AI技术行业报告：从AI基础设施扩张到收益兑现与估值分化

AI Technology Industry Report: From AI InfrastructureExpansion to Earnings Delivery and Valuation Divergence Bin Yang, SFC:BSQ313 harryyang@hfisec.com.hk July 09 2026 (This note is the English translation of a note that was originally written in Chinese and published on 20260708. The original version should prevail.) 本研究报告由华福国际分销，在其许可的司法管辖区内从事证券活动。关于华福国际的分析师证明，重要披露声明和免责声明，请参阅附录。Thisresearch report is distributed by HuaFu International, which is authorized to engage in securities activities in its respective jurisdiction. Please refer to appendix fordetailed analysts certification, conflict of interest disclosures and important disclaimer. Abstract •Core Assessment:The AI tech narrative isn't over, but it has shifted from "theme trading" into a more rigorous validation phase. Themarket no longer just questions whether AI demand exists; it now demands tangible proof that companies can consistently deliveron revenue, orders, gross margins, cash flow, and CapEx ROI. •Demand Side:Cloud providers remain the primary drivers of AI infrastructure demand. Meanwhile, as Agentic AI evolves fromsingle-turn Q&A to executing long-horizon tasks, it will further drive up consumption of computing power and cloud resources. •Supply Side:AI infrastructure has expanded beyond standalone GPUs into comprehensive systems engineering. While GPUs remainthe core, delivery bottlenecks have now spread to ASICs, HBM, CoWoS, advanced node manufacturing, semiconductor equipment,power supply, liquid cooling, and data center capacity. The IEA projects that global data center power consumption will nearlydouble to around 945 TWh by 2030, highlighting tightening constraints on power and infrastructure. •Financials:Multiple segments in the hardware chain have already validated this demand. Nvidia’s Data Center revenue for FY2027Q1 surged 92% YoY, Broadcom’s AI semiconductor revenue in FY2026 Q2 jumped 143% YoY, and Applied Materials expects itssemiconductor equipment business to grow by over 30% in 2026. This confirms that AI demand is successfully cascading throughthe entire chain—from chips and memory to packaging and equipment. For full disclosure of risks, valuation methodologies and target price formation on all HFI rated stocks, please refer to the latest full report on our website atwww.hfisec.com.hk Table of Contents 1.Tech Sector Core Takeaways Update: From AI Infrastructure Expansion to Earnings Realizationand Valuation Divergence 2.Key Company Matrix: AI opportunities are unfolding in parallel across chips, memory, foundry,equipment, and the cloud. 3.Investment Recommendations: Hardware 60%, Cloud Vendors 20%, Temperature Control andPower Supporting Facilities 20% 4.Appendix Updated core conclusions of the technology industry: From the expansion of AIinfrastructure to performance realization and valuation divergence •This report provides updates on AI computing power, semiconductors, capital expenditures of cloud vendors,storage,advanced packaging,semiconductor equipment,power infrastructure and developments of keyenterprises. •The core view of this period is that the AI-driven trend is far from over, while the market is shifting from thematicspeculation to performance realization, supply verification and valuation adjustment. •Key entities to follow in the future: Nvidia, AMD, Broadcom, Micron, TSMC, ASML, Applied Materials, Microsoft,Amazon, Google, Meta, Oracle Market Performance:The AI and semiconductor sectors have seen increased volatility after hitting high levels, indicatingthat capital has raised its requirements for valuations and the ROI of AI capital expenditures. Industrial Chain: The demand for AI computing power has spread from GPUs to multiple links including HBM, advancedpackaging, semiconductor equipment, power supply and heat dissipation. Company Progress:Under the same core AI theme, companies benefit in diverse ways, and it is necessary to distinguishrevenue realization, order flexibility, supply bottlenecks and valuation pressure. For full disclosure of risks, valuation methodologies and target price formation on all HFI rated stocks, please refer to the latest full report on our website atwww.hfisec.com.hk Multiple financial gains have been realized across the AI hardware industry chain,yet the market is now demanding higher-quality performance verification. Latest Changes: Demand for AI infrastructure remains robust, yet the market has shifted from the phase of "AI theme proliferation" to "earningsrealization and verification". The recent financial performances of Nvidia, AMD, Broadcom, Micron and Applied Materials have validated the demandresilience across the AI hardware industry chain from five key segments namely GPUs, data center CPUs/GPUs, customized AI silicon, HBM/storage andsemiconductor equipment respectively. Impact Analysis: The market is not denying AI demand at present; instead, it calls for more rigorous verification, including sustained revenue growth,stable gross pr