Wall Street capital for AI data centers now depends on proven, scalable electricity access. JPMorgan warns that without secured power and permits, even well-backed projects are being sidelined as investors demand hard evidence of grid readiness.
AI infrastructure funding is facing a critical constraint: without guaranteed, scalable electricity, even the most ambitious data center projects are being excluded from Wall Street's capital flows. JPMorgan's investment banking division has established a new standard-power sourcing and regulatory permits now take precedence over hype or technical capability in determining which projects receive funding. The stakes are significant, as the global race to build AI-ready data centers encounters the physical limitations of the electric grid.
According to the International Energy Agency, global data center electricity demand is projected to more than double from 415 TWh in 2024 to around 945 TWh by 2030, with AI as the main driver.
This shift has made power access not just a technical challenge but a financial gatekeeper. Kevin Foley, JPMorgan's co-head of Global Investment Banking, emphasizes that without a credible, permitted power plan, projects are dismissed immediately. While track record and partnerships remain important, the primary filter is now grid access. As detailed in TheStreet, Foley notes that the source of power, permitting, and developer reputation are the main factors under investor review. This has resulted in higher borrowing costs and stricter terms, as more capital pursues a shrinking pool of viable projects.
GE Vernova, which supplies approximately 25% of the world's electricity through its equipment, is experiencing record demand from data center developers. Electrification equipment orders surpassed $5 billion in 2026-more than double the previous year. In Q1 alone, GE Vernova's Electrification segment booked $2.4 billion in data center orders, exceeding the total for 2025. The company's gas turbine and grid infrastructure backlog is on track to reach at least 125 gigawatts by year-end, while free cash flow reached $5.1 billion in the latest quarter, surpassing last year's total. Forbes attributes this surge directly to the "energy bottleneck" created by AI-driven data center expansion.
IEA data shows that U.S. data centers alone could account for nearly half of the country's increase in electricity demand by 2030, making grid access and permitting critical hurdles for new projects. This shift is forcing developers to focus on real supply chains, transformers, substations, and regulatory approvals, not just compute demand.
JPMorgan maintains that financing remains available for AI infrastructure, but only for projects that can demonstrate robust economics based on utility contracts and regulatory approvals. The project pipeline is active, but the threshold for entry is higher than ever. The speculative era-when AI projects could attract funding with little more than a pitch deck-has ended. Now, companies building the physical backbone, particularly those delivering power infrastructure, are emerging as the primary beneficiaries.
According to the U.S. Energy Information Administration, total U.S. electricity generation reached approximately 4,243 terawatt-hours in 2025, with data centers rapidly increasing their share. As AI adoption accelerates, regional grid constraints are prompting utilities and developers to enter long-term power purchase agreements and invest in capacity years ahead of deployment.
While AI and cloud computing dominate headlines, the decisive factor for future growth is now the infrastructure behind the scenes-specifically, power generation and distribution. Developers must navigate a complex landscape of utility negotiations, environmental permitting, and grid bottlenecks. For investors, understanding these operational realities is essential: the difference between a sustainable investment and a stranded asset now depends on the ability to secure and deliver electricity at scale. As the AI buildout intensifies, capital will increasingly flow to those who can guarantee the current, not just the code.
