The rapid growth of artificial intelligence (AI) data centers has caused a global increase in energy and water use, leading to new government regulations and local protests.

人工智慧 (AI) 資料中心的快速增長導致全球能源與水資源使用量增加，引發了新的政府法規與本地抗議。

AI infrastructure requires significantly more resources than traditional data centers. For example, AI GPU clusters use between 80 and 150 kW per rack, whereas standard enterprise racks only need 15 to 20 kW. Consequently, cities like San Francisco, Vancouver, and Visakhapatnam have seen local resistance because the use of drinking water and electricity is seen as a threat to city security. In the United States, these conflicts delayed or stopped about $64 billion in projects between May 2024 and March 2025. Similarly, Chile and Malaysia have taken legal action to prevent environmental damage.

AI 基礎設施所需的資源比傳統資料中心多得多。例如，AI GPU 集群每個機架使用 80 到 150 kW，而標準企業機架僅需要 15 到 20 kW。因此，像舊金山、溫哥華和維沙卡帕特南等城市出現了本地抵制，因為飲用水和電力的使用被視為對城市安全的威脅。在美國，這些衝突在 2024 年 5 月至 2025 年 3 月期間，導致約 640 億美元的項目被延遲或停止。同樣地，智利和馬來西亞已採取法律行動以防止環境破壞。

In India, the sector is moving from a phase of fast growth toward a more unstable period. Experts project that electricity demand for data centers will reach 107 TWh by 2031-32. Furthermore, the United Nations University has noted that India's power grid is one of the most carbon-heavy in the world. To balance industrial growth with nature, analysts emphasize that companies must switch to closed-loop liquid cooling systems and use non-drinking water to avoid competing with farming and home needs.

在印度，該產業正從快速增長階段進入一個較不穩定的時期。專家預計到 2031-32 年，資料中心的電力需求將達到 107 TWh。此外，聯合國大學指出印度的電網是全球碳排放最沉重的電網之一。為了平衡工業增長與自然，分析師強調公司必須轉向封閉式液冷系統，並使用非飲用水，以避免與農業和家居需求競爭。

Meanwhile, the United States shows a contradictory relationship between AI and clean energy. While the high demand has increased investments in renewable energy and battery storage, it has also encouraged the use of old fossil-fuel plants to ensure a steady power supply. Large tech companies are now buying energy firms to secure their own power. Additionally, the rise of fuel-cell technology shows that some companies prioritize keeping their operations running over completely removing carbon emissions.

與此同時，美國在 AI 與清潔能源之間展現出一種矛盾關係。雖然高需求增加了對再生能源和電池儲能的投資，但同時也鼓勵使用舊的化石燃料電廠以確保電力供應穩定。大型科技公司目前正透過購買能源公司來確保自身的電力供應。此外，燃料電池技術的興起顯示，部分公司將維持營運優先於完全消除碳排放。

The global AI infrastructure sector is now at a critical point where the need for more computing power must be balanced with the limited supply of local water and energy.

全球 AI 基礎設施產業目前處於一個關鍵點，對運算能力的需求必須與本地水資源和能源的有限供應之間取得平衡。