据港交所1月30日披露，安徽海马云科技股份有限公司(简称：海马云)向港交所主板递交上市申请，中金公司（601995）、招银国际为联席保荐人。该公司 曾于2025年6月27日递表港交所。 公司简介 招股书显示，海马云是中国最大的云游戏实时云渲染服务(即视觉内容计算及生成)GPUaaS提供商。根据灼识咨询的资料，按2024年的收入计算，公司在中 国云游戏场景实时云渲染服务提供商中排名第一，市场份额为17.9%。 GPUaaS是一种云解决方案，使用户可以通过虚拟化技术远程、经济高效地访问以GPU为基础的计算资源。其主要包括(i)图形GPUaaS，主要是实时云渲染； 及(ii)计算GPUaaS，主要是AI推理和AI训练。公司为游戏、泛娱乐、数字文旅(即利用数字技术虚拟呈现文化遗址或景点，并提供互动体验)、数字教育等行 业的客户提供实时云渲染服务和相关解决方案。 该等服务及解决方案由公司自研及深度一体化的平台软件和基础设施硬件支持，配备以GPU图形计算为核心的深度一体化软硬件。该架构使终端用户可以随 时随地在不同设备上使用基于高性能GPU计算的数字内容及应用，而不受用户设备的限制。 于2022年度、2023年度、202 ...

Strategic Combination - Rumble and Northern Data are combining to build a "Freedom-First Ecosystem" rooted in freedom, privacy, independence, and resilience, aiming to compete with Big Tech[2] - The combination seeks to provide independent access to technology and financial tools, decentralizing these resources to enhance societal resilience[3] - The combined entity aims to create a full-stack cloud platform, from power to GPUaaS[22] Key Assets and Capabilities - Rumble has 47 million Monthly Active Users and a growing cloud business, securing a $775 million strategic investment with Tether[10] - Northern Data possesses over 22,000 GPUs and operates 9 data centers, providing a sovereign AI Cloud platform[13] - The combined entity will have over $1 billion in capital deployed[16] Strategic Rationale - The acquisition of Northern Data will provide Rumble with immediate scale in the enterprise market through GPUaaS and data center businesses[20] - The combination expands the international footprint of both companies, creating globally accessible computing infrastructure[23] - Tether has committed to an initial purchase of up to $150 million of GPU services over a 2-year period from the combined entity[25] Financial Terms and Ownership - The transaction is an all-stock deal where Northern Data shareholders will receive 2.0281 newly issued Rumble Class A shares[31] - Post-transaction, Rumble will own 69.6% and Northern Data will own 30.4% of the combined company[32] Market Positioning - The combined company aims to become a growing player in the high-growth AI and GPU market, with a market capitalization of $3.5 billion[30]

3Q25 Earnings Release Disclaimer This presentation has been prepared by KT Corp.(the "Company") in accordance with K-IFRS. This presentation contains forward-looking statements, which are subject to risks, uncertainties, and assumptions. This presentation is being presented solely for your information and is subject to change without notice. No presentation or warranty, expressed or implied, is made and no reliance should be placed on the accuracy, actuality, fairness, or completeness of the information pre ...

Core View - CoreWeave, Inc. (NASDAQ:CRWV) is highlighted as a top AI stock following an expanded agreement with OpenAI, with a price target of $180.00 set by Citizens JMP analyst Greg Miller [1][4]. Company Developments - CoreWeave has expanded its agreement with OpenAI, increasing the total contract value to approximately $22.5 billion, which includes a recent addition of up to $6.5 billion [2][4]. - The initial contract was valued at up to $11.9 billion, signed in March, followed by an expansion of up to $4 billion in May 2025 [4]. Market Outlook - Analysts anticipate new contracts as hyperscalers are increasingly outsourcing GPU cluster buildouts to GPU-as-a-Service (GPUaaS) providers, which assume financial and technical risks [2][4]. - The GPUaaS market is projected to grow from approximately $3-4 billion currently to around $300 billion, indicating a significant acceleration in demand and infrastructure outsourcing by hyperscalers [3][4].

Summary of Key Points from the Conference Call Industry Overview - **Data Center Market in Asia-Pacific**: The data center market in Thailand is emerging as a new destination for regional deployment, with a pipeline capacity of 2.6GW in Bangkok, which increased by over 1.7GW in 2024 [1][12]. Johor, Malaysia, has a pipeline market of 5.7GW but has seen a slowdown in incremental supply growth due to stricter local regulations [1]. Core Insights - **AI Demand Growth**: AI-related demand is expected to contribute approximately 25-30% of incremental data center demand in markets like Thailand, although current AI capacity in APAC (excluding China) is only about 1GW [2]. Customers remain cost-sensitive regarding AI workloads, leading to deployments in markets like Malaysia and Australia [2]. - **Edge Data Centers**: There is a strong demand for smaller edge data centers at metro edges, which are essential for AI inference due to low latency requirements. Telecom operators are well-positioned to capitalize on this trend [3]. - **GPU as a Service (GPUaaS)**: The GPUaaS market is projected to be significant, with estimates ranging from $35 billion to $70 billion globally by 2030. Companies like Singtel are well-positioned to benefit from this growth [4]. Pricing and Cost Analysis - **Data Center Pricing Variability**: Pricing for data centers varies significantly across the region, with Singapore being the most expensive at up to $300/KW/month, while Malaysia offers lower rates around $90-95/KW/month due to lower land costs [27][30]. - **Cost Structure**: In Japan, MEP (Mechanical, Electrical, and Plumbing) costs constitute 59% of total data center build costs, highlighting the importance of these components in overall expenditure [25]. Market Dynamics - **Regulatory Environment**: In Johor, the local government is becoming more selective in approving new data center projects, focusing on water and power usage efficiency [46]. There are currently 13 live projects with a total approved capacity of 1774MW, but only about 500MW is operational [46]. - **Investment Trends**: The Johor Special Economic Zone (SEZ) agreement between Malaysia and Singapore is expected to enhance cross-border connectivity and attract global investments, particularly in data centers [46]. Emerging Technologies - **Liquid Cooling Solutions**: Liquid cooling is becoming increasingly relevant for data centers hosting AI workloads, as traditional air-cooling systems fail to support higher rack densities required for AI [30][31]. Various liquid cooling methods are being explored, including direct-to-chip and immersion cooling [31]. - **AI Workloads and Infrastructure**: AI workloads require higher densities, with rack densities exceeding 100KW/rack, necessitating advanced cooling solutions [38]. The demand for AI infrastructure is expected to grow rapidly, with significant investments needed to meet this demand [24]. Conclusion - The data center industry in the Asia-Pacific region is experiencing rapid growth driven by AI demand, regulatory changes, and technological advancements. Companies like Singtel and Indosat are positioned to benefit from these trends, particularly in GPUaaS and liquid cooling technologies. The evolving landscape presents both opportunities and challenges, particularly in terms of regulatory compliance and infrastructure development.