Riverwater Partners, an investment management company, released its “Small Cap Strategy” Q3 2025 investor letter. A copy of the letter can be downloaded here. The Small Cap Core Strategy reported solid results for the quarter but did not keep pace with the sharp rally, thus lagging the benchmark. The market’s preference for lower-quality companies persisted in the quarter and contributed significantly to the benchmark’s outperformance. On the other hand, the strategy remains focused on higher-quality compa ...

Core Insights - nVent Electric is experiencing strong demand from data center customers, primarily driven by growth in artificial intelligence (AI) workloads, with organic orders increasing by approximately 65% in Q3 2025 [1][11] - The company has visibility into 2026, with some programs extending into 2027, indicating firm customer spending plans and a solid backlog that grew in double digits sequentially in Q3 [2] Company Performance - Liquid cooling is a key growth driver, with less than 10% of data centers currently utilizing this technology, but the demand is expected to rise due to new graphics processing unit chips requiring advanced cooling solutions [3] - nVent Electric has over a decade of experience in liquid cooling and has deployed more than 1 gigawatt of cooling capacity [3] - A new facility in Minnesota is set to double the company's liquid cooling production capacity by early 2026 [4] - nVent Electric has joined NVIDIA's partner network, enhancing its visibility among global AI infrastructure customers [4][11] Market Trends - The overall trend in AI infrastructure and liquid cooling demand appears robust, with nVent Electric's backlog and customer pipeline suggesting continued growth in data center demand [5] - The Zacks Consensus Estimate for nVent Electric's total revenues for the full year 2025 is projected at $3.83 billion, reflecting a year-over-year increase of 11.1% [5] Competitive Landscape - nVent Electric competes with companies like Vertiv and Hubbell in the electrical and data center markets [6] - Vertiv is advancing large-scale power architectures for AI workloads and is expected to release new power systems in the second half of 2026 [7] - Hubbell reported high single-digit organic growth in its Electrical Solutions segment, driven by strong demand from data center projects [8] Financial Metrics - nVent Electric's shares have increased by 57.2% year to date, outperforming the Zacks Electronics - Miscellaneous Components industry's growth of 50.7% [9] - The company trades at a forward price-to-sales ratio of 3.99X, which is higher than the industry average of 2.31X [12] - The Zacks Consensus Estimate for nVent Electric's earnings per share (EPS) implies year-over-year growth of 33.7% for 2025 and 19.5% for 2026, with upward revisions in EPS estimates over the past 60 days [13]

Summary of Key Points from the Conference Call Industry Overview - The conference call focused on the **liquid cooling technology** within the **data center** industry, particularly the **GB300 cooling system architecture** and its implications for suppliers, especially in China [1][3]. Core Insights and Arguments 1. **GB300 System Architecture**: - The GB300 servers did not utilize the **Cordelia motherboard architecture** due to design challenges, opting instead for a **Bianca-style "T-shaped" layout** with improved liquid-to-air ratios [2][3]. - The current design retains eight fans, indicating it is not fully liquid-cooled [2]. 2. **Product Innovation**: - Liquid cooling is transitioning from a niche to a mainstream solution due to increasing rack power densities and sustainability goals [3]. - Innovations include: - **Electrical System Liquid Cooling**: This approach addresses thermal and electrical challenges as rack power levels exceed 300 kW, with OCP specifications now including liquid-cooled busbar standards [4]. - **Two-Phase Liquid Cooling**: This method enhances thermal performance but presents engineering challenges such as pressure stability and leak detection [4]. 3. **Coolant Distribution Units (CDUs)**: - CDUs are critical for liquid cooling infrastructure, with three dominant architectures emerging [5][9]. 4. **Customer Landscape**: - Google's **Open Rack initiative** promotes interoperability among suppliers, with several companies showcasing prototypes at the OCP summit [6]. - Major integrators prefer local brands for critical components, indicating a trend towards localized supply chains [6]. 5. **Opportunities for Chinese Suppliers**: - Chinese vendors, such as **Envicool**, are positioned to benefit from these trends, with advancements in localized service centers and compliance with Open Rack standards [7]. - Emerging technologies like micro-channel cold plates present additional opportunities for differentiation [7]. 6. **Investment Recommendations**: - **Shenzhen Envicool Technology** is valued at a price target of **Rmb85.64** based on a **60x 2026E PE**, with risks including fluctuations in global AI Capex and competition [10]. - **Shenzhen Kstar Science and Technology** is valued at **Rmb64.86**, implying a **40x 2026E P/E**, with risks related to domestic AI Capex and competition [11]. Additional Important Insights - The evolution of liquid cooling products has been slower than anticipated, with many legacy designs still prevalent, limiting opportunities for local manufacturers [8]. - The integration of advanced leak detection technologies in cooling systems enhances safety and operational efficiency [2][4]. - The report emphasizes the importance of robust safety measures and advanced sealing technologies for the successful deployment of two-phase cooling systems [4]. This summary encapsulates the critical insights from the conference call, highlighting the evolving landscape of liquid cooling technology and the strategic positioning of companies within this sector.

Core Insights - Marvell Technology is a mid-tier player in the digital infrastructure industry with a market cap of $72 billion and revenue of $5.77 billion in fiscal year 2025 [3] - The company specializes in high-quality custom XPUs, CPUs, and DPUs, essential for AI and cloud-computing data centers, along with Ethernet and data center switches [2][3] - Marvell's recent quarterly results for fiscal 2Q26 showed revenues of $2 billion, a 57.5% year-over-year increase, although it slightly missed forecasts by $10 million [8] Company Overview - Marvell produces network and connectivity solutions that support enterprise mobility and cloud computing, promoting a 'borderless campus' concept [1] - The company provides hardware for connecting 5G radio access networks to the cloud, integrating networking and computing technologies [7] - Marvell's solutions are critical for building the infrastructure necessary for AI advancements [7] Financial Performance - In fiscal 2Q26, Marvell reported a non-GAAP EPS of 67 cents, aligning with forecasts [8] - The company is expected to benefit from positive trends in AI and cloud computing, with analysts expressing confidence in its growth prospects for 2026 [9] Analyst Insights - Analysts from JPMorgan have a bullish outlook on Marvell, citing strong commentary from major clients like Amazon and positive earnings guidance from related sectors [9] - The stock has an Overweight rating with a price target of $120, suggesting a potential 44% upside over the next year [9] - The consensus rating for Marvell is a Moderate Buy, based on 32 reviews, with an average price target of $93.23 indicating a 12% upside from current levels [10]

Core Insights - The article discusses the intertwined ecosystem of key players in the AI infrastructure space, including NVIDIA, OpenAI, AMD, and Oracle, highlighting their roles and relationships in the current AI revolution [1][2]. Group 1: Company Roles - OpenAI serves as the AI model developer and central "demand anchor," requiring substantial compute power, data centers, and scale [3]. - NVIDIA provides GPUs and accelerators to OpenAI, invests in OpenAI-linked infrastructure, and benefits from guaranteed demand [3][4]. - AMD also supplies GPUs with large-scale deployment commitments, similar to NVIDIA [7]. - Oracle builds data center infrastructure and cloud hosting for OpenAI, including "Stargate" facilities and power contracts [7]. Group 2: Economic Implications - OpenAI's AI models drive significant demand for high-performance GPUs, creating a symbiotic relationship with NVIDIA and AMD [7]. - NVIDIA's potential $100 billion investment in OpenAI highlights a strategic focus on long-term demand rather than immediate revenue [7]. - Oracle's $300 billion commitment to data center construction and AI hosting indicates a comprehensive infrastructure approach beyond just GPU supply [7]. Group 3: Market Dynamics - The current AI landscape shows speculative traits, with NVIDIA's forward P/E ratio exceeding 40x, suggesting high investor expectations [9]. - The concentration of capital among a few firms creates a feedback loop that reinforces investment and valuation [9]. - Unlike the dot-com era, the AI buildout involves tangible infrastructure, such as chips and data centers, which retain value even during downturns [10]. Group 4: Risks and Signals - Potential risks include order cancellations, delayed deployments, and regulatory scrutiny, which could impact market confidence and valuations [6][13]. - A 50% order cancellation from OpenAI could lead to a 4-5% short-term earnings hit for NVIDIA and 2-3% for AMD, indicating market sensitivity to earnings misses [13]. - The narrative momentum surrounding AI could drive market performance, but a weakening narrative may lead to valuation corrections [10]. Group 5: Cooling Solutions in AI Infrastructure - Companies specializing in advanced cooling technologies for data centers are positioned to benefit from the expanding AI infrastructure [11]. - Vertiv Holdings Co provides thermal management solutions and has recently acquired Great Lakes Data Racks & Cabinets for $200 million to enhance its capabilities [12][14]. - Schneider Electric has bolstered its data center cooling portfolio by acquiring a 75% stake in Motivair Corp for $850 million [17].

Core Insights - The article highlights the importance of lesser-known companies in the AI ecosystem, which are crucial for the foundational growth of AI technology [1][2][18] - These companies, often referred to as "hidden gems," provide essential infrastructure and services that support the AI revolution, contrasting with the more visible chip designers and software giants [1][2][18] Company Summaries Camtek - Camtek specializes in inspection and metrology equipment essential for semiconductor manufacturing, ensuring AI chips function correctly [5][7] - The company reported a 22% year-over-year revenue increase to $118.6 million and a 38% rise in net income to $34.3 million for Q1 2025 [6] - Management projects Q2 revenue between $120 million and $123 million, indicating a growth rate of 17% to 20% [6] - Camtek's market cap is $3.8 billion, and it is recognized as a key player in the advanced packaging market for AI applications [8] FormFactor - FormFactor addresses the challenge of testing AI chips with thousands of connections at high speeds, using advanced wafer probe cards [9][12] - The company reported Q1 2025 revenue of $171.4 million, with Q2 guidance set at $190 million, despite facing export restrictions [10] - FormFactor's acquisition of FICT Limited enhances its capabilities in advanced probe card components, and a $75 million share buyback reflects management's confidence [11] - The company is well-positioned to meet the increasing demands for sophisticated testing as AI chip complexity rises [12] Vertiv Holdings - Vertiv focuses on cooling solutions for AI data centers, which are critical for managing the heat generated by large-scale AI computations [13][16] - The company achieved Q1 2025 net sales of $2 billion, a 24% increase year-over-year, with a backlog 25% higher than the previous year [14] - Vertiv's collaboration with Nvidia and iGenius to deploy a large AI supercomputer in Italy showcases its role in AI infrastructure [15] - Full-year 2025 sales guidance has been raised to between $9.3 billion and $9.6 billion, reflecting the growing demand for thermal management solutions in AI [16] Industry Overview - The article emphasizes the interconnected roles of Camtek, FormFactor, and Vertiv in the AI ecosystem, where each company plays a vital part in the manufacturing, testing, and operational phases of AI chip deployment [17] - As AI adoption accelerates, the demand for inspection equipment, testing technology, and cooling infrastructure is expected to intensify, highlighting the critical nature of these "pick-and-shovel" companies [18]