Broadcom partnerships with Google and with Anthropic. It gave more details in a filing about these long-term agreements to develop Google's TPUs, their tensor processing units, and they're also going to be working with Anthropic on these types of chips. Most analysts pretty positive on this news.VC Arya for example over at Bank of America said that these announcements significantly increase Broadcom's visibility is a longerterm main design partner for TPUs and also remove some stock overhang because Alphabe ...

Welcome back to the exchange. In a deal that could reshape the AI market, shares of Broadcom are moving higher on a deal to develop and supply Google's in-house chips for the next 5 years. Mackenzie Sagalos brings us more in today's tech check.Hi, Mackenzie. Hey, Kelly. So, Broadcom may be getting the bounce today, up around 5% right now as it becomes this new arms dealer for this AI custom chip era, but Google is the one turning that demand into a full stack mode.now monetizing nearly every layer of its AI ...

Anthropic ups compute deal with Google and Broadcom amid skyrocketing demand https://t.co/bZUXyqJnje ...

Broadcom confirmed plans to deliver chips to AI startup Anthropic using Google’s tensor processing units, or TPUs, offering an alternative to technology developed by Nvidia https://t.co/Q5phNMREJR ...

We've signed an agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity, coming online starting in 2027, to train and serve frontier Claude models. ...

Broadcom, one of the most valuable publicly traded semiconductor companies, named Alphabet executive Amie Thuener as its next CFO https://t.co/1YkjMNBvqV ...

Summary of Key Points from the AI Infrastructure Value Chain Conference Call Industry Overview - The discussion centers around the **Artificial Intelligence (AI)** industry, particularly focusing on the **AI infrastructure value chain** and its implications for investment opportunities and risks [1][4][15]. Core Insights and Arguments - The **AI boom versus bubble debate** has been ongoing since 2023, with potential outcomes ranging from significant advancements in AI to a major market crash [1][4]. - AI spending is expected to remain robust in the near future, driven by the dominance of AI believers in decision-making roles at key technology firms [6][27]. - The estimated **AI data center capital expenditure (capex)** is projected at **$36 billion per gigawatt (GW)**, with GPUs accounting for **38%** of total costs and networking comprising approximately **12%** [2][7][35]. - The **total AI infrastructure spending** could exceed **$400 billion** by 2025, contributing significantly to global GDP growth [15]. Potential Winners and Losers - Companies such as **Ibiden**, **Unimicron**, and other PCB and substrate manufacturers are identified as having high upside potential due to their roles in the AI supply chain [3][51]. - In contrast, **Intel**, **Cisco**, and server OEMs like **Dell** and **Hewlett Packard Enterprise** are seen as having lower upside potential relative to their prominence in the AI discussion [3][52]. - The analysis suggests that **NVIDIA** and **Broadcom** remain industry favorites, while companies like **AMD** and **Mediatek** also present significant upside opportunities [9][51]. Important but Overlooked Content - The **US-China AI race** is highlighted, indicating that the US is increasing its compute capacity while China lags behind, which adds a geopolitical dimension to the AI infrastructure landscape [10]. - The **depreciation debate** surrounding GPUs is addressed, emphasizing the need for a reasonable depreciation timeline to avoid overestimating asset values [16]. - The potential for a **digestion cycle** is acknowledged, where overcapacity could lead to reduced investments until demand aligns with supply [16][26]. Financial Projections - The **enterprise AI total addressable market (TAM)** is estimated to range from **$600 billion** in a bear case to **$11 trillion** in a bull case, depending on productivity gains from generative AI and the scaling of model capabilities [26]. - Aggregate revisions from May 2023 to February 2026 indicate a **9% increase in revenue**, **26% in EBIT**, and **22% in free cash flow** for a basket of AI stocks [15][21]. Conclusion - The AI infrastructure landscape presents both significant opportunities and risks, with a focus on identifying which companies are best positioned to benefit from ongoing AI investments. The analysis emphasizes the importance of understanding the underlying dynamics of AI spending and the potential for both growth and contraction in the sector [28][27].

Core Insights - Nvidia and Broadcom are leading investments in the artificial intelligence (AI) sector, with Nvidia being the preferred buy for April due to its growth and valuation advantages [1][10]. Nvidia - Nvidia has established itself as the primary provider of AI computing units since the AI expansion began in 2023, with its GPUs being the industry standard for accelerated computing [3]. - The upcoming Vera Rubin chip architecture is expected to significantly enhance performance, requiring four times fewer chips for training and ten times fewer for inference compared to the current Blackwell generation [3][4]. - Nvidia anticipates lifetime sales of Blackwell and Rubin chips to reach $1 trillion by the end of 2027, a substantial increase from the previous expectation of $500 billion by 2026 [4]. - Nvidia's stock is currently trading at $174.36, with a market cap of $4.2 trillion and a gross margin of 71.07% [5]. Broadcom - Broadcom approaches the AI market differently by partnering with AI hyperscalers to create custom AI chips, which are optimized for specific tasks and can outperform Nvidia's GPUs in those areas [6]. - Broadcom expects its custom AI chips to generate $100 billion in revenue by the end of 2027, with its relevant division growing at a rate of 106% to $8.4 billion in the last quarter [7]. - Broadcom's stock is currently priced at $309.29, with a market cap of $1.5 trillion and a gross margin of 64.96% [8]. Market Outlook - The AI build-out is projected to continue through at least 2030, with global data center capital expenditures expected to rise to $3 trillion to $4 trillion annually by the end of that year, providing a significant growth opportunity for both companies [9]. - Nvidia's growth rate has outpaced Broadcom's in the last quarter, and it is considered cheaper from a forward price-to-earnings perspective [10][12].

Group 1: Company Performance - Sunny Optical (02382) reported an annual revenue of approximately 43.23 billion RMB, a year-on-year increase of 12.9%, and a net profit attributable to shareholders that grew by 71.9%, with a final dividend of 1.206 HKD per share, totaling around 1.301 billion HKD, achieving a dividend yield of about 25%, a recent high [1] - TCL Electronics (01070) saw a rise of over 12% after announcing a framework agreement with Sony to establish a joint venture, acquiring 51% of the new company and fully purchasing Sony's subsidiary in Malaysia for 75.399 billion JPY (approximately 3.781 billion HKD) [1] - Cambridge Technology (06166) reported a revenue of approximately 4.8234 billion RMB for 2025, a year-on-year increase of 32.07%, and a net profit of about 263 million RMB, up 58.08%, with a proposed cash dividend of 0.28 RMB per share [3] - HaiXi New Drug (02637) reported a revenue of approximately 582 million RMB for 2025, a year-on-year increase of 24.79%, and a net profit of about 177 million RMB, up 30.09%, with earnings per share of 2.55 RMB [4] Group 2: Market Trends - Aluminum stocks continued to rise, with China Aluminum (02600) up 4.09%, Nanshan Aluminum International (02610) up 3.83%, and China Hongqiao (01378) up 3.22%, following substantial production cuts from two Middle Eastern aluminum companies, which announced a 20% reduction in output by 2026 [2] - Semiconductor stocks rebounded, with companies like Lanqi Technology (06809) rising by 8.12% and Zhaoyi Innovation (03986) by 5.31%, driven by a strong performance in the South Korean market where Samsung Electronics and SK Hynix saw significant gains [2] - The aviation sector experienced a collective rebound, with China National Aviation (00753) up 6.13% and China Eastern Airlines (00670) up 6.88%, influenced by geopolitical developments in Iran [4] Group 3: Strategic Partnerships - Elysium AI (03696) surged over 10% following a significant partnership with Eli Lilly, granting Eli Lilly exclusive sales rights to a GLP-1 diabetes drug developed using Elysium's AI technology, with an upfront payment of 115 million USD and a total potential value of up to 2.75 billion USD [3]

Market Overview - The tech market sentiment has shifted from bullish to cautious, with increased investor scrutiny on AI capital expenditures and the overall market dynamics influenced by geopolitical factors [4][5][6] - The ongoing war has intensified the tug of war between greed and fear in the market, leading to volatility and a lack of confidence in sustained relief rallies [10][12][16] Geopolitical Factors - The conflict in the Middle East has significant implications for the tech sector, particularly for companies like SK Hynix, Samsung, and TSMC, which are heavily reliant on helium from Qatar for semiconductor manufacturing [19][20] - Recent developments suggest a potential resolution to the conflict, which could alleviate supply chain pressures and positively impact semiconductor companies amid rising demand for chips [24][27] Semiconductor Industry Insights - Micron has reported strong financial performance, but concerns exist regarding the sustainability of its stock rally, as the current price increases are driven by non-AI demand rather than AI-related growth [31][34][36] - The memory market is experiencing significant price fluctuations, with DRAM prices surging over 300% for certain kits, but the outlook for demand in PCs and smartphones has been revised downward, indicating potential challenges ahead [34][40][41] AI and Memory Usage - Google's TurboQuant technology has raised concerns about reduced memory requirements for AI applications, leading to a market reaction that negatively impacted memory and storage stocks [49][50][51] - The evolving nature of AI suggests that while memory efficiency may improve, the overall demand for memory is unlikely to decrease significantly, as the industry continues to scale up for AI applications [52][54] Energy Sector Considerations - The energy requirements for AI infrastructure are becoming a critical issue, with over 500 planned data center projects expected to impact electricity prices and consumption [57][58] - The current geopolitical tensions, particularly related to Iran, exacerbate concerns about energy supply and its implications for tech companies reliant on stable energy sources [60] US-China Relations and Chip Supply - The Super Micro scandal highlights the widening gap between US and Chinese tech capabilities, with increased scrutiny on AI chip exports to China [62][64] - The ongoing geopolitical tensions may lead to more stringent export restrictions from the US, while also opening avenues for negotiations between US and Chinese leaders regarding technology and trade [68][70]