AI Boom Sparks High-End PCB Order Surge, Rippling Through Supply Chains
The printed circuit board (PCB) industry is experiencing an unprecedented surge in orders, with high-end products in short supply and top manufacturers’ order books booked through 2027. The boom, driven by explosive demand from AI computing power infrastructure, signals that the AI investment wave has now reached the foundational hardware layer of the global technology economy.
According to Xinhua News, the PCB industry—often called the “mother of electronic products”—is seeing high-end products being “snatched up” as AI servers demand increasingly complex circuit board architectures. Industry experts say the surge reflects a fundamental shift in how AI demand is being translated into physical infrastructure.
From Chip Concepts to Physical Hardware
The current boom represents the third phase of AI-driven industry chain transmission. As Gong Chao, a researcher at Tongji University’s National Innovation and Development Research Institute, explained: “A chain reaction around ‘computing infrastructure’ has already started, from the earlier ‘AI algorithm and large model craze,’ to ‘GPU and optical module craze,’ now transmitting to the ‘underlying basic hardware craze’ of high-end circuit boards.”
The financial results tell the story. Yibo Technology, a PCB design firm, reported first-half 2026 revenue of 709 million yuan (up 41.63% year-over-year) with net profit surging 1,561.55% to 63.83 million yuan. Xingsen Technology forecast first-half net profit of 100-120 million yuan, up 246.83%-316.19% year-over-year, according to Securities Times.
The Technology Behind the Surge
AI servers require dramatically more sophisticated circuit boards than traditional computing equipment. TrendForce data shows global AI server shipments are expected to grow nearly 31% in 2026, with the nine major cloud service providers’ capital expenditures reaching $886.7 billion—up roughly 90% year-over-year.
Traditional server PCBs typically have 8-12 layers and cost 800-2,000 yuan per unit. AI server PCBs, by contrast, range from 20 to 70+ layers (up to 100 for flagship products) and cost 8,000-12,000 yuan per unit—nearly ten times the value. This technical leap has created massive demand for high-frequency, high-speed, low-loss materials.
Wang Chen, director of the Next-Generation High-Frequency High-Speed CCL Innovation Center at the Greater Bay Area National Innovation Center, noted that AI computing power requires PTFE (polytetrafluoroethylene) copper-clad laminates. “High-end copper-clad laminates are almost being snatched up,” he said, as reported by China News/CCTV Finance.
Price Surges and Supply Constraints
High-speed PCB prices have risen more than fourfold, according to industry sources. Liu Qiming, head of a motherboard production enterprise, told CCTV Finance: “PCB price increases started with high-speed PCBs, which are mainly used in large computing power server motherboards. High-speed PCB prices have already risen far beyond three to four times.”
A Shenzhen PCB supply platform confirmed widespread material shortages, with multiple upstream suppliers issuing price increase notices. The supply crunch is expected to persist due to structural constraints: building a high-end PCB production line takes 18-24 months from planning to mass production, and client certification cycles run 12-36 months for global core customers.
The Next Wave: Computing Power Metals
As the AI infrastructure buildout accelerates, attention is turning to the upstream materials that power it. The China Nonferrous Metals Industry Association reported that non-ferrous metal prices surged in the first half of 2026, driven by AI infrastructure demand. Tin prices rose 40% in six months, indium climbed 60%, and tantalum soared 158%, according to China Economic Net.
Bian Jiang, vice president of the Precious Metals Working Committee of the China National Resources Recycling Association, highlighted the role of platinum group metals in AI infrastructure: “Platinum-rhodium alloys for high-end silicone and high-end electronic cloth, ruthenium for computing storage, and palladium for semiconductor materials.”
Copper demand is also surging. Global data center copper consumption is forecast at 740,000 tons in 2026, projected to reach 1.3 million tons by 2028. AI servers use three times more tin than traditional servers, making tin critical for advanced semiconductor packaging.
Expansion Wave and Structural Differentiation
The boom has triggered a historic expansion wave. As of June 2026, more than 20 A-share PCB listed companies have announced expansion plans totaling over 800 billion yuan in investment, according to Laoyaoba/Jiwei. Key projects include Shengnan Tech’s 200 billion yuan annual plan, Hudian’s 176 billion yuan investment, and Pengding Holdings’ 110 billion yuan Huai’an high-end PCB production base.
However, the industry is experiencing sharp K-shaped differentiation. While high-end AI-specific PCBs face supply shortages and soaring prices, low-end traditional PCB segments remain mired in overcapacity with utilization rates at just 60-70%. This is not a broad electronics recovery—it is a highly concentrated AI-driven boom.
Sina Finance reports that AI-related PCB revenue now accounts for over 70% of business at leading firms like Shengnan Tech, Hudian, and Shennan Circuits, with gross margins stabilizing above 30%. The global PCB market reached $85.2 billion in 2025 (up 15.8%), with AI infrastructure PCB and IC substrate markets projected to grow at a 32.6% CAGR through 2030.
Risks on the Horizon
Despite the current euphoria, experts caution about significant downside risks. Gong Chao warned: “If large model deployment falls short of expectations in 2-3 years, combined with concentrated capacity release, related industries will face significant downside risk.”
Duan Shaofu, deputy secretary-general of the China Nonferrous Metals Industry Association, noted that while the short-term supply-demand pattern for “computing power metals” remains supportive, “macro-level factors create valuation pressure, geopolitical uncertainties exist, market sentiment is highly sensitive, and market volatility risks cannot be ignored.”
The massive expansion wave—with over 800 billion yuan in planned investment—could create structural overcapacity if AI demand growth slows. Global tech giants are increasingly borrowing to fund AI infrastructure, with OFWeek reporting that Meta’s free cash flow fell 91% year-over-year in Q2 2026 despite record revenue, as capital expenditures reached $31.1 billion.
What to Watch
The AI computing power boom has fundamentally transformed the PCB industry from a low-margin, commoditized sector into a strategic bottleneck for global AI infrastructure. With China accounting for 57.5% of global PCB production value, the country is well-positioned to benefit from this buildout.
Key indicators to monitor include: whether AI server shipment growth maintains its trajectory through 2027, how quickly new production capacity comes online, and whether “computing power metals” follow the same explosive trajectory as PCBs. The industry’s ability to manage the transition from shortage to potential oversupply will determine whether this boom becomes a sustainable growth cycle or another cyclical peak.
For now, the “explosive order” state appears set to continue, with supply constraints and multi-year production lead times keeping high-end PCB demand well ahead of supply. The AI infrastructure buildout has moved from concept to concrete reality—and the entire hardware supply chain is feeling the impact.