Thesis:
Observation: Historically, public power credit analysis focused on stable to moderate load growth, predictable generation costs, adequacy of fuel mix diversification, rate-setting flexibility, financial metrics, capital plans, customer diversification and the regulatory environment. The credit shift for utilities may not necessarily be positive, as outcomes will rely heavily on appropriate cost allocations across individual data centers, the utility and existing ratepayers.
The question is whether a disproportionate burden will fall upon ratepayers, utilities, taxpayers or even bondholders, particularly if demand is overestimated with commensurate overinvestment, and actual load growth falls short of expectations. Political risk represents a top credit risk, and if the climate is not properly interpreted, the impact may find its way onto a utility's balance sheet. A clear objective of data center growth is to strengthen a utility's balance sheet, not just grow it.
The concept of stranded assets has returned to the public power narrative and the potential cost implications cannot be ignored. As we think about ways to look at public power credit quality, it is important to assess how effectively a utility can turn data center demand into sustainable credit strength without assuming disproportionate capital, customer concentration, contractual, execution and stranded asset risk. One of the least desired scenarios is a utility stuck with a much larger fixed-cost base without the benefit of corresponding revenue.
Current federal tax-exempt bond private-use rules can produce tax consequences for certain customized, long-term contracts between public power utilities and large private customers. This could render contracts longer than three years problematic when the applicable private-use conditions are met.
Call to Action: Electric utility credit analysis is moving into uncharted territory, given emerging operational, financial and execution risks.
Market stakeholders must determine whether the infrastructure and financing model built around this new demand for electricity will remain politically and economically viable.
Deal team members must stand ready to advise their public power clients regarding debt management practices and strategy, compliance with covenants and other contractual obligations, tax policy and stress-testing scenarios.
Contractual and rate structure evaluation must determine whether they provide for an appropriate cost and risk allocation for serving anticipated load growth to entities creating the demand. For issuers, the question is whether 20- to 30-year municipal debt is being issued against technology-driven demand assumptions that may have a considerably shorter useful forecasting horizon.
Shifting the burden onto ratepayers and bondholders is not a preferred outcome. Investors cannot afford complacency. Investors need to distinguish between headline data center exposure and actual financial exposure and the strength of the structured contract. Critical questions must be asked and the responses may result in portfolio realignment.
Over time, the market should price the financial and contractual framework of a given utility and how that insulates bondholders from unrealized loads. The strongest public-power credits will be those that successfully monetize data center growth while shielding legacy ratepayers and bondholders from development, construction, customer concentration and stranded asset risk.
___________________________________________________________________________
Public power is one of the more complex sectors across the public finance ecosystem. Enveloping an analytical framework of generation, transmission and distribution assets with expectations of future load growth and unprecedented capital-intensive needs is no easy feat. Public power is moving through a transformative phase where customer concentration and strategic execution are more important than ever to credit assessment.
While it is not a far leap to identify the upside to data center development, reconciling the downside is a very different task. Large new electric loads with resultant utility revenue growth, expansive infrastructure investment with accelerated grid and utility modernization, tax-base expansion and potentially accretive credit benefits for public power systems are all top of mind favorables tied to data center construction.
Determining who ultimately pays for associated infrastructure investment and how communities and regulators would view cost shifts onto ratepayers remain subject to debate. The American Public Power Association (APPA), which focuses on stranded costs and cost allocation risks, notes a utility may build generation, transmission and distribution infrastructure for a large load, but potential construction delays and below-forecast power consumption may expose customers to unforeseen financial risk. Utilities may experience higher fixed costs, fuel cost volatility and, potentially, higher debt service.
The artificial intelligence data center buildout conversation cannot catch its breath, and the anticipated load demand raises multiple cost-benefit arguments for municipal governments/enterprise units and their taxpayers as well as for rate payers and electric utilities. Forecasting load growth combines art and science, and the need to get it right is driven by multiple factors that tie together commitment, potential credit impact, infrastructure planning and investment and long-term economic viability.
Researchers are studying whether data center load flexibility can materially lower the need for certain infrastructure investments, rather than having a structure whereby aggregate forecasts are directly matched with permanent capacity. These types of investigations can generate favorable outcomes for multiple stakeholder groups and address some of the risk exposures.
The potential impact of data center growth points to mixed results, with emerging risks and opportunities. The
Perhaps more economically diverse states could better support data center development while being able to absorb potential losses. Local governments must be mindful that although data center growth can create construction employment opportunities, such jobs tend to be transitory. This could limit the impact on tax-revenue receipts. Local governments may also have to commit resources to develop the infrastructure necessary to support data center needs. However, local budgets tend to be less flexible than states', with even greater competitive forces pressuring resources for essential services.
The rise in data center infrastructure requirements will continue to pressure water and electric utility resources, with potential credit implications. While new connections would be accompanied by an accretive impact to sales revenue, the added capital expenditure costs for utilities may be disproportionate.
Growing political and community resistance to data center development is not expected to abate anytime soon. We are already seeing expanded legislation to repeal or restrict tax incentives, a greater sense of bipartisan opposition, expanding regional moratoriums and lengthier approval processes. Migration to friendlier data center regions may occur with greater regularity.
The "build it and they shall come" concept falls far short of validating a suitable risk management strategy. The amount of infrastructure development needed to support data center buildout is extensive, and significant completion needs to occur prior to the new load coming on line.
Utilities cannot wait until a hyperscaler is operating at 100% capacity before building substations, transmission, generation and other infrastructure assets. This is why it is critical to identify correct load forecasts. Community and political support as well as economic feasibility go a long way toward successful data center planning, acceptance, buildout and project completion.
Data center project construction barriers include timing issues, delays, capacity changes and technological obsolescence. While it seems many projects are moving with great speed, unchecked or unrestrained data center buildout is not sustainable. If a data center project falls through, even after the spenddown of significant funds, associated electric utility debt remains.
The data center craze has emerged as a legitimate investment theme, but with concerns, and the force of opposition grows. Data centers have become a political "hot button" issue that voters are likely to take into the November elections against a backdrop of active resistance in many jurisdictions. One must question the level of community support surrounding data center buildout. The permitting and approval process is expected to grow lengthier and environmental concerns run deep.
The amount of debt issued by hyperscalers may not be commensurate with demand, and a recurring question is whether hyperscaler debt is crowding out U.S. Treasury debt. Credit default swaps are widening for a number of these companies, signaling investor concerns over swelling debt loads.
Looking at this through a different lens, hyperscalers tend to boast strong balance sheet liquidity with very high levels of cash and cash equivalents. Such strength and flexibility can help fund data center development until the facility is operational. Hyperscalers can afford to compensate the community while they await construction, buildout and execution. It would be mutually beneficial to provide the community with something meaningful, such as contributions to local education.
The issue of cost allocation resonates across the public power ecosystem, and associated concerns should not be discounted, even though there is great potential for electric utility growth. Research and policy organizations — as well as federal initiatives — recognize the cost allocation distinctions surrounding data centers.
The Federal Energy Regulatory Commission's (FERC) large load orders have taken coordinated action to accelerate integration of large loads onto the grid while protecting ratepayers and promoting reliability. Cost recovery agreements ensure large loads pay for the infrastructure built to serve them, even if the project fails, preventing residential ratepayers from absorbing stranded costs. FERC's large load orders represent a region-specific effort to modernize grid access for high-demand users, potentially leading to faster, cheaper and more flexible large load integration throughout the United States.
Public utilities are taking proactive steps to minimize stranded cost exposure. Baker Tilly, a leading municipal advisory firm, points out utilities are incorporating firm contracts, minimum demand commitments and take-or-pay structures. A strong contractual framework can make a great deal of credit difference and a more standardized use of these features would be welcomed.
Other contractual features may include termination payments, security deposits, collateral posting, construction milestones and specific stranded cost make-up provisions. While the top-line load number is important, in many ways the structure and provisions of the underlying contracts are more relevant, as they provide the legal pathway to risk ownership.
When it comes to data center development, there is clearly a two-sided argument, with ardent opposition pointing to an inevitable bubble. A McKinsey & Co. analysis takes a counter-view, positing that the larger near-term power sector risk may actually be underbuilding as opposed to overbuilding, even if "compute demand" falls short.
For those public power utilities taking on a credit bet, the financing structure must distinguish between real and anticipated demand. For most municipal sectors, diversification is a key credit determinant. In the public power space, customer diversification is critically important. The new load associated with a data center can alter the diversification profile of a utility by creating concentration risk. While this risk can yield revenue growth, margin and debt service coverage expansion, as well as economic development, it should not be underestimated.
Current tax law adds a few wrinkles
Like so many topics highlighted in The Bond Buyer's Market Intelligence content, tax law plays a role in the public power and data center conversation. Data centers have not been granted Internal Revenue Service "exempt facility status" and tax-exempt private-activity bond volume caps have not been expanded to broadly finance data center development.
APPA points out federal private-use rules could constrain long-term contracts between public power utilities and large customers. This could limit a utility's ability to enter into 10-, 15-, or 20-year contracts that would otherwise shield exposure to stranded infrastructure costs.
Given hypothetical investments made by a utility in generation, transmission, substations and distribution infrastructure on behalf of the data center, it makes perfect sense that the utility seeks a long-term commitment to take power and pay for associated expenses. The infrastructure investments could have useful lives of 20 to 40 years following a lengthy construction period, and the utility is concerned about recouping its investment and protecting customers.
APPA argues, under current tax law, a customized retail contract with a large customer that extends beyond three years can potentially constitute "private use" of facilities financed with tax-exempt municipal bonds. Simply put, if enough of the bond-financed property is considered private use, the bonds' tax-exempt status can be revoked. This directly pits the need to preserve tax exemption against establishing long-term contractual protection. This issue is less consequential when a utility is adding incremental load, but the outsized loads required by data centers alter the dynamic.
As previously pointed out, an unexpected withdrawal by the developer/hyperscaler can create many problems, and the specific type of invested infrastructure would not ordinarily lend itself to retrofitting. Utilities would likely confront higher rates, eroding margins, larger debt burden, and reduced liquidity. In this case, while current rules are intended to preserve the tax-exemption on municipal bonds, potential harm to electric utility credit quality may be an unintended consequence.
APPA's proposal advocates for the ability to enter 20-year contracts with large-load customers. Here, the objective is for the utility to legally and financially structure the customer contract in a manner that matches the useful life and debt amortization of the infrastructure.
Stakeholders, particularly investors, must know precisely how much capital is being committed to a data center and the sources of this capital. As outlined, there are clear distinctions between tax-exempt and taxable bond financings. Investors should also be familiar with the terms of the contract with a focus on termination provisions, availability of collateral, minimum demand commitments, rate flexibility and potential economic redeployment of infrastructure. A complete analysis must determine whether the contractual, regulatory and financing structure is able to convert anticipated load into sufficiently viable cash flow to support the capital investment.
Capital investment for data center-related infrastructure is very nuanced. For tax-exempt deals, use of bond proceeds is detailed (or should be) in the official statement far more extensively as compared to use of taxable bond proceeds. The tax provisions outlining the permissible uses of tax-exempt bond proceeds are specific, as are the rules surrounding private use and exempt facility designation.
The issue is not that private use rules are compelling hyperscalers to finance their utility infrastructure largely with taxable municipal bonds. The true observation is the tax-exempt financing tool and the contractual structure needed to support the investment are not perfectly compatible.
For example, tax-exempt bonds are appropriate for a utility's general generation, transmission, and distribution system and multi-customer systemwide improvements. If infrastructure financing is needed for one hyperscaler under a customized long-term contract, private-use concerns come into play. If infrastructure financing exceeds the permitted private-use threshold, this creates a potentially taxable, private-activity situation. If financing is required for a data center's building, servers, equipment and campus, there is no appropriate use of municipal tax-exempt debt.
A combination of tax-exempt financing and a short customer commitment offers lower borrowing costs, but potentially higher stranded cost risk. Taxable financing coupled with a long customer commitment produces higher borrowing costs, but potentially better contractual protection. A hybrid structure combines tax-exempt financing for the broadly public portion of the system with taxable/private financing for the portion that is largely dedicated to the hyperscaler.
Analytically speaking, the unprecedented scale and concentration of new data center load is exposing stress between the enduring nature of tax-exempt public-power infrastructure financing and federal private-use rules that can challenge the duration of contractual commitments from the private customers driving the investment.
Each stakeholder group should consider the following questions:
Issuer: How much infrastructure should we build before the load is contractually committed?
Municipal advisor: Does the financing structure adequately stress-test the load forecast and customer concentration? What is the impact upon the utility's multiyear forecast if the data center experiences one- to two-year delays?
Bond counsel: Do contracts, private-use issues and security provisions protect the tax-exempt financing?
Banker/underwriter: Can the utility explain the required leverage and demonstrate the project is viable under downside scenarios?
Investor: Does projected load growth improve or weaken the utility's underlying credit profile; what is the rate and debt service coverage impact if the load does not materialize? What will the ultimate customer concentration look like? How do we develop pricing distinctions for different types of public power data center exposure; specifically, how do we price different types of credit enhancements? How much load is actually contracted?
All stakeholders: Is the utility building infrastructure for a customer, or is the utility effectively speculating on the arrival of a customer? If the data center does not materialize, who will pay the infrastructure costs? Is economic redeployment of the infrastructure feasible? Is there available collateral? What is the optimal plan to develop market reception? What are the specific contractual provisions; are there termination clauses?
