ACC News Brief
AI & Data Centers Phoenix, Frankfurt, and Singapore
Peer-reviewed planning model - not a universal procurement or reliability conclusion
Data-center model favors batteries over hydrogen for solar balancing
What happened
A peer-reviewed modeling study compared solar-powered battery and green-hydrogen storage pathways for a hypothetical 10-megawatt data center in Phoenix, Frankfurt, and Singapore. Under the study's common assumptions, battery-system levelized costs ranged from $0.097 to $0.206 per kilowatt-hour, compared with $0.155 to $0.295 for hydrogen. A combined storage search selected battery-only configurations, 42.17 to 58.02 percent below the modeled hydrogen costs at the target.
ACC context
The model uses synthetic hourly loads, a minimum renewable fraction, defined equipment costs, and residual grid support. It is useful for comparing pathways within those boundaries, but it does not replace a facility-specific engineering study or establish that batteries are always preferable to hydrogen.
Why it matters
AI and cloud growth is increasing the importance of credible power planning for data centers. The study suggests batteries may be the stronger fit for the mainly daily solar-balancing problem it modeled, while also showing that climate, cooling load, grid price, renewable fraction, and storage duration can materially change system design.
What to watch
- Site-specific studies using measured facility loads, local tariffs, interconnection constraints, land limits, and actual grid carbon intensity.
- Comparisons at longer storage durations and higher renewable fractions, where hydrogen may serve a different role than daily balancing.
- Full reliability engineering, including UPS behavior, redundancy, emergency generation, component failures, and restoration times.
Sources & evidence
- Cost-decarbonisation trade-offs in solar-powered data centresOpen-access International Journal of Hydrogen Energy study, volume dated August 28, 2026, DOI 10.1016/j.ijhydene.2026.156806.
Continue learning with ACC
Understand the energy, water, and emissions footprint of AI and data centers