Data centers, the humming warehouses of the digital age, are facing a growing challenge: how to keep pace with the booming demand for AI while minimizing their environmental impact. A new report, “Accelerating Private-Sector-Led Energy-Infrastructure Investment,” lays out a path for data centers to embrace clean energy sources.
Released by the Tony Blair Institute for Global Change on May 30 at the ATxSummit in Singapore, the Greening AI report identifies a major hurdle: the slow and bureaucratic process of getting approval for renewable energy projects like solar or wind farms. Building new data centers can be a swift process compared to the years it can take to secure permits for clean power sources. This creates a roadblock for data centers seeking to reduce their reliance on fossil fuels.
The report proposes streamlining approvals for renewable energy projects, particularly in areas with ample sunshine or wind. This could involve faster approvals or dedicated pathways for data center development alongside renewable energy infrastructure. Such a joined approach would not only benefit the environment but also ensure a reliable power supply for the ever-growing demands of AI.
Another key strategy involves flexible pricing structures which reward data centers for shifting energy consumption to periods with abundant renewable power. For instance, data-intensive tasks could be scheduled for times with high solar or wind production. The report encourages governments to adopt such tariffs and highlights the success of countries like Ireland, where data centers have optimized energy usage through flexible demand arrangements.
However, change requires a two-way street. Data center operators, particularly co-location facilities, often prioritize maximizing space utilization over energy efficiency due to economic pressures. The report suggests that these operators work together with their customers. Customers can demand greener data center services, pushing operators to prioritize efficiency. This can be achieved by incorporating flexibility requirements into service-level agreements, allowing data centers to adjust operations based on grid conditions.
Looking ahead, the report proposes a Green-AI Certification Scheme which would establish clear benchmarks for data center energy consumption, efficiency, and carbon emissions. By recognizing best practices and setting ambitious targets, the scheme would create a competitive environment driving innovation in sustainable data center design and operation.
Current data center certification programs often focus solely on internal efficiency. The report proposes expanding these programs to encompass the use of renewable energy, investment in clean on-site generation, and integration with the wider energy grid.
The report also singled out Singapore for its efforts on optimising energy efficiency for data centers in tropical climate countries and greening AI software, adding that governments can learn from its experience.
Singapore leads the way in setting sustainbility standards for AI, said the report, with the world’s first sustainability standard for green data centers in tropical settings. It also promotes transparency of and accountability for the carbon emissions of AI software.
The release of its green data centre roadmap “will unlock additional green data center capacity in the coming years”.