October 7, 2025 12:33 AM GMT Investor Presentation | Asia Pacific M Foundation Greater China Technology Hardware: AI Tech Hardware Upgrades Across the Board Morgan Stanley Taiwan Limited+ Sharon Shih Equity Analyst Sharon.Shih@morganstanley.com +886 2 2730-2865 Derrick Yang Equity Analyst Derrick.Yang@morganstanley.com +886 2 2730-2862 Howard Kao Equity Analyst Howard.Kao@morganstanley.com +886 2 2730-2989 Greater China Technology Hardware Asia Pacific Industry View In-Line Morgan Stanley does and seeks to ...

Core Viewpoint - NVIDIA's investment of $5 billion in Intel is seen as a strategic partnership that could reshape the competitive landscape in the semiconductor industry, particularly between NVIDIA and AMD [1][3][16]. Group 1: Investment Details - NVIDIA will acquire Intel shares at a price of $23.28 per share, reminiscent of Microsoft's investment in Apple 28 years ago [3]. - This investment is part of NVIDIA's strategy to enhance its ecosystem and improve investment efficiency, as it has been more active in making investments compared to other tech giants [6][4]. - The investment is viewed as a potential opportunity to "bottom out" Intel's stock, especially after Intel's recent transactions that have raised concerns about its financial health [7]. Group 2: Strategic Implications - The collaboration between NVIDIA and Intel is expected to create synergies in the PC ecosystem, particularly in the high-end GPU market, which has not been fully tapped [13][14]. - NVIDIA's GPUs, while powerful, still require CPUs, and this partnership could enhance the integration of CPU and GPU technologies, benefiting both companies [11][12]. - The partnership is also seen as a strategic move against AMD, which has been gaining market share in the CPU space [16]. Group 3: Intel's Perspective - For Intel, this partnership represents a new beginning, as it seeks to redefine its identity and recover from past challenges [17][19]. - The collaboration is part of a broader strategy under CEO Pat Gelsinger to revitalize Intel's operations and explore new market opportunities [19][21]. - Despite the potential benefits, there are concerns about Intel's ability to secure significant orders for its foundry services, which may limit the impact of NVIDIA's investment [23][25].

Core Insights - CoreWeave, Inc. (CRWV) stock has increased by 24.5% in the past month, outperforming the Zacks Internet-Software Market's gain of 1.9% and the S&P 500 Composite's gain of 3.6% [1] - The momentum is driven by strong AI demand trends and company-specific developments, including a new order form valued at $6.3 billion with NVIDIA Corporation [4][8] - CoreWeave's management has raised 2025 revenue guidance to $5.15-$5.35 billion, citing accelerating demand and a robust pipeline [7] Company Developments - CoreWeave launched a Ventures Group to invest in AI start-ups, enhancing its competitive position in the AI infrastructure space [5] - The company has a significant revenue backlog of $30.1 billion, bolstered by strategic partnerships with OpenAI and NVIDIA [8] - CoreWeave is expanding its data center network, targeting over 900 MW of active power by year-end, with key projects including a $6 billion data center investment [12] Financial Performance - CoreWeave's second quarter capex was $2.9 billion, an increase of $1 billion from the previous quarter, with full-year capex guidance reaffirmed at $20-$23 billion [16] - The company reported a net loss of $291 million in the second quarter, primarily due to high interest expenses, which surged to $267 million compared to $67 million a year ago [15] Competitive Landscape - CoreWeave faces tough competition in the AI cloud infrastructure space from major players like Amazon and Microsoft, as well as emerging competitors like Nebius, which reported a revenue growth of 625% in the last quarter [17][18] - The competitive environment is intensifying, with Nebius securing a deal with Microsoft valued at approximately $17.4 billion [18] Valuation Concerns - CoreWeave appears overvalued with a Price/Book ratio of 20.77X, significantly higher than the industry average of 6.93X [21] - Analysts have revised earnings estimates downward for the current quarter, indicating potential challenges ahead [19]

AI Infrastructure & Market Opportunity - Aethir is positioned as an AI infrastructure cash cow, similar to Pump but in the AI sector [1] - Aethir operates as a decentralized cloud platform, providing enterprise-grade GPU-as-a-Service [1] - NVIDIA H100/200 chips are identified as a key bottleneck for AI training, highlighting Aethir's potential role in addressing this constraint [1] Aethir's Business Model - Aethir's business model is based on delivering enterprise-grade GPU-as-a-Service [1] - The company's valuation is implied to be attractive, with a comparison to Pump at 13x revenue [1]

AI Infrastructure & Market Opportunity - Aethir is positioned as an AI infrastructure cash cow, similar to Pump but in the AI sector [1] - Aethir operates as a decentralized cloud platform, providing enterprise-grade GPU-as-a-Service [1] - NVIDIA H100/200 chips are identified as a key bottleneck for AI training, highlighting Aethir's potential role in addressing this constraint [1] Aethir's Business Model - Aethir's business model is based on delivering enterprise-grade GPU-as-a-Service [1] - The company's valuation is implied to be attractive, with a comparison to Pump at 13x revenue [1]

Core Insights - China's electricity consumption reached 1.02 trillion kilowatt-hours last month, marking the first time it has surpassed the trillion-kilowatt-hour threshold in a month [1] - This achievement reflects China's strategic prowess in energy management, emphasizing that "electricity is national power" [2][4] - Unlike many countries that faced energy crises, China maintained stable electricity supply during extreme weather, showcasing the effectiveness of its "strong grid" strategy developed over the past decade [6] Electricity Generation and Global Standing - By 2024, China's annual electricity generation is projected to reach 10.1 trillion kilowatt-hours, more than double that of the United States, accounting for 32.3% of global electricity generation [7] - This substantial generation capacity positions China as a cornerstone of the global energy system, earning it the title of the first true "electric power empire" in human history [7] Implications for AI Development - The increasing electricity supply is crucial for the development of AI, which relies heavily on power for its infrastructure [9] - For instance, a single NVIDIA H100 AI server consumes nearly 1.2 kilowatts per hour, and a cluster of 10,000 such servers would require over 100 million kilowatt-hours annually, equivalent to the annual electricity consumption of a small city [9][10] - The International Energy Agency predicts that global data centers will consume over 10 trillion kilowatt-hours by 2026, surpassing Germany's total electricity consumption for 2023 [10] Competition for Energy Resources - Major tech companies are competing for stable electricity supplies, with firms like Meta signing long-term nuclear energy procurement contracts [10][11] - Companies such as Microsoft are actively seeking reliable power generation projects, while others face project delays due to insufficient electricity [11] Transition to Electric Vehicles - Electricity is becoming a key factor in reducing dependence on oil, particularly in the context of electric vehicle (EV) development [13] - Many European countries struggle with low EV adoption rates due to inadequate grid capacity and high electricity costs, while China has successfully supported millions of EVs and established extensive high-speed charging networks [14][15] - China's ability to provide stable electricity across urban and rural areas, including remote regions, enables its EV market to thrive, contrasting with the challenges faced by other countries [14][15] Conclusion - Over the past few decades, China has built the world's strongest electricity grid and generation capacity, positioning itself favorably in the upcoming technological competition [17] - In an era where "electricity equals national power," China's early strategic investments in energy infrastructure have set it apart as a leading player in the global energy landscape [17]

Core Viewpoint - IREN Limited has significantly expanded its GPU fleet by procuring an additional 4.2k NVIDIA Blackwell B200 GPUs, bringing the total to approximately 8.5k GPUs, and has secured $102 million in financing for prior GPU purchases, positioning the company for growth in AI Cloud services [1][2][4]. Financing Details - IREN has secured $102 million in financing structured as a 36-month lease for 100% of the purchase price of NVIDIA Blackwell GPUs, with lease payments based on a high single-digit interest rate [2]. - Financing discussions are ongoing for the newly acquired 4.2k NVIDIA Blackwell B200 GPUs, with initial funding sourced from existing cash [3]. Capacity and Growth - The new GPUs will be installed at IREN's Prince George campus, maintaining a total installed mining capacity of approximately 50 EH/s, utilizing spare data center capacity efficiently [3]. - The Prince George campus has a total power capacity of 50 MW, allowing for phased growth to support up to 20,000 Blackwell GPUs [4]. Strategic Positioning - The expansion of GPU capacity is aimed at capturing strong demand and driving revenue growth in the AI Cloud sector, leveraging competitively priced, non-dilutive capital [4]. - IREN operates a vertically integrated data center business focused on Bitcoin, AI, and other high-performance computing applications, utilizing 100% renewable energy [10].

Core Viewpoint - NVIDIA has urged customers to enable Error-Correcting Code (ECC) to defend against a new variant of RowHammer attacks targeting its GPUs, known as GPUHammer, which can manipulate data in GPU memory [3][4][5]. Group 1: GPUHammer Attack Details - GPUHammer is the first RowHammer exploit specifically targeting NVIDIA GPUs, allowing malicious users to flip bits in GPU memory and alter data of other users [3]. - The most alarming consequence of this attack is a drastic drop in AI model accuracy, from 80% to below 1% [4]. - Unlike CPUs, which have benefitted from side-channel defense research, GPUs lack parity checks and instruction-level access control, making them more vulnerable to low-level fault injection attacks [5]. Group 2: Impact on AI Models - In a proof-of-concept, single-bit flips were used to corrupt an ImageNet deep neural network model, reducing its accuracy from 80% to 0.1% [5]. - GPUHammer poses a broader threat to AI infrastructure, encompassing various attacks from GPU-level faults to data poisoning and model pipeline intrusions [5][6]. Group 3: Shared GPU Environment Risks - In shared GPU environments, such as cloud machine learning platforms, malicious tenants can launch GPUHammer attacks against adjacent workloads, affecting inference accuracy and corrupting cached model parameters without direct access [7]. - This introduces cross-tenant risks that are often overlooked in current GPU security considerations [7]. Group 4: Recommendations and Mitigations - To mitigate the risks posed by GPUHammer, enabling ECC is recommended, although it may reduce the performance of A6000 GPUs by 10% and decrease memory capacity by 6.25% [9][10]. - Monitoring GPU error logs for ECC-related corrections can help identify ongoing bit-flip attempts [9]. - Newer NVIDIA GPUs, such as H100 or RTX 5090, are not affected due to on-chip ECC capabilities [9]. Group 5: Broader Implications - The implications of GPUHammer extend to edge AI deployments, autonomous systems, and fraud detection engines, where silent corruption may be difficult to detect or reverse [9]. - Organizations deploying GPU-intensive AI must incorporate GPU memory integrity into their security and audit frameworks to comply with regulatory standards [10]. Group 6: AMD Vulnerabilities - AMD has warned of a new side-channel attack, Transient Scheduler Attack (TSA), affecting multiple chip models, which could lead to information leakage [11][12]. - The vulnerabilities are rated as medium to low severity, but their complexity means only attackers with local access can exploit them [11][13]. - AMD suggests updating to the latest Windows versions to mitigate these vulnerabilities, although the attacks are difficult to execute [19].

Core Viewpoint - The article discusses the advancements in AI data center architecture, particularly focusing on Huawei's CloudMatrix384, which aims to address the limitations of traditional AI clusters by providing a more efficient, flexible, and scalable solution for AI computing needs [5][12][49]. Group 1: AI Computing Demand and Challenges - Major tech companies are significantly increasing their investments in GPU resources to enhance AI capabilities, with examples like Elon Musk's plan to expand his supercomputer by tenfold and Meta's $10 billion investment in a new data center [1]. - Traditional AI clusters face challenges such as communication bottlenecks, memory fragmentation, and fluctuating resource utilization, which hinder the full potential of GPUs [3][4][10]. - The need for a new architecture arises from the inability of existing systems to meet the growing computational demands of large-scale AI models [10][11]. Group 2: Huawei's CloudMatrix384 Architecture - Huawei's CloudMatrix384 represents a shift from simply stacking GPUs to a more integrated architecture that allows for high-bandwidth, peer-to-peer communication and fine-grained resource decoupling [5][7][14]. - The architecture integrates 384 NPUs and 192 CPUs into a single super node, enabling unified resource management and efficient data transfer through a high-speed, low-latency network [14][24]. - CloudMatrix384 achieves impressive performance metrics, such as a throughput of 6688 tokens/s/NPU during pre-fill and 1943 tokens/s/NPU during decoding, surpassing NVIDIA's H100/H800 [7][28]. Group 3: Innovations and Technical Advantages - The architecture employs a peer-to-peer communication model that eliminates the need for a central CPU to manage data transfers, significantly reducing communication overhead [18][20]. - The UB network design ensures constant bandwidth between any two NPUs/CPUs, providing 392GB/s of unidirectional bandwidth, which enhances data transfer speed and stability [23][24]. - Software innovations, such as global memory pooling and automated resource management, further enhance the efficiency and flexibility of the CloudMatrix384 system [29][42]. Group 4: Cloud-Native Infrastructure - CloudMatrix384 is designed with a cloud-native approach, allowing users to deploy AI applications without needing to manage hardware intricacies, thus lowering the barrier to entry for AI adoption [30][31]. - The infrastructure software stack includes modules for resource allocation, network communication, and application deployment, streamlining the process for users [33][40]. - The system supports dynamic scaling of resources based on workload demands, enabling efficient utilization of computing power [45][51]. Group 5: Future Directions and Industry Impact - The architecture aims to redefine AI infrastructure by breaking the traditional constraints of power, latency, and cost, making high-performance AI solutions more accessible [47][49]. - Future developments may include expanding node sizes and further decoupling resources to enhance scalability and efficiency [60][64]. - CloudMatrix384 exemplifies a competitive edge for domestic cloud solutions in terms of performance and cost-effectiveness, providing a viable path for AI implementation in Chinese enterprises [56][53].

Core Insights - CoreWeave (CRWV) and Microsoft Corporation (MSFT) are key players in the AI infrastructure market, with CRWV focusing on GPU-accelerated services and Microsoft leveraging its Azure platform [2][3] - CRWV has shown significant revenue growth driven by AI demand, while Microsoft maintains a strong position through extensive investments and partnerships [5][9] CoreWeave (CRWV) - CRWV collaborates with NVIDIA to implement GPU technologies and was among the first to deploy NVIDIA's latest clusters for AI workloads [4] - The company reported revenues of $981.6 million, exceeding estimates by 15.2% and increasing 420% year-over-year, with a projected global economic impact of AI reaching $20 trillion by 2030 [5] - CRWV has a substantial backlog of $25.9 billion, including a strategic partnership with OpenAI valued at $11.9 billion and a $4 billion expansion agreement with a major AI client [6] - The company anticipates capital expenditures (capex) between $20 billion and $23 billion for 2025 to meet rising customer demand, with interest expenses projected at $260-$300 million for the current quarter [7] - A significant risk for CRWV is its revenue concentration, with 77% of total revenues in 2024 coming from its top two customers [8] Microsoft Corporation (MSFT) - Microsoft is a dominant force in AI infrastructure, with Azure's global data center coverage expanding to over 60 regions [9] - The company invested $21.4 billion in capex in the last quarter, focusing on long-lived assets to support its AI initiatives [10] - Microsoft has a $315 billion customer backlog and is the exclusive cloud provider for OpenAI, integrating AI models into its services to enhance monetization opportunities [12] - The company projects Intelligent Cloud revenues between $28.75 billion and $29.05 billion for Q4 fiscal 2025, with Azure revenue growth expected at 34%-35% [14] Share Performance - In the past month, CRWV's stock surged by 69%, while MSFT's stock increased by 8% [17] - Current Zacks Rank indicates MSFT as a better investment option compared to CRWV, which has a lower rank [18]