CITIC SEC released a research report stating that NVIDIA's FY26Q3 earnings and FY26Q4 guidance exceeded expectations, while Google unveiled Gemini 3, showcasing significantly improved model performance beyond projections. This reflects the continued acceleration of the AI industry, reinforcing the investment appeal of PCBs as a critical upgrade component for AI chips. The PCB sector is also witnessing multiple catalysts, prompting CITIC SEC to reiterate its recommendation for key players in the AI PCB segment.

**Catalyst 1: NVIDIA's Earnings and Guidance Surpass Expectations** On November 20, NVIDIA reported its FY26Q3 results for the period ending October 26, 2025, posting revenue of $57 billion, up 62% YoY and above the consensus estimate of $55.2 billion. Adjusted operating profit reached $37.8 billion, also up 62% YoY and exceeding the $36.5 billion forecast. For FY26Q4, NVIDIA projected revenue of $65 billion, a 65% YoY and 14% sequential increase, surpassing market expectations ($62 billion). The company also guided Q4 gross margins to 74.8% (±50 bps), signaling an upward trend. The stronger Q4 growth outlook compared to Q3 actuals underscores the overall outperformance.

**Catalyst 2: Google's Gemini 3 Outperforms, Validating Scaling Laws** On November 18, Google launched its next-gen AI model, Gemini 3.0, which demonstrated industry-leading performance in multiple benchmarks. For instance, Gemini 3 Pro scored 37.5% in Humanity's Last Exam (HLE), significantly ahead of GPT-5.1's 26.5%. In the ScreeSpot-Pro test, it achieved 72.7%, outperforming Claude Sounnet 4.5 (36.2%) and GPT-5.1 (3.5%). The model exhibited marked improvements in vision, reasoning, programming, and multimodal capabilities, enhancing real-world usability. Gemini 3 is already available on AI Studio, Gemini CLI, and major developer platforms like Cursor, GitHub, JetBrains, and Cline.

**Sector Outlook: Sustained Compute Demand Bolsters AI PCB Growth** Token demand growth continues to outpace expectations, with Google reporting monthly inference token usage of 1.3 quintillion in October—more than double May’s figure. Gemini 3.0 was trained on Google’s proprietary TPUs, whose cost efficiency implies higher compute output per capex, driving demand for supporting components. CITIC SEC remains bullish on the AI PCB sector’s growth potential, noting ongoing iterations/validation of orthogonal backplane solutions, with further feedback expected in 1–2 months. Additionally, NVIDIA’s Rubin chip has entered tape-out, and its PCB specifications (materials, design, etc.) are expected to clarify soon, providing further catalysts for the sector.

**Investment Strategy** CITIC SEC highlights PCBs as the most synchronized upgrade segment for AI chips, benefiting from structural demand for high-end HDI and multilayer PCBs driven by AI chips and high-speed networking. The sector’s competitive landscape and supply-demand dynamics are likely to remain favorable near-term, with leading players poised for sustained high earnings growth. Valuation upside persists, and the firm recommends focusing on: 1) Leading manufacturers with advanced technology, strong client positioning, and high earnings visibility. 2) Companies demonstrating notable margin improvements, capacity expansion, and reinforced AI business prospects. 3) CCL (copper-clad laminate) leaders poised to benefit from rising CCL prices and high-end material adoption.