2026.01.31 本文字数：1178，阅读时长大约2分钟 作者 |第一财经 郑栩彤 黄仁勋春节前访华到达上海、北京、深圳后，近日又到达中国台湾，宴请当地供应链厂商，并在接受采 访时回答了关于投资OpenAI、AI基础设施建设的话题。 据当地媒体报道，1月31日，黄仁勋到达中国台湾第三天，出现在台北四平商圈，并采购了蜜饯。黄仁 勋现场称，水果干、各类花生糖和其他零食都非常棒，是他儿时的味道。 此次行程中，黄仁勋与当地供应商的交流备受关注。1月31日晚间，黄仁勋宴请了英伟达的供应链伙 伴，参加晚宴的包括纬创董事长林宪铭、台积电董事长魏哲家、广达创始人林百里及副董事长梁次震、 和硕集团董事长童子贤、鸿海董事长刘扬伟、仁宝董事长陈瑞聪、联发科执行长蔡力行、华硕创始人施 崇棠等。晚宴结束后，黄仁勋送魏哲家离开。 针对ASIC（专用集成电路）带来的竞争，黄仁勋表示，ASIC一直有需求，但英伟达做的事情非常不一 样。一方面，英伟达不只做一种芯片，而是参与到整个AI基础设施建设过程，做的产品包括CPU、 GPU、网络芯片、交换器芯片等。另一方面，英伟达与几乎所有AI公司合作，包括谷歌。此外，英伟 达与每一个云都相关，一些 ...

本文字数：1178，阅读时长大约2分钟 作者 | 第一财经 郑栩彤 黄仁勋春节前访华到达上海、北京、深圳后，近日又到达中国台湾，宴请当地供应链厂商，并在接受采 访时回答了关于投资OpenAI、AI基础设施建设的话题。 2026.01. 31 据当地媒体报道，1月31日，黄仁勋到达中国台湾第三天，出现在台北四平商圈，并采购了蜜饯。黄仁 勋现场称，水果干、各类花生糖和其他零食都非常棒，是他儿时的味道。 此次行程中，黄仁勋与当地供应商的交流备受关注。1月31日晚间，黄仁勋宴请了英伟达的供应链伙 伴，参加晚宴的包括纬创董事长林宪铭、台积电董事长魏哲家、广达创始人林百里及副董事长梁次震、 和硕集团董事长童子贤、鸿海董事长刘扬伟、仁宝董事长陈瑞聪、联发科执行长蔡力行、华硕创始人施 崇棠等。晚宴结束后，黄仁勋送魏哲家离开。 在接受采访时，黄仁勋谈到供应链的动态。他表示，今年英伟达需求非常强劲，英伟达正在全力生产 Blackwell芯片，同时生产Rubin芯片。台积电今年必须要非常努力工作，因为英伟达需要很多晶圆和 CoWoS（一种先进封装技术）产能，台积电现在做得非常好。他还表示，台积电在未来10年可能会增 加100%产能 ...

Summary of Broadcom Inc. Conference Call Company and Industry - **Company**: Broadcom Inc (Ticker: AVGO) - **Industry**: U.S. Semiconductors Core Points and Arguments 1. **Investor Concerns**: Investors have expressed worries about rising competition and customer-owned tooling (COT) potentially impacting Broadcom's AI-dominant position. However, the company believes these concerns are overblown and unlikely to dethrone them in the ASIC space anytime soon [2][11]. 2. **Technological Advantages**: Broadcom claims to have unique technological, scale, and supply chain advantages, particularly in their XPU roadmaps. They are positioned to keep pace with NVIDIA's innovation in the AI space, which is seen as a critical factor for success [2][3][12]. 3. **TPU Shipping Projections**: Broadcom anticipates shipping "many millions" of TPUs in 2026, with hundreds of thousands of TPU v8 units expected to ship monthly by year-end. The previously mentioned $73 billion order number is now considered "significantly higher" [4][15]. 4. **Financial Outlook**: The company remains bullish on its AI story, with current valuations providing an attractive entry point. The stock is rated as Outperform with a price target of $475 [5][16]. 5. **Supply Chain Management**: Broadcom is actively managing its supply chain, working with all HBM vendors and ensuring that they have dedicated substrate supply. They are focused on a limited number of large LLM customers, which allows for tighter management of resources [14]. Additional Important Information 1. **Innovations in Chip Technology**: Broadcom is innovating with 3D chip stacking and 400G serdes, which are expected to provide significant performance advantages over competitors. They have also built a substrate factory in Singapore to secure supply and manage costs effectively [3][13]. 2. **Financial Metrics**: - **Adjusted EPS**: Expected to grow from $6.82 in FY2025 to $14.86 in FY2027, indicating a strong CAGR [8]. - **Market Cap**: Approximately $1,576.38 billion [6]. - **Performance**: The stock has shown a 45% increase over the past 12 months [6]. 3. **Risks**: Potential risks to the price target include unexpected weakness in AI demand, share losses at key customers, and failure to execute on merger synergies [27]. This summary encapsulates the key takeaways from the conference call, highlighting Broadcom's strategic positioning, financial outlook, and the competitive landscape within the semiconductor industry.

Core Insights - The company announced new orders of 2.5 billion yuan for Q4 2025, representing a 130% year-over-year increase and a 56% quarter-over-quarter increase, confirming the high growth trend in the company and industry [1] - The company is expected to benefit from the ongoing transformation in the industry as internet companies increase their investment in AI computing power and develop stronger computing systems [1] - The stock price currently corresponds to an 8x price-to-sales ratio for 2027, and recent adjustments due to shareholder reductions and unlocked shares present a long-term buying opportunity, maintaining a buy recommendation [1] New Orders Growth - In Q4 2025, the company secured new orders of 2.5 billion yuan, with over 84% related to AI computing orders and nearly 76% in the data processing sector [1] - The total new orders for 2025 are expected to reach 5.8 billion yuan, reflecting a 150% year-over-year growth, further validating the high growth trend in the company and industry [1] - The company’s performance is expected to see significant growth over the next two years [1] Shareholder Reduction Impact - The major shareholder, the Big Fund, plans to reduce its stake by 1.7% through centralized bidding and block trading, with a reduction period from January 21, 2026, to April 20, 2026 [2] - Although the reduction may create short-term pressure, the long-term impact on the company is expected to be limited [2] Profit Forecast - The company has been focusing on Chiplet technology and its applications in generative AI and smart driving for the past five years, aiming to expand into emerging markets and attract leading industry clients [2] - Revenue projections for 2025, 2026, and 2027 are 3.06 billion yuan, 5.41 billion yuan, and 8.47 billion yuan, with year-over-year growth rates of 32%, 77%, and 57% respectively [2] - Net profits are forecasted to be 26 million yuan, 270 million yuan, and 640 million yuan for the same years, with corresponding EPS of 0.05 yuan, 0.51 yuan, and 1.21 yuan [2]

Core Insights - Intel's AI strategy is focusing on two main areas: ASIC and edge AI, aiming to regain competitiveness in the AI sector where it lags behind Nvidia and AMD [2][4] - The establishment of the Central Engineering Group (CEG) is intended to consolidate engineering talent and enhance the company's capabilities in ASIC and design services [4][5] Group 1: AI Strategy and Market Position - Intel has acknowledged its shortcomings in AI strategy, with former CEO Pat Gelsinger admitting that the company's approach has not been satisfactory [2] - The company is developing a power-optimized GPU for inference as part of its edge AI strategy, with products like Meteor Lake and Lunar Lake aimed at enhancing mobile SoC performance [2][3] - Intel's new ASIC department, led by Srini Iyengar, aims to provide customized chips for specific workloads, competing with established solutions like Google's TPU and Amazon's Trainium [3][4] Group 2: ASIC Business Development - The ASIC business is expected to play a crucial role in Intel's operations, with plans to offer a "one-stop" solution for customers seeking custom AI chips [4][6] - Intel's CEG will lead the development of ASIC and design services, expanding the application of its core x86 IP and leveraging its design advantages [5][6] - The company aims to differentiate itself by providing internal foundry services, which is a unique offering compared to competitors like Broadcom and Marvell [6][7] Group 3: Future Prospects and Challenges - If executed effectively, the custom chip business could become a significant revenue stream for Intel, positioning it as a system foundry responsible for every supply chain segment [7] - The competitive landscape in the AI market is intense, with companies like Broadcom continuously evolving, posing challenges for Intel to capitalize on this opportunity [7]

Core Viewpoint - Intel's AI strategy is shifting focus towards ASIC and edge AI, aiming to regain competitiveness in the AI sector where it lags behind Nvidia and AMD [2][6]. Group 1: AI Strategy and Market Position - Intel has historically dominated general computing and server-level computing but has struggled in the AI domain, as acknowledged by former CEO Pat Gelsinger [2]. - The company is working on a power-optimized GPU for inference as part of its edge AI strategy, with products like Meteor Lake, Lunar Lake, and the upcoming Panther Lake series [2]. - Intel plans to expand its edge product line with Crescent Island, a processor focused on inference, featuring LPDDR5X memory [2]. Group 2: ASIC Business Development - Intel has established a dedicated ASIC department under Srini Iyengar, aiming to provide customized chips for specific workloads, similar to Google's TPU and Amazon's Trainium [3][5]. - The ASIC business is expected to play a crucial role in Intel's operations, potentially opening new revenue streams and addressing past missteps in the AI market [5][6]. - The company aims to offer a "one-stop" solution for customers seeking custom AI chips, leveraging its chip technology expertise and internal foundry services [6]. Group 3: Competitive Landscape and Future Prospects - Intel's ASIC business is positioned to compete with companies like Broadcom and Marvell, focusing on custom network ASIC chips for network-intensive workloads [4][6]. - The centralized engineering group (CEG) is expected to reduce costs by integrating design services with manufacturing and packaging [6]. - If executed effectively, the custom chip business could become a significant revenue source for Intel, enhancing its position in the AI supply chain [7].

Core Viewpoint - The market is increasingly focused on AI underlying computing power, with Baidu Group announcing plans to evaluate the spin-off and independent listing of its subsidiary Kunlun Chip, which could potentially be submitted for IPO in early 2026 and completed by early 2027 [1][2]. Group 1: Company Overview - Kunlun Chip, originally part of Baidu's AI chip and architecture department, was established as an independent entity in 2021 and has completed multiple rounds of financing, with a valuation of approximately 13 billion yuan in its first round [2][3]. - The company has developed its first-generation AI chip, Kunlun 1, which was launched in July 2018 and achieved mass production in 2020, utilizing Baidu's self-developed XPU architecture [2][3]. Group 2: Market Dynamics - The demand for AI chips in China is surging as the industry is in a rapid growth phase, creating a favorable window for domestic AI chip manufacturers [3]. - Baidu's decision to consider a spin-off aligns with the current industry cycle and Kunlun Chip's development stage, as it has begun external sales and received positive feedback from domestic applications [3][4]. Group 3: Competitive Landscape - The competition in the domestic AI chip market is evolving from a focus on technology to engineering capabilities, ecosystem development, and scalability [4][5]. - Major internet companies in China, including Baidu, Alibaba, and Tencent, are increasingly investing in chip development to reduce costs and ensure supply chain security, with each adopting different strategies for chip development [6][7]. Group 4: Strategic Implications - The spin-off of Kunlun Chip is seen as a strategic move to shed its identity as a Baidu subsidiary, allowing it to compete more effectively in the market [7]. - The trend of large companies developing their own AI chips is common globally, driven by the need to enhance efficiency and meet internal demands, but it also presents challenges in terms of market competition and collaboration [7].

Nvidia graphics processing units like these latest Blackwell [music] GPUs are inside server racks all over the world. Nvidia has catapulted [music] from gaming giant to the very core of generative AI, training the models, running the workloads, and sending Nvidia's valuation soaring. [music] With 6 million Blackwell GPUs shipped over the last year, >> this [music] connects all 72 GPUs, allowing to act as a single GPU to power the most advanced AI workloads.[music] GPUs are the generalpurpose workhorse stars ...

Core Viewpoint - The AI chip market is experiencing significant shifts, with Google accelerating the commercialization of its self-developed AI chip, TPU, which may disrupt the dominance of NVIDIA's GPUs in the computing power market [1][3]. Group 1: Google's TPU Development - Google has been developing TPU since 2013, primarily for internal AI workloads and Google Cloud services, but is now pushing for external commercialization, with potential contracts worth billions [3]. - Meta is considering deploying Google's TPU in its data centers starting in 2027, with the possibility of renting TPU capacity through Google Cloud as early as next year [3]. - Google's TPU strategy aligns with its long-term "soft-hard integration" approach, aiming to reduce energy consumption and control costs amid rising training costs for large models [3]. Group 2: NVIDIA's Market Position - NVIDIA currently holds over 90% of the AI chip market share and emphasizes its "one generation ahead" and "all-scenario advantages" in response to competition from Google's TPU [3][4]. - Despite the potential entry of TPU into large-scale data centers, NVIDIA maintains that GPUs will not be replaced in the short term, as both TPU and NVIDIA GPUs are experiencing growing demand [1][4]. Group 3: Industry Trends - The industry is moving towards a heterogeneous deployment of ASICs and GPUs, rather than a single architecture dominating the market [2][5]. - Major tech companies, including AWS and Microsoft, are also developing their own AI chips, indicating a broader trend of companies seeking to control their computing power [5][6]. - The collaboration between Anthropic and both NVIDIA and Google highlights a shift towards a diversified supply chain for AI computing power, as companies are reluctant to rely solely on one chip architecture [6]. Group 4: Market Reactions - Following news of Google's TPU commercialization, NVIDIA's stock experienced significant fluctuations, reflecting market reassessment of GPU's future share and profitability in AI infrastructure [7]. - The AI infrastructure industry is transitioning from hardware competition to system-level competition, influenced by changes in software frameworks, model systems, and energy efficiency [7].

Core Insights - Major international companies are investing heavily in AI self-developed chip markets, creating new business opportunities. MediaTek is leveraging its years of R&D strength to enter the ASIC design service market, targeting high-end orders and expanding into the AI sector within cloud data centers [1][2]. Group 1: Market Potential and Growth - MediaTek has revised its total addressable market (TAM) for data center ASICs from $40 billion to $50 billion, driven by increased capital expenditures from cloud service providers [2]. - The company aims to capture a market share of approximately 10% to 15% within the next two years, with expectations of stable growth even if its market share remains constant [2]. - The first ASIC project is expected to contribute several billion dollars in revenue starting in 2027, with a second project anticipated to begin generating revenue in 2028 [2][6]. Group 2: Technological Advancements - MediaTek is actively investing in high-speed interconnects and silicon photonics, focusing on chip-to-chip and chip-to-rack connectivity, while also advancing 2nm process technology and 3.5D packaging [3]. - The company emphasizes its long-term technological foundation and R&D investments as key advantages in the ASIC field, enhancing its capabilities in data center technology and communication with local customers [2][6]. Group 3: Competitive Landscape - The AI ASIC market is projected to grow from $12 billion in 2024 to $30 billion by 2027, with a compound annual growth rate (CAGR) of 34% [5]. - Major tech giants, including Google, Tesla, Amazon, Microsoft, and Meta, are all investing in ASIC chip development, indicating a competitive and rapidly evolving market [5]. - MediaTek's collaboration with Google to develop the next-generation TPU, expected to be produced by 2026, highlights the strategic partnerships forming within the industry [6].