Core Viewpoint - AMD is facing significant challenges in the AI chip market, primarily due to supply chain constraints and competition from NVIDIA, despite recent stock price recovery and product launches [1][3][12]. Group 1: AMD's Product Developments - AMD's MI300 series was launched in December 2023, featuring memory 2.4 times that of NVIDIA's H100 and memory bandwidth 1.6 times higher, positioning it as a strong competitor in the AI chip market [3]. - The MI325X chip, intended to compete with NVIDIA's H200, is expected to start production in Q4 2024, but faces competition from NVIDIA's B200 [7][8]. - The MI350 series, released in June, utilizes TSMC's 3nm process and integrates 1,850 billion transistors, with a performance capability of 20 PFLOPS, significantly enhancing AMD's competitive stance [17][24]. Group 2: Market Performance and Stock Movements - AMD's stock price rebounded from a low of $76.48 in April to over $170, a rise of more than 100%, driven by the MI350 series launch and expectations of increased market share in server CPUs [12][15]. - Despite the stock price increase, AMD's sales expectations for the MI300 series remain below market expectations, primarily due to supply chain limitations rather than demand [6][21]. Group 3: Competitive Landscape - NVIDIA's dominance in the AI chip market is underscored by its significant market share and superior product performance, with NVIDIA's data center revenue growing by 154% compared to AMD's 14% [24]. - AMD's market share in the server market has increased from 2% to 36.5% over eight years, but it still lags behind NVIDIA in overall revenue and profitability [12][24]. - The software ecosystem remains a critical disadvantage for AMD, with NVIDIA's CUDA framework covering over 90% of AI developers, while AMD's ROCm7 lacks the same level of community support and optimization [27]. Group 4: Future Outlook - The upcoming MI400 series aims to compete directly with NVIDIA's DGX SuperPod ecosystem, with significant investments in AI infrastructure expected from major tech companies [19][20]. - AMD's strategy focuses on being a viable alternative in the AI chip market, with the belief that the market can support multiple players, rather than solely aiming to defeat NVIDIA [30].

Core Viewpoint - AMD is facing significant challenges in the AI chip market, primarily due to supply chain constraints and competition from NVIDIA, despite recent stock price recovery and product launches [2][3][5]. Group 1: AMD's Market Position and Product Launches - AMD's MI300 series was launched in December 2023, aiming to compete directly with NVIDIA's H100 series, featuring 2.4 times the memory and 1.6 times the memory bandwidth of H100 [10][11]. - The stock price of AMD surged nearly 75% from Q4 2023 to Q1 2024, driven by high expectations for MI300's sales [12]. - Despite the initial boost from MI300, AMD's sales expectations fell short, leading to market skepticism [15]. Group 2: Supply Chain and Competitive Landscape - Supply chain limitations, rather than market demand, are constraining MI300's output, as NVIDIA has secured a significant portion of TSMC's CoWos packaging capacity [16]. - AMD's MI325X, intended to compete with NVIDIA's H200, faces challenges due to slower iteration speed and competition from NVIDIA's B200, which has superior hardware specifications [18][40]. - AMD's market share in the server segment has increased from 2% to 36.5% over eight years, aided by the release of the MI350 series and easing export restrictions [22]. Group 3: Financial Performance and Future Outlook - AMD's financial performance remains below that of NVIDIA, with a 14% year-over-year growth in data center revenue compared to NVIDIA's 154% [40]. - The company reported a gross margin of 40%, which could rise to 54% if inventory write-downs are excluded [26]. - Analysts express concerns over AMD's ability to meet the strong demand for AI infrastructure, with unclear revenue projections for the second half of the year [36][38]. Group 4: Competitive Advantages and Challenges - AMD's MI350 series, utilizing TSMC's 3nm process, integrates 185 billion transistors and offers competitive performance at a lower price point than NVIDIA's offerings, appealing to cost-sensitive customers [28][42]. - However, NVIDIA's established ecosystem and rental services provide a competitive edge, making it difficult for AMD to gain traction in the mid-term rental market [43]. - The software ecosystem remains a critical weakness for AMD, with NVIDIA's CUDA dominating over 90% of AI developers, while AMD's ROCm lacks depth and community support [44]. Group 5: Market Potential and Strategic Positioning - The AI chip market is large enough to accommodate multiple players, and AMD aims to position itself as a viable alternative rather than a direct competitor to NVIDIA [49][50]. - The ongoing demand for AI infrastructure is expected to drive higher revenue ceilings for AMD, with potential market share recovery anticipated by the end of 2023 or early 2024 [47].

Group 1: AMD - The MI350 series is expected to drive revenue growth in the second half of 2025, with management guiding a median revenue of $8.7 billion for Q3 2025, exceeding previous expectations [1][2] - Q2 2025 revenue reached $7.7 billion with a Non-GAAP gross margin of 43.3%, both surpassing forecasts [1] - The target price for AMD has been raised to $196, reflecting a potential upside of 12.4% based on a 35x P/E ratio for 2026 [2] Group 2: Property Trust - The property trust reported a slight revenue decline of 2.0% year-on-year to HKD 854.5 million for the first half of 2025, with net property income down 3.2% to HKD 612.6 million [3][7] - Distributable income and distribution per unit (DPU) increased by 2.1% and 1.0% year-on-year, respectively, aligning with expectations [3] - The average interest cost decreased significantly to 3.5%, down 60 basis points year-on-year, which is expected to support further increases in DPU [3] Group 3: Internet Industry - Tencent's domestic revenue decreased by 12% year-on-year, while overseas revenue showed a positive growth of 5% [8][9] - New game launches are expected to mitigate the impact of high revenue bases in the second half of the year, with an estimated annual game revenue growth of 12% [9] - NetEase's domestic revenue grew by 5%, driven by new game contributions, indicating a stable performance in the mobile gaming sector [8][9]

Core Insights - AMD reported Q2 2025 revenue of $7.7 billion, a 32% year-over-year increase, exceeding market expectations of approximately $7.4 billion [1] - The company maintains an optimistic outlook for Q3 2025, projecting revenue of $8.7 billion, above market expectations of $8.3 billion, with a forecasted gross margin recovery to 54% [2] - Despite strong performance in client and gaming segments, the data center business growth was below market expectations, primarily due to export restrictions impacting MI308 chips [1][2] Financial Performance - Total revenue for Q2 2025 reached $7.7 billion, with a non-GAAP EPS of $0.48, and a gross margin of 43%, which could rise to 54% when excluding $800 million in costs from export restrictions [1] - Client business revenue was $2.5 billion, up 57% year-over-year, driven by strong sales of the Zen 5 architecture Ryzen processors [1] - Gaming revenue surged to $1.1 billion, a 73% increase year-over-year, significantly contributing to overall revenue growth [1] Business Segment Analysis - Data center revenue was $3.2 billion, a 14% year-over-year increase, but fell short of market expectations due to MI308 chip export limitations [1][3] - Embedded business revenue declined by 4% year-over-year to $824 million [1] - The client and gaming segments were the main drivers of revenue growth, with the PC market recovery and new Ryzen processors boosting client sales [1][3] Market Reaction - Following the earnings report, AMD's stock price fell approximately 3%, attributed to concerns over data center growth lagging behind competitors like NVIDIA, ongoing export restrictions, and profit-taking after significant stock price increases earlier in the year [2][3] - AMD's year-to-date stock price increase exceeds 120%, prompting some investors to realize gains post-earnings [2] Competitive Landscape - In comparison, Intel reported Q2 revenue of $12.86 billion, a mere 0.8% increase, with an EPS of -$0.10, indicating a loss [3] - NVIDIA is expected to report Q2 revenue of approximately $45 billion, a 66% year-over-year increase, with an EPS of around $0.93 [3] - Overall, while AMD's Q2 performance and future guidance surpassed expectations, challenges in the data center segment and export uncertainties have pressured stock performance [3]

Overall Performance - AMD reported Q2 2025 revenue of $7.69 billion, a year-over-year increase of 31.7%, exceeding market expectations of $7.43 billion. The growth was driven by client and gaming segments, as well as data center business [1][6] - The company's GAAP gross margin was 39.8%, significantly down due to an $800 million inventory impairment related to MI308. Excluding this impact, the actual gross margin would be 50.2%, flat compared to the previous quarter [1][6] Operating Expenses - R&D expenses for the quarter were $1.89 billion, up 19.6% year-over-year, while sales and administrative expenses rose 52% to $990 million [1][2] - The core operating expense ratio reached 37.5%, affecting profit margins [2] Profitability - AMD achieved a net profit of $872 million, influenced by non-recurring items. The core operating profit was $174 million, reflecting a significant decline with a core operating margin of only 2.3% [2][6] Business Segments - Client segment revenue grew to $2.5 billion, a 67.5% increase year-over-year, capturing over 50% of the desktop CPU market [2][7] - Data center revenue reached $3.24 billion, up 14.3% year-over-year, driven by increased server CPU shipments and improved product competitiveness [2][9] - The gaming segment saw revenue of $1.12 billion, a notable increase compared to previous quarters [5] Future Guidance - AMD expects Q3 2025 revenue to be between $8.4 billion and $9 billion, with a midpoint of $8.7 billion, reflecting a 13.2% sequential growth. This guidance does not include any revenue from MI308 sales to China [4][6] - The company anticipates a non-GAAP gross margin of around 54% for the next quarter [4] AI and Product Development - The MI350 series is expected to significantly boost AI GPU revenue, with projections of over $1.7 billion for the next quarter, aligning closely with market expectations [12][18] - AMD's market share in server CPUs has increased from 10% to approximately 30%, driven by the combination of CPU and GPU offerings [9][11] Market Trends - The overall demand for AI-related products remains strong, with major cloud companies expected to increase capital expenditures significantly in 2025 [14][17] - AMD's valuation reflects optimistic market expectations, with a projected CAGR of 30.3% for core operating profit from 2025 to 2029 [16][17]

Core Viewpoint - Goldman Sachs maintains a "Neutral" rating on AMD (AMD.US) with a target price of $140, citing impressive returns and revenue but concerns over growth potential due to intense competition in the AI and GPU sectors [1] Financial Performance - AMD's revenue has grown by 21.71% over the past 12 months, reaching $27.75 billion, with a return of 19.28% over the last six months, reflecting strong investor enthusiasm for AI [1] - The company's current P/E ratio stands at 101.36, indicating a historically high valuation [1] Competitive Landscape - AMD faces challenges in the server CPU market as the adoption of ARM-based processors increases, which are favored for their efficiency and scalability [2] - The company’s commercial GPU products are under significant competition from NVIDIA, which benefits from a strong AI software stack and established ecosystem [2] - Increased funding is flowing towards ASICs designed for specific AI tasks, presenting another challenge for AMD [2] Analyst Sentiment - Analysts express uncertainty regarding AMD's AI future, with mixed reports on its prospects. KeyBanc Capital Markets maintains a "Market Perform" rating, projecting $7-8 billion in AI revenue from the MI355 AI GPU by 2025 [3] - Truist Securities holds a "Hold" rating, citing uncertainties in the company's data center GPU strategy and customer acceptance [3] - Mizuho raises its target price to $152, adjusting revenue expectations for the June quarter to $7.4 billion [3] Optimistic Outlook - Melius Research upgrades AMD to a "Buy" rating with a new target price of $211, based on potential EPS of $8 within two years [4] - HSBC also upgrades AMD to "Buy," doubling its target price from $100 to $200, citing the premium pricing of the new MI350 series products [4] - The MI350 chips' compatibility with existing data center infrastructure is seen as an attractive feature [4] Upcoming Financial Disclosure - AMD is set to release its Q2 fiscal year 2025 earnings report on August 5, which is anticipated to provide insights into its AI plans, revenue trajectory, and profitability amid significant R&D investments [5]

证券研究报告 2025 年 6 月 23 日 湘财证券研究所 | 1.《Marvell上调算力ASIC市场 | | | --- | --- | | 规模预期，算力ASIC需求强 | | | 劲》 | 2025.06.21 | | 2.《AMD推出旗舰新品MI350系 | | | 列，AI推理需求强劲增长》 | | | 2025.06.15 | | | 3.《博通公布二季报，XPU需求 | | | 强劲》 | 2025.06.08 | 行业评级：增持（维持） 近十二个月行业表现 % 1 个月 3 个月 12 个月 相对收益 0.24 -6.51 14.66 绝对收益 -1.08 -9.74 24.46 -20.00% 0.00% 20.00% 40.00% 60.00% 24-06 24-07 24-08 24-09 24-10 24-11 24-12 25-01 25-02 25-03 25-04 25-05 25-06 电子(申万) 沪深300 注：相对收益与沪深 300 相比 联系人：李杰 证书编号：S0500521070001 Tel：(8621) 50293520 Email：lijie5@xcsc. ...

Core Viewpoint - AMD has launched the MI350X and MI355X GPUs, claiming they outperform NVIDIA's B200 in AI model inference by 30% and offer significant improvements in performance and efficiency [1][4][24]. Group 1: Product Launch and Specifications - AMD introduced two new GPUs, MI350X and MI355X, built on a 3nm process with 185 billion transistors and equipped with HBM3E memory [2]. - The MI350 series boasts a fourfold increase in computing power and a 35-fold increase in inference speed compared to the previous MI300X [3]. - Both models feature the fourth-generation Instinct architecture (CDNA 4), with 288GB of HBM3E memory and 8TB/s memory bandwidth, which is 1.6 times that of NVIDIA's B200 [9]. Group 2: Performance Comparison - The MI350X and MI355X have peak FP64 performance of 72 TFLOPs and 78.6 TFLOPs, respectively, which is twice that of NVIDIA's equivalent [12]. - In lower precision formats (FP16, FP8, FP4), the MI350 series performs comparably or slightly better than NVIDIA [13]. - The MI355X can process 40% more tokens per dollar spent compared to the B200, indicating higher cost efficiency [5][28]. Group 3: Future Developments - AMD plans to release the MI400 series next year, developed in collaboration with OpenAI, which is expected to be 10 times faster than the MI300 series [7][41]. - The MI400 series will feature the next-generation CDNA architecture, with up to 432GB of HBM4 memory and 19.6TB/s memory bandwidth [42]. - AMD's roadmap includes the introduction of the MI500 series GPUs and Verono CPUs by 2027, aiming to further enhance performance and scalability [52].

Core Viewpoint - AMD has shown significant growth in revenue and market share, particularly in the data center and AI sectors, driven by strong demand for its EPYC CPUs and Instinct GPUs [1][3][5]. Financial Performance - In the first quarter, AMD reported revenue of $7.4 billion, a 36% year-over-year increase, marking the fourth consecutive quarter of accelerating revenue growth [1]. - The data center segment generated $3.7 billion in revenue, up 57% year-over-year, largely due to increased sales of EPYC CPUs and Instinct GPUs [1]. - Customer revenue reached a record $2.3 billion, reflecting a 68% year-over-year growth, driven by strong demand for the latest "Zen 5" AMD Ryzen processors [1]. Market Share and AI Outlook - AMD's market share in server CPUs rose from 2% in 2018 to 40% in the first quarter of this year, indicating a strong presence in the market [3]. - AMD's CEO, Lisa Su, projected that the data center AI accelerator market could exceed $500 billion by 2028, with particularly rapid growth in AI inference demand [3]. AI Strategy and Acquisitions - AMD has made several strategic acquisitions to enhance its AI capabilities, including Mipsology, Nod.ai, Silo AI, ZT Systems, Enosemi, Brium, Untether AI, and Lamini [5][10]. - The company has developed a comprehensive AI stack, leveraging its CPU and GPU expertise along with its acquisitions to strengthen its position in the AI market [5][10]. Product Development and Innovations - AMD introduced the MI350X and MI355X AI GPUs, which offer up to 4 times the performance compared to the previous generation and 35 times faster inference speeds [13][16]. - The new GPUs utilize a 3nm process and feature up to 288GB of HBM3E memory and 8TB/s memory bandwidth, significantly outperforming competitors [15][16]. - AMD's upcoming MI400 series, set to launch in 2026, will double the performance of the MI355X and support HBM4 memory, enhancing memory capacity and bandwidth [30][35]. Ecosystem and Software Development - AMD has established an open-source ecosystem, including the ROCm software stack, which has been widely adopted by industry leaders like OpenAI, Microsoft, and Meta [9][43]. - The latest ROCm 7 version focuses on improving inference performance by up to 3.5 times compared to its predecessor, ROCm 6 [43]. Networking and Scalability - AMD's acquisition of Pensando has bolstered its networking capabilities, allowing for enhanced system scalability [37]. - The company is introducing the Pollara 400 AI NIC, which integrates UltraEthernet technology to optimize high-speed network reliability and scalability [37][39].

Core Viewpoint - AMD has shown significant growth in revenue and market share, particularly in the data center and AI sectors, driven by strong demand for its latest products and strategic acquisitions [1][3][4]. Financial Performance - In the first quarter, AMD reported revenue of $7.4 billion, a 36% year-over-year increase, marking the fourth consecutive quarter of accelerating revenue growth [1]. - The data center segment generated $3.7 billion in revenue, up 57% year-over-year, largely due to increased sales of AMD EPYC CPUs and AMD Instinct GPUs [1]. - Customer revenue reached a record $2.3 billion, reflecting a 68% year-over-year growth, driven by strong demand for the new "Zen 5" AMD Ryzen processors [1]. Market Share and AI Outlook - AMD's market share in server CPUs rose from 2% in 2018 to 40% in the first quarter of this year, indicating a strong competitive position [3]. - AMD's CEO, Lisa Su, projected that the data center AI accelerator market could exceed $500 billion by 2028, with particularly rapid growth in AI inference demand [4]. Strategic Acquisitions - AMD has made several strategic acquisitions to bolster its AI capabilities, including Mipsology, Nod.ai, Silo AI, and ZT Systems, enhancing its position in the AI ecosystem [7][8]. - Recent acquisitions include Enosemi, Brium, Untether AI, and Lamini, aimed at improving AMD's AI capabilities, especially in the data center market [8]. AI Market Strategy - AMD has developed a diverse computing foundation for the AI market and is promoting an open-source ecosystem to foster collaboration and innovation [10]. - The ROCm software stack has been widely adopted by industry leaders like OpenAI, Microsoft, and Meta, enhancing AMD's AI deployment capabilities [10][11]. Product Innovations - AMD introduced the MI350 series GPUs, which offer up to 4 times the performance compared to the previous generation and 35 times faster inference speeds [13][15]. - The MI350X and MI355X GPUs feature 288 GB of HBM3E memory and 8 TB/s memory bandwidth, significantly outperforming competitors [15][19]. - AMD's next-generation MI400 series, set to launch in 2026, will double the performance of the MI355X and increase memory capacity to 432 GB [31][32]. Future Developments - AMD's upcoming EPYC "Venice" processors will utilize TSMC's 2nm process and feature up to 256 cores, promising a performance increase of up to 70% over the current generation [37][38]. - The Helios AI rack, powered by the MI400 series, aims to achieve exaflop-level performance, positioning AMD competitively against Nvidia's upcoming platforms [34][36].