Core Insights - The core contradiction in AI infrastructure construction is shifting from merely pursuing GPU quantities to competing for power supply speed, with "Time-to-Power" becoming the most severe bottleneck in AI development [1] - Chinese power solution providers with rapid delivery capabilities and large-scale production advantages are experiencing a historic revaluation opportunity [1] Group 1: Power Supply Challenges - The average lifespan of power grids in the US and EU has reached 35 to 40 years, and the infrastructure is increasingly fragile due to the explosive energy demands of AI data centers (AIDC) [1] - Current domestic power equipment capacity in the US can only meet about 40% of local demand, with interconnection waiting times extending to nearly five years [1] - Goldman Sachs predicts that the power supply shortage in the US will persist until 2030, with a projected 175% increase in electricity consumption by data centers by 2030 [2][3] Group 2: Market Dynamics and Growth - The overall addressable market for AI data center power products is expected to expand at a compound annual growth rate (CAGR) of approximately 39% from 2025 to 2030 [3] - The growth is driven by continuous capacity construction, increasing power density, and a shift from AC to DC architecture, with 800V DC distribution becoming the standard for many AI data center projects [4] Group 3: Competitive Advantages of Chinese Suppliers - Chinese suppliers are gaining a decisive competitive advantage not only through lower costs but also through shorter delivery cycles, which have become a primary decision factor for data center operators [5] - Companies like Siyi Electric and Yinglite are positioned to benefit from the supply shortages in the US market, with Siyi Electric's revenue from the US expected to grow from 26% in 2026 to 28% in 2028 [5][6] Group 4: Product Prioritization - Goldman Sachs has provided a clear preference ranking for Chinese power supply-related product categories: gas turbine blades > power transformers > electrical components > uninterruptible power supplies/power racks > liquid cooling systems > server power [6][7] - Gas turbine blades rank highest due to high material science and manufacturing barriers, while power transformers follow due to labor-intensive manufacturing and lengthy certification processes [7]

Core Insights - The essence of the AI competition is fundamentally a race for physical infrastructure, particularly data centers, which are expanding at an unprecedented pace, projected to exceed $400 billion in capital expenditure by 2024 and reach $506 billion by 2025, driven by AI demand [1][2] - The data center market is expected to grow at a remarkable compound annual growth rate (CAGR) of 23% from 2024 to 2028, ultimately forming a market exceeding $900 billion by 2028 [1][2] Market Overview - The data center market's growth is no longer driven by traditional enterprises; since 2017, cloud service providers and colocation companies have contributed nearly all new capacity, with major players like Amazon AWS and Microsoft Azure leading the way [5][7] - North Virginia, USA, accounts for nearly 15% of the global hyperscale data center capacity, making it the largest single concentration globally, followed by Beijing, China, at approximately 7% [9] Investment Dynamics - Data centers are viewed as high-value infrastructure assets with clear and attractive return models, exemplified by a typical wholesale colocation project that can yield an internal rate of return (IRR) of 11% over a 20-year holding period [12][13] - The average annual rent per megawatt (MW) is projected to increase from $2.50 in 2022 to $4.38 by 2028, with occupancy rates remaining stable at around 90% [14] Technological Evolution - The "density revolution" in data centers is driven by the exponential increase in power consumption of AI chips, leading to a significant rise in rack power density, with current standards reaching up to 120 kW per rack [15][17] - The global infrastructure for data centers is lagging, with only 5% of existing centers capable of supporting the latest high-power AI chips, necessitating massive upgrades and new constructions [18][19] Infrastructure Transformation - The shift from air cooling to liquid cooling systems is becoming essential due to the limitations of traditional cooling methods, with the market for cooling solutions expected to grow significantly [21][23] - The transition to high-voltage direct current (DC) power systems is underway, promising to reduce copper usage and improve efficiency in power delivery to data centers [25][28] Supply Chain Analysis - The rise of AI is increasing the unit construction costs of data centers, with next-generation AI architectures projected to cost approximately $5.2 million per MW, a 33% increase from traditional setups [30] - Key players in the thermal systems market include Vertiv, Johnson Controls, and Carrier, while Schneider Electric leads in the electrical systems market [32][36] - The IT equipment segment is the largest component of data center investments, with the global server market expected to reach $280 billion in 2024, driven significantly by AI servers [39] Conclusion - The ongoing AI-driven construction boom in data centers is set to surpass the total construction spending of all general office buildings in the coming months, highlighting the critical role of infrastructure in the AI landscape [43][46]