Data Center Power Demand, 165% Growth Forecast, $1.4 T US Investment, and 49 GW Supply Shortfall (2025 to 2030)

Industry Risks, Gartner Projects 26% Electricity Growth, Highlighting a 49 GW Power Shortfall

The rapid expansion of artificial intelligence is creating a structural power deficit, moving the primary industry risk from generation cost to generation availability and speed of deployment. The challenge for energy leaders is no longer solely about finding the lowest-cost electron but about securing a reliable, high-volume power source that can be brought online within the compressed timelines of data center construction. This shift invalidates traditional load forecasting and elevates the “time-to-power gap” to the most critical risk for both technology companies and the utilities that serve them.

AI-Driven Demand Acceleration

The energy demand from data centers is expanding at a non-linear rate, driven by the intense computational requirements of AI workloads. Before 2025, data center demand growth was a predictable component of utility planning. However, recent analysis confirms a dramatic acceleration that legacy infrastructure cannot support.

• Gartner projects that worldwide data center electricity consumption will reach 565 terawatt-hours (TWh) in 2026, a 26% increase from 447 TWh in 2025.
• Goldman Sachs forecasts a 165% increase in data center power demand by 2030, attributing the growth primarily to AI, which is expected to account for 27% of the 84 gigawatts (GW) total data center demand by 2027.
• In 2023, U.S. data centers already consumed 4% to 5% of the nation’s total electricity, a figure that is now on a much steeper trajectory, creating a significant challenge for grid operators as highlighted in the EUDCA report on data center power.

The Time-to-Power Gap

A fundamental mismatch exists between the rapid construction of data centers and the protracted timelines for developing and interconnecting new large-scale power generation. This “time-to-power gap” represents a major bottleneck and a significant financial risk, as multi-billion dollar data centers may sit idle waiting for energy.

• Morgan Stanley Research forecasts that U.S. data center power demand could reach 74 GW by 2028, while projecting a potential shortfall of approximately 49 GW in available power access.
• This supply-demand imbalance, identified by Enverus as the “time-to-power gap, ” is the Achilles’ heel of the AI build-out, creating a substantial opportunity for agile energy providers who can deploy assets quickly, including through private power infrastructure.
• While the issue was emerging between 2021 and 2024, it has become an acute constraint from 2025 onward, forcing a strategic re-evaluation of grid planning and investment across the sector.

$1.4 T in US Power Sector Investment, Deloitte Forecasts Unprecedented Capital Allocation by 2030

Capital is aggressively flowing toward solutions that can close the power gap, with massive investments targeting both generation assets and the enabling AI infrastructure. The sheer scale of the required capital, equivalent to more than a decade of prior spending compressed into a five-year window, signals a historic opportunity for developers, financiers, and equipment suppliers capable of executing projects at speed.

Capitalizing on the Power Deficit

The financial markets have recognized the data center power constraint as a primary investment theme for the latter half of the decade. The numbers indicate a fundamental repricing of power generation and infrastructure assets that can serve this specific demand.

• Deloitte estimates that U.S. power sector investments could reach as high as USD 1.4 trillion between 2025 and 2030, a figure equal to the industry’s entire capital expenditure over the previous 12 years.
• This is driven in large part by AI-related capital expenditures, which a Goldman Sachs model implies could reach USD 765 billion annually in 2026, growing to USD 1.6 trillion.
• This spending is not just on servers but on the entire power and cooling infrastructure, with the offsite data center power market projected to grow from USD 13.4 billion in 2024 to USD 35.4 billion by 2030.

Financing Structures for New Generation

Project finance experts anticipate a robust deal flow in 2026, as developers leverage the strong credit of data center offtakers to secure funding for new energy projects. The focus is on bankable contracts that provide revenue certainty.

• The availability of long-term Power Purchase Agreements (PPAs) from high-credit technology companies is the critical enabler, de-risking projects and attracting debt financing for everything from natural gas peaker plants to large-scale Banpu Power energy storage systems.
• While the period from 2021 to 2024 saw a focus on securing PPAs for traditional renewables, the 2025-2026 period is characterized by a search for firm, 24/7 power, increasing the bankability of assets like geothermal, advanced nuclear, and carbon capture projects.

Table: Forecasted Capital Allocation to Address Energy Demand

On-site vs. Grid-Scale, Data Center Power Solutions Shift Toward Commercial Readiness

The power deficit is accelerating the commercialization of alternative and on-site power solutions, moving technologies like fuel cells and advanced energy storage from pilot stages to critical infrastructure components. As grid-scale solutions struggle with interconnection queues and permitting delays, technologies that can be deployed “behind the meter” offer a path to faster deployment and greater energy resilience for data center operators.

Validating Dispatchable Power Sources

The 24/7 operational needs of data centers have exposed the limitations of relying on intermittent renewables and the isolated metric of Levelized Cost of Energy (LCOE). This has renewed focus on firm, dispatchable power sources that provide reliability, even if their standalone LCOE is higher.

• Analysis from the 2021-2024 period highlighted the challenge of integrating renewables, showing that low LCOE for solar and wind does not account for the high system-level costs of storage and grid balancing.
• As a result, technologies like geothermal, advanced nuclear, and natural gas with carbon capture are gaining commercial traction post-2025 as essential components for a reliable, low-carbon grid capable of supporting AI workloads.

The Rise of On-Site Generation

On-site power generation is emerging as a key strategy to bypass grid constraints. Technologies like Solid Oxide Fuel Cells (SOFC) offer a clean, compact, and scalable solution that can be deployed directly at a data center campus, providing both primary and backup power.

• The need for rapid deployment makes on-site solutions like those from Power Cell and Plug Power increasingly attractive, as they can be installed on timelines that align with data center construction.
• Companies like Mitsubishi Power and Wechai Power are developing SOFC technology, while major tech companies like Microsoft are exploring hydrogen fuel cells as part of their strategy to power future data centers.
• The nascent Long Duration Energy Storage market, valued at USD 964.8 million in 2025 and projected to nearly triple by 2033, is another critical enabler for integrating both on-site and grid-scale renewables.

US Power Grid Faces 49 GW Shortfall, Highlighting Regional Concentration of Demand (2025 to 2028)

While AI data center demand is a global phenomenon, the most acute power constraints and investment activities are currently concentrated in specific regions of the United States. This geographic focus is creating unique regional markets for power generation and infrastructure, driving localized policy and investment responses that differ significantly from those in Europe and Asia.

The US as the Epicenter

The United States is the primary market where the “time-to-power” gap is most pronounced. The combination of a massive installed base of data centers and the headquarters of major AI developers has concentrated new demand in a few key states.

• The Morgan Stanley forecast of a potential 49 GW power shortfall by 2028 is a U.S.-centric figure, reflecting the intense build-out in markets like Virginia, Texas, and Ohio.
• Reports from bodies like the Texas comptroller and analysis from Wood Mackenzie on U.S. gridlock confirm that regional transmission and generation capacity is the main limiting factor for data center expansion in the near term.

European and Asian Responses

Other regions are facing similar challenges but with different market dynamics and policy tools. In Europe, the primary constraint is often cross-border transmission bottlenecks, while Asia is focused on managing the carbon intensity of its manufacturing supply chains.

• The European Commission has identified internal grid bottlenecks as a key impediment to matching new renewable supply with demand centers, a problem exacerbated by new data center loads.
• In Asia, the focus remains on reducing the LCOE of renewables and addressing the carbon footprint of manufacturing, such as the coal-powered manufacturing of solar PV components noted by the IEA. Blockchain-based trading platforms, involving companies like Power Ledger, offer a potential path to optimizing grid usage in these complex markets.

SWOT Analysis, AI Data Center Power Demand Creates Both Opportunity and Systemic Risk

The surge in data center power demand presents a generational investment opportunity for the energy sector, but it is constrained by significant weaknesses in grid infrastructure and threats from supply chain vulnerabilities. Strategically navigating this landscape requires a clear understanding of the interplay between the immense market pull from AI and the physical limitations of the power grid.

Table: SWOT Analysis for Data Center Power Demand

2026 Scenario, On-Site Power Adoption Will Accelerate if Grid Interconnection Delays Persist

If grid-scale generation and transmission projects continue to face multi-year delays, data center operators will increasingly adopt on-site and behind-the-meter generation solutions to meet their aggressive deployment timelines. This will mark a significant shift in infrastructure strategy, moving from a centralized utility model to a distributed, customer-driven one.

• If this happens: The market will see an acceleration of corporate PPAs and direct equipment orders for on-site technologies, particularly solid oxide fuel cells (SOFCs) and long-duration battery storage systems.
• Watch this signal: State and federal regulators beginning to introduce specific, fast-track permitting pathways for data-center-adjacent power projects, separate from traditional utility-scale review processes.
• This could be happening: Major technology firms will move beyond simple offtake agreements to form direct joint ventures or make strategic investments in energy technology providers, as seen with Microsoft’s exploration of hydrogen, to secure their power supply chain.

The questions your competitors are already asking

This report covers one angle of the new power market created by AI. The questions that matter most depend on your work.

• new power plants being built for data centers
• tech companies signing contracts for 24/7 power
• Virginia power grid capacity for new data centers
• data centers using on-site power generation

This report does not answer these. Enki Brief Pro does.

Your question, your angle, your framework. SWOT, PESTL, scenario modelling. The same niche depth, built around the decision your work actually depends on.

Run your first brief in Enki Brief Pro