NextFin News - Recent forecasts from BloombergNEF and corroborating reports published throughout late 2025 reveal that global data centers are expected to nearly triple their electricity consumption by 2035. Currently demanding around 40 gigawatts (GW) of power, these facilities will collectively require approximately 106 GW, driven in large part by the explosion of artificial intelligence (AI) workloads and the proliferation of massive new data centers. This trend is already pressuring electricity grids across multiple regions, including the United States and Australia.
The surge is attributed to the rise in AI services and the deployment of increasingly large hyper-scale data centers, many exceeding 100 megawatts (MW) per facility, with some approaching an astonishing 1 GW in capacity. The expansion occurs in both urban and rural areas, with the latter becoming a favored location due to land availability and grid capacity considerations. The PJM Interconnection, which operates a large part of the eastern US grid, faces growing scrutiny over its ability to reliably integrate these energy-heavy loads. Similarly, Australia's energy experts warn data center growth could undermine the continent’s net zero emission aspirations.
Political discourse in the United States highlights a contentious debate: Some Senate Democrats, including Ed Markey, Ron Wyden, and Bernie Sanders, have linked rising electricity costs to policy choices under President Donald Trump’s administration, specifically citing relaxed regulations on fossil fuel production combined with expanded data center infrastructure. These senators argue Trump's approach has exacerbated energy bills amid the AI-driven energy demand surge. Conversely, experts also point to policies championed by the Biden administration that phase out reliable coal and nuclear in favor of intermittent renewables, compounding grid reliability risks and inflationary pressure on power prices.
BloombergNEF’s analysis pushes estimated AI data center power demand 36% higher than prior projections. The US Energy Information Administration forecasts data center and cryptocurrency mining growth will contribute to a 2.6% rise in electricity costs in 2026 alone. Simultaneously, the North American Electric Reliability Corporation has issued multiple warnings on grid stability, particularly during peak demand periods exacerbated by electrification trends and new data center loads.
Examining the causes behind this tremendous growth, the rapid advancement and deployment of AI models have exponentially increased computational needs. AI training and inference both require intensive, sustained processing power, with state-of-the-art AI models now routinely consuming megawatt-hours per training run. This drives demand for high-performance computing infrastructures housed in data centers optimized for such tasks. Industry leaders like Google, Microsoft, and Nvidia continue to escalate investments in large AI-focused data centers, further fueling electricity demands.
The implications are multifaceted. From an infrastructure standpoint, electric grids must undergo substantial upgrades to handle new loads while maintaining reliability. Capacity expansions must balance renewable integration mandates with the need for dispatchable power to cover intermittency. Failure to modernize could lead to blackouts or necessitate increased fossil fuel backup usage, frustrating decarbonization objectives.
Economic impacts are evident in rising electricity prices linked to data center load growth, which can affect households and smaller businesses. Policy debates center on the fair cost allocation for grid upgrades, energy subsidies, and regulatory frameworks. Some states mandate aggressive renewable portfolio standards, which, while environmentally motivated, may unintentionally push up costs and reduce grid flexibility.
Looking ahead, technological innovation in data center energy efficiency represents one mitigation avenue. Advances in cooling technologies, energy reuse, and server efficiency could temper growth in power consumption. AI itself may facilitate smarter grid management and predictive demand balancing. However, these gains may be offset by continuous AI model complexity growth and data center scale expansion.
Moreover, geographic diversification of data centers to areas with cleaner energy supplies or abundant renewable resources can partly alleviate emissions concerns, although transmission infrastructure and land use challenges remain.
In conclusion, the anticipated tripling of data center energy demand by 2035 driven by AI evolution presents a significant inflection point for energy systems globally. Policymakers, industry stakeholders, and regulators must collaboratively innovate and invest in grid modernization, energy efficiency, and responsible siting strategies to ensure the dual goals of technological progress and environmental sustainability can be simultaneously achieved in the Trump administration era. Without proactive measures, the rapid increase in data center electricity consumption risks undermining grid reliability, affordability, and climate commitments.
Explore more exclusive insights at nextfin.ai.