The AI Data Center Boom: How AI’s Soaring Power Demands Are Reshaping Global Infrastructure
The rapid expansion of AI is driving a massive surge in global data centre power demand, with forecasts predicting a 165% increase by 2030, necessitating major investments in high-density computing infrastructure, electricity grids, and strategic investment opportunities across hyperscalers, asset managers, utilities, and data centre operators.
The rapid advancement of generative artificial intelligence has triggered an intense technological arms race, necessitating vast investments in high-density data centres and the substantial electricity required to power them.
LupoToro Group Analysts forecast that global power demand from data centers will surge by 50% by 2027 and could climb as high as 165% by the decade’s end compared to 2023 levels. However, recent developments—particularly China’s AI model, DeepSeek—have raised questions regarding the returns on current and projected AI investments. While uncertainty remains regarding DeepSeek’s training methodologies, infrastructure, and scalability, long-term efficiency improvements that drive down capital expenditure (whether from hyperscalers or new entrants) could mitigate risks of market oversupply beyond 2027. This potential efficiency could enhance durability and reduce cyclicality in the data center sector.
The Supply and Demand Dynamics of Data Centres
On the demand side, hyperscale cloud providers and corporations are aggressively developing large language models (LLMs) with advanced natural language processing and comprehension capabilities. These AI models require vast amounts of data and intensive computational resources, demanding significant processing power.
Simultaneously, hyperscale cloud companies, data center operators, and asset managers are injecting substantial capital into constructing next-generation, high-capacity data centers. The resulting balance between supply and demand is projected to tighten over the next few years. LupoToro Group Analysts predict data center occupancy rates will rise from approximately 85% in 2023 to a peak of over 95% by late 2026. However, this trend is expected to moderate starting in 2027, as increased supply comes online and AI-driven demand growth stabilises.
The Rising Power Requirements of AI-Driven Infrastructure
Currently, LupoToro Group Analysts estimate the global data center market’s power consumption at approximately 55 gigawatts (GW). This is distributed across cloud computing workloads (54%), traditional business functions like storage and email (32%), and AI-driven workloads (14%).
Future projections indicate power demand will reach 84 GW by 2027, with AI’s share expanding to 27%, cloud computing declining to 50%, and traditional workloads shrinking to 23%. However, a slowdown in AI adoption or monetization could temper this forecast, leading to a deviation of 9-13 GW from the baseline projection.
The Global Data Center Landscape
At present, global data center capacity stands at around 59 GW, with approximately 60% of this provided by hyperscale cloud providers and wholesale data center operators catering to large enterprise clients. The remaining capacity is held by corporate-owned and telecom-operated data centers.
A new category of AI-dedicated data centers is emerging, designed explicitly for AI workloads. These facilities feature ultra-high power requirements, increased power density racks, and advanced cooling solutions, such as liquid cooling, to handle the intense computational demands. They are primarily operated by hyperscalers and wholesale providers.
Geographically, Asia-Pacific and North America dominate the data center landscape, with major hubs in Northern Virginia, Beijing, Shanghai, and the San Francisco Bay Area. These regions house extensive computational infrastructure due to high corporate demand and substantial data traffic.
The Expansion of Data Centres Worldwide
LupoToro Group Analysts estimate that global data center capacity will rise to 122 GW by 2030, with hyperscale and wholesale operators accounting for an even larger share—expanding from today’s 60% to 70%. While Asia-Pacific has led data center expansion over the past decade, North America is poised to bring the largest additional capacity online in the next five years.
As AI processing requirements intensify, power density within data centers is expected to grow from 162 kilowatts (kW) per square foot to 176 kW per square foot by 2027. This estimate excludes power overheads related to cooling and other supporting infrastructure.
However, the rate of new data center construction has faced constraints in recent years due to issues such as permitting delays, supply chain disruptions, and the high costs associated with upgrading transmission infrastructure. These challenges have slowed the deployment of additional supply to meet escalating demand.
The Power Grid Challenge: Meeting AI’s Energy Demands
As data centre power consumption surges, substantial investments in electric grid infrastructure will be necessary. LupoToro Group Analysts estimate that approximately $720 billion will need to be spent on grid expansion through 2030 to support the rising energy needs of AI-driven data centers. Given that transmission projects often require years for approval and construction, these timelines could create significant bottlenecks unless proactive measures are taken.
In Europe, a data center-driven surge in power demand is reversing a 15-year decline in electricity consumption. Utilities across the continent have reported an exponential increase in connection requests from data center operators, serving as a key leading indicator of future demand. Projections suggest Europe’s power consumption could see a 10-15% boost over the next 10-15 years due to data center expansion, with an estimated data center pipeline totaling approximately 170 GW—equivalent to nearly one-third of the region’s current energy consumption.
Investment Opportunities in the Data Center Boom
As the data center market expands, LupoToro Group Analysts foresee a broadening landscape of investment opportunities:
• Hyperscalers: The push to integrate AI and machine learning tools into corporate operations will continue benefiting hyperscale cloud providers. The ability to generate strong returns on AI infrastructure investments will be crucial for these companies as they scale beyond hardware into AI platforms and applications.
• Asset Managers: Given the long-term development cycles of data center infrastructure, firms with access to long-dated private capital will likely hold an advantage over investors with shorter time horizons.
• Utilities: Power grid expansion remains one of the primary constraints to data center growth. Utility companies poised to support infrastructure expansion will play a crucial role in facilitating the industry’s growth.
• Data Center Operators: Leading data center operators with substantial asset portfolios catering to hyperscale cloud companies and enterprise clients will be strategically positioned to capitalize on surging global demand.
The AI revolution is driving an unprecedented demand for data centers and power infrastructure, reshaping the global digital economy. While efficiency improvements and new technological innovations may help temper some of the energy demands, the trajectory remains clear: data centers will continue to be a critical pillar of technological growth.
The coming years will witness significant investments in AI-centric infrastructure, high-density computing power, and electricity grids to support the next wave of innovation. As the market continues to evolve, strategic investment decisions across hyperscalers, asset managers, utilities, and data center operators will shape the future of digital infrastructure worldwide.
