According to market intelligence, major global technology giants have recently released intensive signals of accelerating AI infrastructure development. On September 10 Eastern Time, Oracle (ORCL.US) announced its Q1 FY2026 financial results. While revenue and earnings per share fell slightly short of market expectations, the company delivered a market-shaking bombshell: Remaining Performance Obligations (RPO) reached $455 billion, surging 359% year-over-year. This stunning figure completely ignited market enthusiasm, with Oracle's stock price soaring nearly 36%, marking its largest single-day gain since 1992, and market capitalization reaching $933 billion at one point.
Similar to Microsoft and other major cloud service providers, Oracle has become one of the primary beneficiaries of the artificial intelligence boom through its cloud infrastructure business and access to NVIDIA's graphics processing units (GPUs) for large-scale workloads. Oracle management emphasized that the current AI inference market significantly exceeds the AI training market in scale, while inference computing power in the market is being depleted. Leveraging its leadership position in enterprise databases, the company enables enterprise clients to securely combine private and public data while maintaining data privacy, utilizing large language models for inference.
Oracle provided even more optimistic performance guidance, projecting that the company's cloud infrastructure revenue for FY2026 will grow 77% year-over-year to $18 billion. Over the subsequent four years, the company's cloud infrastructure revenue is expected to grow to $32 billion, $73 billion, $114 billion, and $144 billion respectively. "Most of the revenue in our projections has already been included in RPO, and Oracle is entering FY2026 in brilliant fashion," stated Oracle CEO Safra Catz.
Industry observers are optimistic about the sustainability of domestic companies directly supporting overseas supply chains. Given China's most comprehensive industrial production system globally, advantages in AI implementation are becoming prominent, and domestic AI industry chain segments may be undervalued.
Notably, AI giant NVIDIA also made new moves. NVIDIA announced on Tuesday that the company plans to launch a new product designed to handle complex tasks such as video generation and software development, while its chips and systems are currently at the center of the artificial intelligence computing boom. NVIDIA stated that this product, named Rubin CPX, will be available by the end of 2026. It will take the form of a card that can be embedded in existing server computer designs or used in standalone computers capable of running parallel with other hardware in data centers. This design is a derivative of the new Rubin product line launching next year, enabling certain types of AI work to be more efficient. NVIDIA indicated that Rubin racks equipped with Rubin CPX can deliver up to 6.5 times better performance than the current flagship rack GB300 NVL72 when processing large context windows. NVIDIA CEO Jensen Huang stated that CPX is the first chip built specifically for models that need to process large amounts of knowledge (millions of tokens) at once and perform AI inference.
Additionally, Microsoft reached a $19.4 billion multi-year AI cloud computing partnership with Nebius (including a $2 billion additional option, valid until 2031); Meta founder Zuckerberg announced plans to invest approximately $600 billion in AI infrastructure by 2028. Leading manufacturers are intensively revising capital expenditures upward, driving the AI computing power industry chain from "expectation fulfillment" into the "accelerated performance release" phase.
Recently, supportive policies in the AI sector have been intensively introduced. On September 9, Vice Minister of Industry and Information Technology Zhang Yunming stated at a State Council Information Office press conference that the Ministry of Industry and Information Technology will promote high-quality development of the artificial intelligence industry, accelerate high-level empowerment of new industrialization, research and introduce special action implementation plans for AI + manufacturing, deploy intelligent transformation tasks for key industries, key links, and key areas, formulate AI + manufacturing transformation roadmaps, and release and implement artificial intelligence application guidelines for manufacturing enterprises.
On the same day, Hangzhou Economic and Information Technology Bureau publicly solicited opinions on the "Hangzhou Three-Year Action Plan for Accelerating the Development of Artificial Intelligence Terminal Industry (2025-2027) (Draft for Comments)." It proposed that by 2027, Hangzhou's artificial intelligence terminal industry scale aims to reach 300 billion yuan. Based on DeepSeek and "Tongyi Qianwen" series open-source models, the plan involves developing high-performance large model integrated machines such as inference-integrated machines and training-inference integrated machines, as well as edge computing servers and AI servers equipped with domestic artificial intelligence chips.
Guojin Securities indicated that according to IDC calculations, China's AI market is expected to reach 815.9 billion yuan by 2028, with a five-year CAGR of approximately 33%. The industry chain shows a gradient of "computing power first - model follow-up - application explosion": upstream GPU/memory/network infrastructure expansion, midstream model and MaaS acceleration of commercialization, downstream industry applications blooming across multiple sectors, with application-side focus on enterprise-level AI Agent implementation. Comprehensively, technological breakthroughs, domestic and overseas demand, and capital expenditure resonance establish an upward cycle, with artificial intelligence layout around three main lines: upstream computing power supply, MaaS platformization, and industry-level Agent solutions.
Related Concept Stocks:
Asia-Pacific Select ETF (159687): The Southern Fund Asia-Pacific Select ETF is primarily composed of two types of assets: "Asia-Pacific quality dividend assets and Asia-Pacific semiconductor leading companies." This ETF includes quality dividend assets from the Asia-Pacific region (represented by Japanese and Chinese markets), such as Toyota Motor, Tencent Holdings, Alibaba, and Mitsubishi Group. These companies have long-term stable cash flows and stable profitability, providing relatively steady return contributions to the portfolio. The Asia-Pacific Select ETF also includes quality semiconductor companies such as TSMC, Samsung, Tokyo Electron, and MediaTek, with TSMC being the largest component stock by index weight at over 8%. Asia-Pacific is the core region of the global semiconductor industry chain. Most intuitively, according to the 2024 Deloitte & GSA report, Asia-Pacific semiconductor industry revenue accounted for 57.6% of global semiconductor industry revenue in 2022.
Baidu Group-SW (09888): On September 9, according to Baidu's official WeChat account, at the WAVE SUMMIT Deep Learning Developer Conference 2025, ERNIE Bot Large Model X1.1 was officially released. According to the introduction, users can currently use ERNIE Bot Large Model X1.1 on the ERNIE Bot official website and Wen Xiaoyan APP. ERNIE Bot Large Model X1.1 has also officially launched on Baidu Intelligent Cloud Qianfan Platform, fully open to enterprise customers and developers. Compared to ERNIE Bot Large Model X1, X1.1 improved factuality by 34.8%, instruction following by 12.5%, and intelligent agents by 9.6%. Additionally, in multiple authoritative benchmark evaluations, ERNIE Bot Large Model X1.1's overall performance exceeded DeepSeek R1-0528, showing leading advantages in some tasks. Simultaneously, compared with international top models GPT-5 and Gemini 2.5 Pro, the effects are comparable.
Alibaba-SW (09988): In early September, Guosen Securities issued a research report maintaining Alibaba-W (09988) "outperform" rating. The company currently has extensive AI model and application layouts, with the Qwen series models leading domestic models, and actively promoting AI transformation of core products including Taobao, Quark, DingTalk, and AutoNavi on the application side, achieving comprehensive layout for B2B and B2C scenarios.
SenseTime-W (00020): At the end of August, Goldman Sachs issued a research report stating that the State Council issued "Opinions on Deeply Implementing 'Artificial Intelligence +' Actions," emphasizing AI commercialization and ecosystem expansion. The policy implementation is expected to promote user consumption in AI, driving SenseTime-W (00020) generative AI business development. Combined with market reassessment of China's AI supply chain, Goldman Sachs raised SenseTime's target price from HK$1.83 to HK$2.72, upgrading the rating from "neutral" to "buy." The report noted that SenseTime's first-half revenue performance exceeded expectations, growing 36% year-over-year. Management indicated strong growth in generative AI business, up 73% year-over-year, mainly driven by productivity tools and interactive tools. Under favorable policy background and SenseTime's comprehensive product portfolio support, the firm holds a more optimistic attitude toward SenseTime's generative AI business, expecting revenue contribution to increase from 64% in 2024 to 91% in 2030.
Kingboard Laminates Holdings (01888): Kaiyuan Securities pointed out that the company's copper clad laminate prices have already led price increases in the first half, and strong PCB demand may support price rigidity, with second-half performance expected to benefit; high-end CCL and material capacity layout acceleration in 2026, AI options expected to enter realization period, driving valuation center upward, with actual financial statement contributions likely to be concentrated in 2027, maintaining "buy" rating.
Kingboard Holdings (00148): Layout covers upstream raw materials and downstream PCB full industry chain. Kingboard Holdings announced 2025 interim results with revenue of approximately HK$21.608 billion, up 6% year-over-year; net profit attributable to company shareholders of approximately HK$2.582 billion, up 71% year-over-year; basic earnings per share of HK$2.329, interim dividend per share of HK$0.69. Kingboard Laminates Holdings also released 2025 interim results, achieving revenue of HK$9.588 billion in the first half, up 11% year-over-year; net profit attributable to company shareholders of HK$933 million, up 28% year-over-year; basic earnings per share of 29.9 HK cents, proposing interim dividend of 15 HK cents per share.