24-Hour AI News: OpenAI Reorganizes and Adds 30GW of Computing Power; UK Launches AI Skills Toolkit

In the past 24 hours (October 29), North America and Europe have made significant moves regarding AI infrastructure and governance: OpenAI announced a restructuring and proposed 30GW of computing power and new financing pathways; Amazon launched the "Rainier" hyperscale cluster for Anthropic; the EU opened its platform data for research; and the UK released an AI skills toolkit. In China, regulators emphasized using AI to drive the digital transformation of the capital market; Beijing launched a "challenge-based" intelligent manufacturing initiative; and the establishment of intelligent computing centers and their integration with industry projects, along with standardization efforts, are progressing simultaneously.

I. OpenAI's Reorganization and Computing Power Expansion Path is Clear

1. On October 29, OpenAI completed its governance and financing structure adjustment, lifted key fundraising restrictions, proposed to achieve 30GW-level computing power, and explored the IPO path.
2. The company's annual revenue operating rate is said to be close to $20 billion, and management said it needs to move towards "annual revenue of hundreds of billions of dollars" to support its infrastructure ambitions.
3. It also proposes a long-term capital expenditure plan of approximately US$1.4 trillion and an expansion target of "1GW per week", emphasizing collaboration with ecosystem partners.

II. Amazon's "Rainier" AI cluster deployed; Anthropic plans to use one million Trainium 2 processors this year.

1. On October 29, Amazon announced that its "Project Rainier" spanning multiple data centers was officially launched, and Anthropic will expand to more than 1 million Trainium 2 chips this year.
2. This cluster is used for training and deployment of current and subsequent versions of the Claude model, forming a complementary and distributed supply of NVIDIA/GPUs.
3. Cloud vendors are accelerating the supply of MaaS by using self-developed chips and dedicated clusters, releasing a signal of linkage between scale, cost and energy efficiency.

III. IBM Releases AI Model for Defense Scenarios, Emphasizing "Air-isolated/Locally Deployable"

1. On October 29, IBM announced that a dedicated model for defense and national security is now officially available, supporting deployment in isolated, edge, and classified environments.
2. The model is based on the Granite family and wassonx.ai, emphasizing the alignment of domain data with tasks to reduce illusions and improve usability.
3. The industry model approach of "small but refined" is similar to the logic of "general large model + security domain adaptation" in government and military industries.

IV. With the EU DSA data access package now in effect, researchers can apply for access to VLOPs platform data.

1. Starting October 29, the “access to research data” authorized by the EU Digital Services Act (DSA) came into effect, allowing qualified researchers to apply for data from large platforms/search engines in accordance with the law.
2. The research focuses on "systemic risks" such as recommendation systems, and incorporates governance issues such as the protection of minors, illegal content and fraud.
3. Platforms must cooperate in accordance with the law, and digital service coordinators of member states shall conduct unified reviews to strengthen the "transparency-accountability" framework.

V. The UK releases an AI skills framework and toolkit, targeting £400 billion in growth potential by 2030.

1. On October 29, Skills England released a three-piece set of documents: the AI Skills Framework, the Adoption Path Model, and the Employer List.
2. It also stated that through cooperation with NVIDIA, Google, IBM, Microsoft and other companies, it will provide AI skills training to 7.5 million workers by 2030.
3. The document is positioned as a "starter tool" for employers, focusing on industry pain points and feasible implementation paths for SMEs.

VI. China Securities Regulatory Commission: Promoting high-quality digital transformation of the capital market with new technologies such as artificial intelligence

1. On October 29, the Vice Chairman of the China Securities Regulatory Commission (CSRC) stated at the Financial Street Forum that the CSRC will adhere to the principle of "actively promoting + preventing and controlling risks" and promote the deep integration of AI in investment research, risk control, operations and other businesses.
2. Propose to carry out special actions focusing on "artificial intelligence +" and "data elements ×" to improve the level of digitalization and intelligence in supervision.
3. Clearly define information security, data security, and business risk prevention and control requirements, and emphasize compliance boundaries.

VII. Beijing launches a "challenge-based" competition for intelligent manufacturing system solutions by 2025.

1. On October 29, a notice was issued focusing on bottlenecks in key industries such as raw materials, high-end equipment, and electronic information, and strengthening the "deep integration of artificial intelligence technology and scenarios".
2. Clarify the unit's qualifications, task focus, and acceptance mechanism. Online applications will be accepted until November 10. Priority will be given to projects that are "independent and controllable and can be replicated and promoted".
3. Propose to carry out integrated research and application verification around typical smart factory scenarios, and promote the integration of "equipment-software-system" into a chain.

8. Accelerated Construction of Intelligent Computing Centers: AI IaaS Up 122.4% Year-on-Year in the First Half of the Year

1. On October 29, industry reports cited IDC data stating that China's AI IaaS market will grow by 122.4% year-on-year in the first half of 2025.
2. High power density, ultra-fast delivery, green and low-carbon, and safe and reliable have become the "four key words" of intelligent computing centers.
3. The integration of data collection in the east and computing in the west with localized applications is driving computing power supply from "resource supply" to "value creation".

9. 360 wins bids for projects including the Wuhan AI Demonstration Base, securing orders exceeding 300 million yuan in a single month.

1. On October 28, 360 won the bid for the Wuhan Artificial Intelligence Innovation Application Demonstration Base (Phase I) for 132 million yuan; previously, it won the bid for the Hohhot project with its partners for approximately 160 million yuan.
2. The company's AI-related projects totaled over 300 million yuan in October, covering sectors such as "AI Security Cloud, AI Application Factory, and AI Academy".
3. Supporting computing power/platforms and industry applications in tandem, strengthening the in-depth layout of "AI + security".

10. Shanghai has led the development of 251 international standards, with artificial intelligence listed as a key area.

1. On October 29, it was reported that Shanghai has formed a standard-industry collaborative development pattern in three leading industries: integrated circuits, artificial intelligence, and biomedicine.
2. In the past three years, we have led/participated in the formulation of 126 national standards, cultivated 64 "Shanghai standards", and built a standardization collaboration platform for foreign-invested enterprises.
3. Encourage active leadership in the formulation of international standards in cutting-edge fields such as quantum mechanics, AI, and space information, and enhance global influence.

11. Anthropic releases new features for the financial industry, enhancing "real-time data + agent skills".

1. On October 27, Anthropic announced that "Claude for Financial Services" has added Excel plugins, LSEG and other real-time data connectors.
2. Launch pre-built Agent skills such as DCF modeling, financial statement key point extraction, and due diligence package, and open enterprise/team previews first.
3. Reflects the industry trend of "engineering financial workflows + multi-source data access".

Frequently Asked Questions (Q&A)

Q: What is the most critical change in OpenAI's restructuring?

A: With the removal of core fundraising restrictions and the public outlining of a 30GW computing power and approximately $1.4 trillion in long-term capital expenditure plans, management has stated that they need to achieve revenue of "hundreds of billions of dollars per year," and an IPO is seen as a "more likely path."

Q: What is the industry significance of Amazon deploying "Rainier" for Anthropic?

A: Within the year, we will expand to one million Trainium2 processors to form a massive dedicated computing power, enhance the elasticity of Claude training/inference, further reduce the single reliance on GPU supply, and promote the "MaaS + self-developed chip" paradigm.

Q: What are the key signals from the Chinese side today?

A: Regulators have clearly stated that AI will drive the digital transformation of the capital market; Beijing has launched a "challenge-based" approach to intelligent manufacturing; industry data shows that AI IaaS increased by 122.4% year-on-year in the first half of the year; enterprise-side implementation projects and local standardization construction are progressing in tandem.

Q: What will happen when the EU's DSA "access to research data" comes into effect?

A: Qualified researchers can apply for VLOPs data in accordance with the law to assess systemic risks (such as recommendation systems, protection of minors, illegal content and fraud), and platforms need to cooperate by strengthening transparency and accountability.

Q: What is the direct value of the AI skills toolkit launched in the UK to businesses?

A: We provide a three-piece toolkit of “framework, adoption path, and employer list” to address industry pain points and implementation processes for SMEs, aiming to support the £400 billion growth potential and the training program for 7.5 million people by 2030.