According to Laura Cozzi, the Director of Technology at the International Energy Agency (IEA), the intensifying global tariff conflict is anticipated to pose significant hurdles for the burgeoning data center industry, potentially decelerating its growth prospects.
In a report released on Thursday, the IEA highlighted that the United States, China, and the European Union are collectively projected to contribute to 80% of the anticipated rise in data center demand by 2030, a trend predominantly driven by the increasing use of Artificial Intelligence (AI).
Cozzi pointed out that the IEA’s “headwind scenario” accounts for various current challenges, such as slower economic growth and expanded tariffs across multiple countries. In such a scenario, AI could experience reduced growth rates compared to the base case assumptions.
The agency’s data suggests that global electricity consumption by data centers might increase to around 945 terawatt-hours (TWh) by 2030 under the base scenario, while the headwind scenario estimates a decrease to 670 TWh.
The U.S. is expected to lead global data center development and is projected to account for nearly half of the electricity demand growth from now until 2030. U.S. utilities are receiving substantial requests for new capacity, surpassing current demand or existing generation capabilities.
This surge is raising concerns that technology firms are simultaneously seeking excess capacity from multiple power providers, potentially distorting demand forecasts.
The report emphasizes collaboration with tech companies and the industry to clarify the actual demand for data centers, stressing the importance of ensuring AI has sufficient electricity.
Additionally, supply grid constraints could result in project delays, with approximately 20% of planned data center projects facing potential risks. Demand for transmission lines and essential grid infrastructure is soaring, underscoring this concern.
Lastly, nearly half of the U.S. data centers under development are sited within existing large clusters, which may exacerbate local bottleneck issues, according to the report.
