Counterpoint Research says the race to build custom AI silicon is accelerating, with hyperscalers scaling internal chips to meet surging demand. • Broadcom stock is showing upward movement. What’s driving AVGO shares up?Hyperscalers Ramp Custom AI ChipsLeading cloud and AI providers, including Alphabet Inc.’s (NASDAQ:GOOGL) Google; Amazon.com, Inc.’s (NASDAQ:AMZN) Amazon Web Services; Microsoft Corp (NASDAQ:MSFT) ; OpenAI; ByteDance and Apple Inc (NASDAQ:AAPL) , are rapidly expanding deployments of AI serve ...

Key Points Broadcom expects strong growth from its custom AI accelerators. Custom AI accelerator revenue is expected to double in Q1. 10 stocks we like better than Broadcom › Few stocks had as strong a 2025 as Broadcom (NASDAQ: AVGO). Broadcom's stock caught fire following the tariff announcement recovery, and finished the year up 49.3%. While 2025 was great, investors need to look forward and see if it's worth buying more shares now or if there is another compelling investment opportunity. I thi ...

Core Viewpoint - A fresh analysis of global semiconductor supply chains indicates that hyperscaler demand is increasing, leading to tighter memory supply and reshaping the competitive landscape in the industry. Group 1: Hyperscaler Demand and Memory Pricing - Hyperscalers are securing Dynamic Random-Access Memory (DRAM) and NAND capacity in anticipation of a 50% growth in data center bits by 2026, resulting in higher contract prices [2] - DRAM contract prices are expected to rise by approximately 25% in Q1 2026 and 10%-12% in Q2 2026, while NAND prices are projected to increase by about 20% in Q1 2026 and 10%-15% in Q2 2026 [2] Group 2: Company Performance and Upgrades - Intel and AMD are highlighted as leading companies, with both upgraded to Overweight due to being largely sold out of 2026 server CPU capacity, with potential price increases of 10%-15% in Q1 2026 [4] - Micron Technology's price forecast has been raised to $450 from $325, reflecting positive views on AI compute and infrastructure demand [5] Group 3: Custom Silicon and Market Dynamics - Custom silicon is identified as a key battleground for hyperscalers, with Broadcom's Chip on Wafer on Substrate (CoWoS) supply for 2026 revised upward by 30% to 250K units, a significant increase compared to 2025 [6] - OpenAI's custom ASIC timeline has shifted to Q1 2027, with an expected lifetime unit opportunity of 1.5 million to 2 million, potentially adding $8 billion to $10 billion to Broadcom's AI backlog [7] Group 4: Challenges and Market Outlook - Higher memory prices and shortages are negatively impacting handset demand and margins, leading to lowered estimates for Qualcomm and highlighting Arm Holdings as a potential headwind due to anticipated smartphone market contraction in 2026 [9] - Memory constraints and price increases may affect PCs, smartphones, and automotive sectors, despite strong demand in AI and data center markets [9]

Core Viewpoint - Nvidia has experienced significant growth in recent years, particularly in the AI sector, and continues to show strong potential for future performance as it heads into 2026 [1][2]. Group 1: Market Position and Competitive Advantage - Nvidia holds a dominant position in the data center GPU market with a 92% market share, making it difficult for competitors to catch up [5]. - The company has established a competitive moat through its Compute Unified Device Architecture (CUDA), which is widely recognized as the gold standard for GPU programming and is utilized by over 4.5 million developers [10][11]. - CEO Jensen Huang has strategically positioned Nvidia to capitalize on the AI adoption wave since 2013, contributing to its current strength [4]. Group 2: Competition Landscape - Competitors such as Amazon, Alphabet's Google, and Advanced Micro Devices (AMD) are developing their own AI processors, with notable advancements in performance and efficiency [7][8][9]. - Amazon's Trainium 3 chip claims to be four times faster and 40% more energy-efficient than its predecessor, while Google's latest TPU offers peak performance improvements [7][8]. - AMD's MI350 processors provide four times better performance than previous versions, with upcoming MI400 series expected to be highly competitive [9]. Group 3: Valuation and Growth Prospects - Nvidia's stock is currently trading at 46 times earnings, which some investors consider high, but the company's growth trajectory justifies this premium [12][13]. - Over the past three years, Nvidia's stock has risen 1,200%, indicating strong growth that supports a higher valuation [13]. - For fiscal 2026, Nvidia is projected to increase revenue by 63% and earnings per share (EPS) by 59%, with expectations of 50% revenue growth in fiscal 2027 [14].

Key Points Alphabet has a lot to gain by developing its quantum technology in-house. The tech giant is already seeing real-world applications for its technology. 10 stocks we like better than Alphabet › Quantum computing is a difficult area to invest in. It's similar to investing in early-stage biotechs, where there are some promising results, but nothing indicative of true future success. Couple that with how many companies are competing for quantum computing dominance, and it turns into a very di ...

Core Insights - Nvidia has entered a non-exclusive licensing agreement with Groq for its inference technology, which allows Nvidia to eliminate a potential competitor while acquiring new chip technology [1][4] - The deal includes key personnel from Groq, indicating an "acqui-hire" strategy that closely resembles a full acquisition [4] - Groq will continue to operate under its CFO, but significant advancements in its technology will now be made under Nvidia [5] Deal Size and Valuation - The deal size is reported to be around $20 billion, making it Nvidia's largest deal to date, surpassing its previous $6.9 billion acquisition of Mellanox Technologies [7][8] - Groq's valuation was $6.9 billion following a $750 million financing round in September, indicating that the deal represents a substantial premium [8] Market Context - Nvidia dominates the AI chip market, particularly in AI training and inference, but faces increasing competition from companies like AMD and custom ASICs from Broadcom and Marvell [10] - Major tech companies are exploring alternatives to Nvidia's GPUs, such as Meta considering Google's TPU, to reduce costs and diversify supply chains [11] Groq's Technology - Groq specializes in language processing units (LPUs) designed for AI inferencing, which is the second step in the AI process following training [9] - Groq claims its technology is faster for specific inference applications and aims to sell its chips at lower prices than Nvidia's offerings [12][13]

Core Viewpoint - Alphabet and Amazon are identified as strong investment opportunities, particularly in light of their potential growth in 2026, despite differing performances in 2025 [2][6]. Group 1: Company Performance - Alphabet's stock increased by approximately 60% in 2025, overcoming initial doubts regarding its capabilities in artificial intelligence and concerns about potential monopoly issues [6][7]. - Amazon's revenue grew by 13% in the third quarter, with its cloud computing unit, Amazon Web Services (AWS), accounting for 66% of operating income despite only representing 18% of total revenue [10][11]. Group 2: Investment Insights - Berkshire Hathaway's portfolio includes Alphabet, which makes up about 1.7% of its total holdings, and Amazon, which constitutes 0.7% [5]. - The performance of Alphabet in 2025 suggests continued dominance in generative AI, potentially leading to new revenue streams in 2026 [8][9]. - Amazon's AWS has shown its fastest growth quarter in several years, indicating a strong rebound potential for the company in 2026 [11].

Core Insights - The artificial intelligence (AI) market is projected to grow from over $270 billion today to more than $5.2 trillion in the next decade, indicating significant investment opportunities in the sector [1]. Company Summaries - **Nvidia**: Dominates the accelerator chip market with a 92% market share in data centers, providing essential compute power for AI. The company has a $500 billion order backlog, showcasing its strong momentum [5][7]. - **Alphabet**: Leverages its extensive ecosystem, including Google apps and YouTube, to enhance its AI capabilities. The company has developed its own AI chip, the Tensor Processing Unit (TPU), and is positioned as a competitor to Nvidia in the AI chip market [8][9]. - **Microsoft**: Operates Azure, the second-largest cloud services platform, and holds a 27% stake in OpenAI, the creator of ChatGPT. The company benefits from a stable software business and has a history of increasing dividends for 23 consecutive years [10][12]. - **Amazon**: Operates Amazon Web Services (AWS), the leading cloud services provider, and has an $8 billion stake in Anthropic, a competitor to OpenAI. The company's diverse business model supports long-term growth, with AI serving as an additional growth driver [13][14]. - **Palantir Technologies**: Focuses on custom software applications and has seen growth since launching its AI platform, AIP. The company has a significant opportunity for customer acquisition, which could drive future growth [15][17].

Merry Christmas and happy holidays! It’s a quiet week at the SGX but we did have a couple of news to share plus a major acqui-hire across the pond by NVIDIA (NASDAQ: NVDA).Volare Group bids for Alpha Integrated REIT Volare Group has entered into a sales and purchase agreement with ESR Group (SEHK: M1GU) to acquire ESR’s 241.6 million units in Alpha Integrated REIT (AIR, SGX: M1GU), representing 21.5% of AIR, at S$0.40 per unit. AIR was formerly called Sabana REIT. Post-acquisition, Volare will control appro ...

For more on what the deal means for the chip space overall and Nvidia in particular, let's bring in Stacy Rascin, the Bernstein Research US semiconductors analyst. Stacy, it's great to have you with us on this day after Christmas. Uh, this is an interesting proposition because Jonathan Ross and his team of engineers were some of the ones behind the Google Alphabet Tensor Processing Unit type products out there.Just how much does this change the landscape within chips overall. Yeah. So, what what they're doi ...