By Kennedy Maize
Data centers and AI have been prominent in the news, focusing largely on the impact on electric prices, incidents of political push back, and the thorny question of how they connect, or don’t, to the grid.
Less attention has been paid to how these big money infrastructure finance their operations and how the potential costs of those investments could arise.Â
Show us the money. New reports from two management consulting firms have done just that.
CSC’s Project Finance Report 2026 identifies “three main shifts in AI data center financing. These shifts are not just about who provides capital, but also about how projects are structured and executed as power demand accelerates and transactions become more complex.”
The first is a move“from cloud-era self-funding to AI-era capital intensity.” Christian Oakley-White, managing director and head of project finance at CSC, said, “Earlier waves of expansion were often funded by Big Tech’s internal cashflows. But as intended data center usage evolves from supporting cloud services to generative AI, the scale of computing power, energy, and financing will grow exponentially.”
The next trend is that “financing structures are becoming more sophisticated.” According to CSC, “Financing structures are no longer just about raising capital, they are also crucial in managing inherent risks, and whether that affects the bankability of a project.
Financial innovation is becoming increasingly important as financing methods evolve, including the use of financial engineering and insurance products. Ensuring the bankability of new technology and infrastructure requirements is top challenge according to 60% of respondents, especially are project finance adapts to future trends.”
Third, “a broader investor base means greater execution pressure.”
CSC observes that as “participation widens across private equity, sovereign wealth, banks, public debt markets, and private credit, the $1.5 trillion gap will less likely be filled by traditional bank lending alone.
CSC comments, “While many practitioners still consider development finance institutions (DFIs) and infrastructure funds and platforms to be primary sources of equity, private equity (53%) and private credit (38%) are fast catching up.
On the debt side, private debt (45%) is already a significant component, closely aligned with infrastructure funds and platforms (48%) and both international and local syndicated loans (42% and 45%).”
What’s ahead for data center finance? “Since no single financing approach is likely to dominate, the teams that will gain are those that pair access to capital with the ability to run complex transactions cleanly.”
BRG looks at the skyrocketing demand for power created by AI and data centers, “creating a spike in electricity demand and necessitating costly investments that new and/or existing utility customers must pay.” This puts state and federal policymakers “in a bind: data centers are a strategic imperative, but consumer backlash is becoming a major political concern and leading to proposed moratoriums on new buildouts….Put simply, demand for power is outpacing supply and deliverability at a speed the grid has never experienced.”
Two approaches are approaching: bring your own generation (BYOG) and backstop generation procurement. CSC looks at the pros and cons of each.
The report notes that “BYOG projects still often rely on the grid for backup power and restart capabilities. Therefore, the Federal Energy Regulatory Commission (FERC) favors co-located customers paying for the transmission, distribution, and grid services they consume.”
If the generators fail and the data centers switch to the grid “the result can be stress on grid reliability. For generators: If the data center exits or fails, the generator may have no easy alternative buyer without transmission deliverability.” For non-data center customers — the vast majority of customers — what if the data centers don’t pay for their grid usage? “FERC is trying to mitigate these risks by, for instance, ordering PJM to establish clear, nondiscriminatory interconnection rules for generator-connected loads and offer new transmission services to prevent cost-shifting.”
The alternative to BYOG is backstop generation procurement, “when regional transmission organizations (RTOs), independent system operators (ISOs), or adjacent procurement authorities obtain generation capacity outside existing mechanisms.” PJM has announced that they are developing a “backstop reliability” procurement, but no details are available.
There is a downside to backstoping, the report notes, “If even a portion of AI load is transient, rapid backstop procurement results in significant overbuilding. Accurately forecasting load growth is crucial in this approach.” If backstop generation is procured through long-term contracts instead of yearly auctions, “regular customers’ payment obligations likely will shift to remaining customers in the event of data center exits.” If utilities build new transmission to serve data center loads that don’t materialize, “they and/or their customers will have to cover the costs.”
Neither path — BYOG or backstop generation — “offers a silver bullet. New firm generation cannot be deployed overnight, regardless of regulatory structure. Market forces—not regulators—determine the price of natural gas, turbines, and other critical equipment. Meeting rising data center demand will require pursuing additional strategies in parallel, including greater reliance on distributed energy resources and storage. Both technical opportunities to strengthen the grid and the long‑standing debate over how to allocate costs will persist even after the immediate surge in demand is addressed.”