Hon Hai has agreed to sell an EV plant in Ohio for $375 million, under an arrangement in which the Taiwanese company aims to shift toward assembling AI servers at the US facility https://t.co/L9nyiXFmY4 ...

Summary of Conference Call Records Company: Inspur (000977.SZ) Key Points 1. **Upgrade to Buy**: Inspur has been upgraded from Neutral to Buy due to improving GPU supply in the China market, new GPU platforms expected in 2H25, and the growth of local chipset platforms to meet generative AI demand in China. The 12-month price target (TP) has been raised to Rmb77.8 from Rmb55, indicating a potential upside of 39.5% [1][9][10]. 2. **Revenue and Earnings Growth**: Revenue projections for Inspur have been revised upwards by 8% for 2025, 25% for 2026, and 27% for 2027, primarily driven by higher AI training server revenues. The company expects a significant ramp-up in shipments of AI training servers due to better GPU supply and new product cycles [14][18]. 3. **Market Position**: Inspur is positioned as a leading AI server ODM in China, supplying major Chinese cloud service providers (CSPs) and expanding its offerings from global-tier GPU-powered servers to local chipset-powered servers. This transition is expected to enhance its market share and profitability [10][13][14]. 4. **Earnings Revision**: Earnings have been revised upwards by 7% for 2025, 11% for 2026, and 9% for 2027, despite a lower gross margin (GM) and higher operating expenses (Opex) ratio. The expected GM is projected to decline slightly due to the higher contribution from AI training servers, which typically have lower margins [14][18]. 5. **Sensitivity Analysis**: A sensitivity analysis indicates that a 10% increase in AI training server shipments could lead to a 6% increase in total revenues and a 2% increase in gross profit. Conversely, a decrease in shipments would have the opposite effect [17]. 6. **Investment Thesis**: The investment thesis for Inspur is based on the anticipated increase in capital expenditure (Capex) from Chinese CSPs, telecom operators, and government clients focused on AI training and inferencing. The emergence of advanced AI applications is also expected to drive growth [29][32]. 7. **Risks**: Key risks include slower-than-expected ramp-up of AI servers in China, potential GPU supply constraints, and intensified pricing competition among ODM suppliers, which could negatively impact earnings [27][28][32]. Company: Piotech (688072.SS) Key Points 1. **Downgrade to Neutral**: Piotech has been downgraded from Buy to Neutral due to pressures on near-term profitability stemming from new product ramp-up. The revised TP is set at Rmb185, reflecting a 17% upside potential [2][33]. 2. **Profitability Concerns**: The company's gross margin fell to 20% in Q1 2025 from 47% in Q1 2024, attributed to higher delivery and service costs associated with new products. The expectation is that it will take time for margins to recover to the historical average of around 40% for deposition tools [35][37]. 3. **Earnings Revision**: Earnings estimates have been reduced by 3% to 5% for 2025-2027, primarily due to lower margins and adjustments in government subsidies. The target P/E multiple has been reset from 42.5x to 37.7x to reflect these changes [34][38]. 4. **Market Position**: Piotech is recognized as a local leader in deposition tools, expanding into high-end products. However, the company faces challenges due to its exposure to competitive mature nodes and the costs associated with new product delivery [34][39]. 5. **Government Subsidies**: Expected government subsidies have been lowered, reflecting the company's shift towards self-funding for new production facilities. Despite this, there is an ongoing effort to deepen cooperation with government entities [37][39]. Conclusion Inspur is positioned for significant growth in the AI server market, supported by favorable supply conditions and increasing demand. Conversely, Piotech faces challenges that have led to a downgrade, highlighting the importance of managing new product costs and maintaining profitability in a competitive landscape.

Investment Rating - The report does not explicitly provide an investment rating for the industry or specific companies involved in AI ASIC servers. Core Insights - The AI ASIC market is expected to grow significantly, with Meta's MTIA AI servers potentially marking a milestone in 2026. The total AI ASIC volume could exceed nVidia's AI GPU volume by 2026 [1][3][12] - nVidia currently dominates the AI server market with over 80% market value, while ASIC AI servers hold approximately 8-11% market value share. However, unit comparisons suggest that Google and AWS's AI ASICs could reach 40-60% of nVidia's GPU volume by 2025 [1][3] - Meta's MTIA AI server is projected to ramp up significantly, with expectations of 1 to 1.5 million units of MTIA V1 and V1.5 by late 2025 to 2026 [10][12] Summary by Sections AI ASIC Market Dynamics - The AI ASIC market is experiencing aggressive growth, with more cloud service providers (CSPs) developing their own ASIC solutions, including Meta and Microsoft [1][3] - nVidia is responding to competition by unveiling NVLink Fusion, which allows inter-chip connections between its GPUs and third-party CPUs, indicating a proactive approach to maintain its market position [2] Meta's MTIA AI Server Development - Meta's MTIA AI server is set to launch its first ASIC (MTIA T-V1) by late 2025, with subsequent versions (V1.5 and V2) expected in mid-2026 and 2027, respectively [9][10] - The MTIA T-V1.5 is anticipated to be significantly more powerful than V1, with a larger interposer size and complex design [9][10] Supply Chain and Component Insights - Key suppliers for Meta's MTIA projects include Quanta, Unimicron, EMC, WUS, and Bizlink, which are expected to benefit from the growing demand for AI ASICs [12][13][14][15][19] - The report highlights the importance of baseboard management controllers (BMCs) in Meta's ASIC AI server development, with an estimated 23 BMCs per rack [17][18] Competitive Landscape - Despite nVidia's current leadership in AI computing, the report suggests that the gap is narrowing as ASIC solutions improve in specifications and performance [3][7] - The specifications of AI ASICs from companies like Google and AWS are catching up to nVidia's offerings, although nVidia still holds advantages in connectivity and ecosystem [7][8]

Investment Rating - The report assigns a "Buy" rating to several companies within the Taiwan Technology sector, including Wiwynn, Wistron, Hon Hai, Gigabyte, and various AI server components [25]. Core Insights - The Taiwan Technology sector is experiencing strong growth in ASIC AI servers and related components, with notable revenue increases reported by key players such as Wiwynn (+187% YoY) and Gigabyte (+108% YoY) [1][8]. - The report highlights a transition in server models expected in 3Q25, with a cautious outlook on rack-level AI server shipments due to macro uncertainties [9]. - The demand for components like liquid cooling, silicon photonics, rail kits, and chassis is anticipated to rise, driven by specification upgrades and the increasing adoption of AI technologies [9][10]. Summary by Sections ASIC AI Servers and Components - ASIC AI servers are showing robust growth, with Wiwynn's revenues up 187% YoY, significantly exceeding estimates [1][8]. - Baseboard-based AI servers also demonstrate strong performance, with Gigabyte achieving 108% YoY revenue growth [1][8]. - Liquid cooling solutions are gaining traction, with AVC reporting 87% YoY revenue growth [1][8]. Market Outlook - The report expresses caution regarding rack-level AI server shipments in the second half of 2025, while maintaining a positive outlook for ASIC and baseboard-based AI servers [9]. - The global market for AI servers is projected to reach 19,000 racks by 2025, indicating a significant growth opportunity [9]. Non-AI Sector Insights - The smartphone supply chain is expected to strengthen in 3Q25, with new high-end models being prepared for launch [10]. - AI PCs are anticipated to see continued penetration, with brands like ASUS and Lenovo leading the charge [10]. Investment Opportunities - Recommended investments include AI server ODMs and brands such as Wiwynn, Wistron, Hon Hai, and Gigabyte, as well as components like AVC, Fositek, and LandMark [11].

Investment Rating - Maintain Buy on Chenbro, Auras, and E Ink [2] Core Insights - Chenbro's May revenues declined by 7% MoM to NT$1.8 billion, primarily due to FX impact and a high base, but new rack products are expected to support growth in 3Q25 [4][10] - Auras experienced a 20% MoM revenue decline in May due to FX impact and diversification of production sites, but is expected to see growth from liquid cooling adoption and AI server expansion [23][26] - E Ink's April revenues exceeded expectations, with a 50% YoY increase, driven by rising demand for e-readers and e-labels, although FX impacts may affect future growth [42][43] Summary by Company Chenbro - May revenues were NT$1.8 billion, 12% below estimates, with expectations for June revenues to remain stable [4][10] - 3Q25 revenues are anticipated to be supported by new noise-cancellation rack products and increasing market share in ASIC AI servers [4][10] - Earnings revisions reflect a 3% reduction in net income estimates for 2025-27E due to lower revenues and higher operating expenses [11][14] Auras - May revenues fell to NT$1.62 billion, with expectations for June revenues to remain at similar levels [29][30] - The company is projected to see a 19% QoQ revenue growth in 2Q25, driven by liquid cooling components and AI server expansion [23][26] - Earnings revisions indicate a slight increase in revenue estimates for 2025-27E, with a target price raised to NT$825 based on a P/E multiple of 19.9x [41][30] E Ink - April revenues were NT$3.31 billion, 14% above estimates, with a 50% YoY growth attributed to increased demand for e-readers and e-labels [42][43] - Future revenues are expected to be stable, with a projected 15% QoQ growth in 2Q25 [47] - Earnings revisions show a reduction in net income estimates for 2025-27E by 19% due to lower revenue expectations and higher operating expenses [49][51]