Core Insights - The exponential growth of AI is transforming data centers into power-hungry entities, overwhelming aging grid infrastructure [1][2] - Goldman Sachs predicts that by 2030, "behind-the-meter" (BTM) solutions could meet 25-33% of the incremental power demand from data centers, translating to 20-25 gigawatts (GW) of installed capacity [1][3] Group 1: Power Demand and Infrastructure Challenges - The global power demand from data centers, driven by both AI and non-AI applications, is reshaping the electricity market, with an estimated need for an additional 82 GW of power capacity in the U.S. alone by 2030 [2] - The construction pace of new high-voltage transmission lines in the U.S. has significantly declined, from an average of 1,700 miles per year (2010-2014) to only 350 miles per year (2020-2023) [2] - The median time for a project to go from application to commercial operation has reached nearly 5 years, exacerbating the urgency for data center operators to seek alternative power solutions [2] Group 2: Emergence of Behind-the-Meter Solutions - BTM solutions, which allow for on-site generation and consumption of power without relying on the public grid, are emerging as a critical alternative for data centers facing grid reliability issues [3] - From 2024 to 2030, the total incremental power demand for data centers is projected to be approximately 730 terawatt-hours (TWh), with BTM solutions expected to satisfy 25-33% of this demand [3] Group 3: Fuel Cell Technology Potential - Fuel cells, particularly solid oxide fuel cells (SOFC), are anticipated to capture 25-50% of the BTM solutions market, equating to 8-20 GW of installed capacity by 2030 [4][6] - SOFCs are favored for their structural advantages over traditional gas turbines, including delivery time, noise, emissions, and flexibility, making them well-suited for data center applications [4][6]
美国数据中心增量电力需求，“电网外方案”解决“1/4到1/3”，其中燃料电池满足“25-50%”
Hua Er Jie Jian Wen·2025-11-02 11:45