24-Hour AI News: Microsoft Secures $97 Billion for GB300; Mininglamp Technology's Listing Ignites the Agentic Sector

In the past 24 hours (November 2-3), the Chinese side saw a dual-engine drive from the capital market and policies: Minglue Technology listed on the Hong Kong Stock Exchange, and ministerial articles and water-saving equipment solutions directly targeted "AI + manufacturing/data center"; in North America, Microsoft signed a $97 billion supply agreement with IREN, wearable AI medical research was included in the conference agenda, and semiconductor testing capacity cooperation was upgraded.

I. Minglue Technology listed on the Hong Kong Stock Exchange, and is known as the "world's first agentic AI stock".

1. Listed on November 3, stock code 2718, issue price HK$141 per share, issue 7,219,000 Class A shares.
2. The Hong Kong public offering was oversubscribed by 4,452.86 times, and the international offering was oversubscribed by 13.93 times; the stock price rose by more than 100% at one point during the day, and the market value exceeded HK$40 billion.
3. Positioning itself as a complete technology stack provider of "Agentic AI", the funds will be used for product and R&D expansion, marking a new stage of capitalization for the Agentic AI sector.

II. Microsoft and IREN sign a five-year, $9.7 billion cooperation agreement: gain access to AI chip capabilities such as NVIDIA GB300.

1. The agreement includes a 20% upfront payment, under which IREN will purchase approximately $5.8 billion worth of equipment through Dell for deployment at its 750MW Childress campus in Texas.
2. The newly built liquid-cooled data center is expected to provide approximately 200MW of critical IT capacity, alleviating Microsoft's AI computing power shortage until mid-2026.
3. Stimulated by the news, IREN's stock price rose more than 20% in pre-market trading, demonstrating the strong cyclical nature of "computing power as capital expenditure".

III. Minister of Industry and Information Technology Li Lecheng issued an article: Systematically promote "AI + Manufacturing" and accelerate the research and development of AI terminals and humanoid robots.

1. An article published on November 3rd proposed strengthening computing power, algorithms, and data infrastructure; and accelerating breakthroughs in key technologies such as high-end chips and high-speed interconnects.
2. Clearly define the need to accelerate the replacement of consumer terminals such as AI mobile phones and AI computers, and promote the research and development of next-generation intelligent terminals such as humanoid robots and brain-computer interfaces.
3. Deploy a public service platform for intelligent transformation and large-scale modeling in key manufacturing industries to strengthen "modular-digital resonance".

IV. Two departments released the "Implementation Plan for High-Quality Development of Water-Saving Equipment (2025-2030)": specifically mentioning water-saving technologies for data centers/intelligent computing centers.

1. The document issued on November 3rd emphasizes the water conservation needs of new infrastructure such as data centers, general computing centers, intelligent computing centers, and supercomputing centers.
2. Encourage the use of efficient heat dissipation equipment such as liquid cooling, evaporative cooling, heat pipes, and refrigerant pumps to improve the utilization rate of natural cold sources.
3. By 2027, a number of leading enterprises and single-item champions will be cultivated, which will help alleviate the pressure on water and electricity supply and operation and maintenance costs brought about by the expansion of AI computing power.

V. AHA Releases: Hand-Swept One-Lead ECG + AI Can Screen for Multiple Types of Structural Heart Disease (Research Abstract)

1. According to information released at the meeting on November 3, among 600 subjects, the AI algorithm based on a smartwatch achieved a detection performance of approximately 88% for a 30-second single-lead ECG.
2. The training data came from 266,054 12-lead ECGs, and the external validation covered 4 hospitals and a Brazilian population cohort.
3. Provides evidence for the application of "wearable + AI" in early screening, but it is a conference abstract and still awaits peer review.

VI. Aehr Test Systems partners with ISE Labs to provide AI/HPC wafer-level aging and testing services.

1. On November 3, a strategic partnership was announced, focusing on advanced packaging such as CoWoS and "Known Good Product" (KGD) delivery.
2. The goal is to accelerate the ramp-up of AI/HPC processor mass production yield and control costs.
3. Provide testing support for the ramp-up of downstream AI accelerator/GPU production capacity.

VII. OpenAI to move into Michigan's Saline Township hyperdata center (Stargate project progress)

1. According to local public radio reports, the project will create approximately 2,500 construction jobs and 450 park jobs.
2. Adopt a closed-loop cooling system to reduce daily water consumption and address community concerns about resource use.
3. As a supplement to the ">1GW-level" park announced in late October, it highlights the concentration of computing power in the Midwestern United States.

8. Baidu released its "Top Ten Cutting-Edge Technological Inventions of 2025" and announced that Baidu World 2025 will be held on November 13.

1. Cutting-edge inventions cover large models, deep learning frameworks, AI computing power, intelligent agents, AI search, digital humans, autonomous driving, etc.
2. Baidu World 2025, themed "Emerging Effects," will unveil the latest advancements in AI applications, smart cloud, and autonomous driving.
3. This shows that leading manufacturers are advancing simultaneously across multiple levels: "infrastructure, platform, and application".

9. Ningbo "Interstellar Power" Established and Unveils Humanoid Robot "Interstellar No. 1"

1. Established jointly by Hongrun Construction and Matrix Intelligence, focusing on the industrialization of humanoid robots.
2. "Interstellar-1" publicly demonstrated its dynamic control and interaction capabilities, and the team's background covers Tesla, NVIDIA, OpenAI, Huawei, and others.
3. Local industrial parks undertake the "AI + Robotics" manufacturing chain, driving regional industrial synergy.

10. (Trend Observation) Leading model companies and cloud vendors join forces to strengthen AI security and compliance.

1. Overseas media focus today: Google, OpenAI, Anthropic and others are accelerating the patching of security vulnerabilities and indirect suggestion injection paths that are being exploited by chatbots.
2. With the accelerated adoption of enterprise-level security, security baselines, DLP, and jailbreak protection have become hard requirements for "compliance upon deployment".
3. It is expected to give rise to an integrated implementation solution of "model security base + enterprise strategy".

Frequently Asked Questions (Q&A)

Q: What does Microsoft's $9.7 billion deal with IREN mean for the AI industry?

A: In the short term, approximately 200MW of critical IT capacity and GB300 chip channels can be provided by mid-2026 to alleviate the bottleneck of Microsoft's limited AI capabilities; it also verifies that "outsourced computing power + liquid-cooled new factory" is the mainstream path for capacity expansion.

Q: Why is Minglue Technology called the "first stock of Agentic AI"?

A: Its disclosed autonomous intelligent agent technology stack and ToB application positioning, coupled with oversubscription and strong performance on the first day of listing, show that the capital market recognizes the combined logic of "Agent + industry knowledge graph + multimodal data".

Q: What are the key points of China's policy today?

A: Emphasizing the "AI + Manufacturing" main battlefield positioning, promoting the upgrading of AI terminals, the research and development of humanoid robots and brain-computer interfaces, and consolidating the computing power/algorithm/data foundation; at the same time, providing energy consumption and water consumption support for the expansion of data centers/intelligent computing centers through water-saving solutions.

Q: What are the clinical value and limitations of wearable AI ECG screening?

A: The value lies in using "single-lead ECG + AI" to expand early screening coverage at a lower cost; the limitation is that it is currently a conference abstract, the positive rate in the samples is low, and further peer review and real-world validation are needed.

Q: Why is the latest development of OpenAI's Michigan data center attracting so much attention?

A: By clearly defining "tenant + job scale + water-saving technology", we provide a concrete implementation node for the previous >1GW computing power park plan, and alleviate the computing power and power capacity gap in the Midwestern United States.