difficulty. "We consider the loss of PCB production yield in the coming years (expected to drop from 73% in 2025 to 65%/62% in 2026/27)... This will force an increase in the long-term usage of CCL, leading to the growth rate of the AI CCL market potentially exceeding that of the AI PCB market." Yield "Trap" Boosts Raw Material Demand, Goldman Sachs Targets High-End Taiwanese Suppliers For investors, this means the industry is undergoing a brutal yet beneficial "cleansing": the more complex the board, the more expensive it is, and leading manufacturers capable of handling low yield challenges will dominate the market In this feast, GPUs remain the absolute main character, but ASICs (Application-Specific Integrated Circuits) cannot be ignored. Goldman Sachs predicts that the market size for GPU-related PCB/CCL will surge nearly tenfold within two years, while the ASIC market will also double. Although new suppliers are trying to enter the GPU market to grab a share, leading to a slightly crowded competitive landscape (with the number of PCB suppliers expected to increase from 5 to 10), the ASIC supply chain remains relatively stable. Goldman Sachs maintains a "Buy" rating for Taiwan Semiconductor Manufacturing Company (EMC), Global Unichip Corp (GCE), and Taiwan Semiconductor (TUC), and has raised the target prices across the board. For capital seeking certainty, these high-end players who master core materials and processes remain the biggest winners in this round of AI infrastructure frenzy