This article presents S&P Global Ratings' global criteria framework for rating data center securitizations. (See Appendix 1 for details on the scope of these criteria and Appendix 2 for a glossary of relevant terms and concepts). For information about the initial publication of this article as of June 13, 2024, including key features, the impact on ratings, and superseded criteria, see "New Global Data Center Securitizations Criteria Published."

METHODOLOGY

Overview

We use these criteria to analyze data center securitizations (see Appendix 1 for further details on scope). Data center securitizations are transactions backed by income generated from data center operations. When the issuer owns the related properties, the transactions also benefit from the proceeds arising from their sale. In general, we analyze both the recurring lease income and proceeds related to property sales. For issuers that don't own the data centers or corresponding land, only the recurring revenue streams are analyzed because noteholders have no recourse to the physical assets; instead, issuers are assigned both real estate leases and tenant contracts.

For each transaction, we analyze exposure to events that could interrupt or reduce the expected cash flow, including factors that could impede finding new tenants or selling properties. We make assumptions regarding:

• The value of properties at the time of sale;
• Revenue streams from contracts signed after closing; and
• Tenants' ability to meet their contractual obligations.

Although transactions are often refinanced at the anticipated repayment date (ARD), which are typically five to seven years after closing, the time to maturity is longer (up to 30 years). Our analysis extends to the legal maturity of the liabilities because there is no obligation to redeem the liabilities at the ARD. Typically, bond amortization is accelerated after the ARD.

We take a step-by-step approach to analyzing pools of data center assets and their related cash flows (see chart above):

• Step 1: Determine the utility score. This allows us to incorporate our view of the relative strength of the properties in the pool into our assessment of the timing of asset liquidations and capital improvement needs. A high score indicates that a property is more likely to maintain its competitive advantage over time, and less likely to become obsolete.
• Step 2: Determine the liquidation value and timing. This step applies only to transactions where the issuer owns the data center and land, owns the data center with exposure to a ground lease that has a sufficiently long remaining term, or owns the land and not the data center. We estimate the sustainable sale value of each property using our CMBS property evaluation methodology and then apply a haircut. Liquidation timing is informed by the property's utility score, determined in Step 1.
• Step 3: Estimate how lease income may alter over time, and the size and stability of any other revenue sources available to the issuer.
• Step 4: Apply a utilization stress to account for fluctuations in delinquency and occupancy over time.
• Step 5: Assess the cost to the transaction of operating its data centers. Data centers require strong security systems and all critical mechanical, cooling, electrical, and network infrastructure, including back-up systems, needs regular maintenance. Typically, we model such expenses, and the cost of property management, as senior to liability interest. In addition, we make assumptions regarding the capital improvements necessary to maintain a facility's competitive position.
• Step 6: Estimate exposure to broader risks, such as construction or development risk. We also assess insurance coverage and forward starting leases.
• Step 7: Consider our analysis under these criteria alongside our wider analytical framework for rating structured finance securitizations. In particular, we assess operational and administrative risk, counterparty risk, and legal and regulatory risk using our existing approaches.

Criteria Framework

Step 1: Determining The Utility Score

The utility score measures the long-term competitive position of a given data center in its respective market, using the following scale: 5 (strong), 4 (above average), 3 (average), 2 (below average), 1 (low), and 0 (weak). A lower utility score could indicate that a property is more exposed to obsolescence risk, suggesting a shorter liquidation time horizon. By using a utility score, we address the potential for weaker properties to have an outsized credit effect on the overall performance and value of the portfolio over time. The score is typically informed by the following five utility attributes, each scored on scale from 5 (strong) to 1 (low):

• Location;
• Age of the facility, taking into account any major capital improvements;
• Cost of power;
• Power efficiency; and
• Redundancy.

We determine a utility score for each property by combining these subscores. The relative importance of the subscores varies for each area; to capture this, we apply a weighting that ranks the relevance of the different attributes to the property's long-term competitive position. Our weightings may differ from market to market. For example, location may play a lesser role in marketing data centers in smaller jurisdictions or regions, given that distances between major cities and business centers are likely to be shorter. Similarly, where energy prices are high, we may believe power efficiency is likely to have a stronger-than-usual impact on a data center's long-term competitive position. We may assign an overall score of zero to certain properties, in rare instances. For example, a utility score of zero can be used for a property that has little to no future value, due to very high vacancy rates or poor property condition.

Appendix 3 includes our current weighting of the attributes. Periodically, we update the weighting of the attributes to more-accurately differentiate properties as market conditions change and the data center industry evolves. In addition, if the characteristics of the collateral pools that back data center securitizations shift, we may introduce new attributes or remove attributes and update the weighting. We develop weightings for new jurisdictions and markets as the asset pools expand. Our aim is to identify the factors affecting the likelihood that a data center would maintain its value over time.

Where relevant, we take interconnectivity into account when determining a given property's utility score. Capacity at data centers that exhibit significant physical interconnectivity and operate as internet exchange points (also known as carrier hotels) is highly sought after. Typically, carrier hotels are assigned a utility score of at least 3.

Utility attributes

Appendix 3 includes our current benchmark assumptions for the five standard utility attributes within the U.S. and Canada. As asset pools expand, it will be updated to add our benchmark assumptions for additional jurisdictions and markets. The property level analysis we perform could lead us to assign an attribute score that differs from the benchmark assumption. For example, our view of the location score may be bolstered if a facility in a tertiary market nevertheless benefits from strong demand and a diversified tenant base. When we are unable to assign a subscore to a given property, we will benchmark the score against similar facilities in similar markets.

We set our benchmark assumptions so that a subscore of 3 represents the average for a given market. The benchmark assumptions are updated periodically in light of market conditions, including long- or short-term trends in the property market, changing patterns of supply and demand, evolving efficiency standards, or the relative cost of power (for the current assumptions, see appendix 3). If the characteristics of the collateral pools that back data center securitizations shift, we may add or remove attributes to/from the list and establish benchmark assumptions.

Location:  We consider location key to the long-term utility of a data center because it affects the speed of connections to other users. The time it takes for a data packet to go from one place to another, known as the latency, largely depends on the distance between the two locations. In particular, latency-sensitive use cases require closer proximity to the end user. Proximity is therefore essential to our assessment of the location subscore.

The need for low-latency cross-connections drives the development of major data center hubs. This creates a clustering effect that supports self-perpetuating ecosystems of data centers. The size of the cluster (or market) is a significant factor when assessing location as a utility attribute.

Other factors that support a higher location subscore include:

• Ability to act as, or proximity to, an internet exchange point;
• Access to robust and reliable sources of power;
• Low risk of natural disaster; and
• Favorable tax incentives.

Age of build:  At a given price point and location, new properties with modern designs typically have a competitive edge over older facilities. Many older data center facilities were converted from telecom buildings. These properties were not built to accommodate today's need for high power and uninterrupted services, making them less attractive to tenants. By contrast, modern purpose-built data centers tend to be more efficiently designed and are better able to support the industry's high-density computation and cooling needs. Therefore, we tend to assign better age-of-build subscores to newer properties.

That said, if a property has undergone major capital improvements--such as structural changes that improve the integrity of the building, updates to cooling systems, or infrastructure improvements that reinforce redundancy--we may assign a better subscore. To help us determine an appropriate age-of-build subscore, we consider external assessments, such as appraisals and property condition reports. Where facilities are built out in phases, we will consider the average age of the building.

Cost of power:  Data centers are power-hungry facilities. Just cooling the infrastructure takes a significant amount of power, on top of that needed for daily operations. In addition, data centers employ redundant systems to ensure that they can maintain consistent, reliable services. Major data center markets therefore tend to be located in regions where electricity is relatively inexpensive and the power supply is robust. Low-cost electricity is critical to managing operational expenses, and to the value proposition for tenants.

We assign higher cost-of-power subscores to facilities that have access to low-cost power, because it offers them a robust competitive advantage. Energy pricing trends at the national and state level inform our cost-of-power benchmark assumptions. The nature of the energy sources can also affect our cost-of-power score, although this is typically a secondary consideration. Nevertheless, if the cost of power is the same, facilities that have a greener energy matrix rank higher. Many tenants favor green energy because it supports their net emission targets, which benefits data centers that run on green power.

Power efficiency:  Tenants typically pay for the cost of the power they use, making the power efficiency of a property very visible to them, especially for large-scale deployments. In addition, power efficiency is an important element in the global transition to clean energy. Both aspects may motivate tenants to factor power efficiency into data center deployment decisions.

The data center industry has adopted a metric to measure the energy efficiency of infrastructure--power usage effectiveness (PUE). PUE is a calculation of how much incoming power is utilized by IT equipment. For each facility, we consider its PUE, or equivalent measurement of power efficiency, in our evaluation of its power efficiency subscore.

Until properties ramp up to full occupancy, they may exhibit lower-than-targeted power efficiency--we take this into account when analyzing a property's overall power efficiency. We also consider the power needed for cooling, both now and in the future, when determining the power efficiency subscore. In part, this is to account for the trend toward cooling systems that use less water but more energy than the older evaporative cooling methods. Typically, we account for the efficiency level indicated by infrastructure design while considering actual performance.

Redundancy:  Tenants see data centers as mission-critical facilities and typically expect operators to maintain a continuous service, with no interruptions. Data center operators achieve this by implementing redundant systems for power and cooling, which would mitigate the effects of disruptive events, such as regional power outages.

A data center's level of redundancy affects its capacity to continue its operations--greater redundancy provides more security from disruption events. The level of operational redundancy is a key part of our assessment of a data center's utility, given the impact of service interruptions on tenants' operations. Therefore, data centers that have higher redundancy credentials receive higher subscores. Where facilities have a mix of redundancy features, our score will reflect the majority exposure.

In determining the level of redundancy at a specific data center, we consider the level of redundancy across critical components and third-party data.

Step 2: Property Liquidation And Timing

Property value

We generally determine property values using the expected-case value (S&P Value) described in our CMBS property evaluation methodology and the associated guidance (see related criteria and other related publications). This incorporates an analysis of the lease terms and the property's exposure to tenants rated 'BBB' and above. Subject to any relevant transaction covenants and the presence of longer lease terms, we may modify the credit given to investment-grade tenants in our property valuation, to allow for the longer time horizon in data center securitizations, compared with a typical CMBS transaction.

If a property's ground lease, including any renewals included in its contract, has a term that does not extend well beyond the legal final maturity date, we do not calculate a liquidation value for the property. Typically, we would only ascribe full valuation credit when the ground lease term--including renewals--exceeds 20 years past the legal final maturity of the rated liabilities. If the remaining term of the ground lease is less than 20 years past the legal final maturity of the liabilities, we will apply a discounted cash flow approach to recognize the potential reversionary rights of the data center to the ground owner. Over time, given the revolving nature of data center master trusts, a reduced buffer period between the maturity of the ground lease and that of the issued debt could lower the valuation credit ascribed to properties. In addition, if the property has a utility score of zero, we typically assume the property has a value of zero at liquidation, which occurs when the majority of the initial leases have expired.

Property value haircuts:  For each property, we estimate the sales proceeds by applying a rating-specific haircut to the S&P Value. The haircuts are derived from the values found in the relevant CMBS criteria (see related criteria section). For example, table 1 summarizes our haircuts for industrial properties such as data centers in the U.S. and Canada through the 'A' category.

Table 1

Property value haircuts for industrial properties
Rating category	Haircut as a percentage of S&P Value (%)
A	37.5
BBB	30.0
BB	17.5
B	5.0
The haircut is equal to 100% minus the value in "Rating Methodology And Assumptions For U.S. And Canadian CMBS," Sept. 5, 2012. Given the longer-term maturity profiles in data center securitizations, we do not include the benefit for diversification represented by the diversification tables. A linear interpolation is calculated for ratings that fall between categories.

If a portfolio exhibits significant property diversification (such as property count), we may apply a smaller haircut to the property value. Conversely, where portfolios exhibit significant property concentration (e.g., property count), and have long-dated legal final maturities, we may apply an increased haircut. Similarly, we may apply a larger property value haircut if the transaction is more dependent on liquidation of the property portfolio in the long term; for example, if it has a longer ARD period and minimal amortization.

Liquidation timing

In our analysis, we assume properties are liquidated based upon their long-term competitive nature and likely exposure to obsolescence risk. We use the property-level utility score to estimate the timing while considering the legal final maturity of the shortest-dated liability. Properties that have lower utility scores are liquidated earlier than those that have moderate or high utility scores (see table 2).

Table 2

Assumed time to liquidation
Utility score	Years
5	25.0
4	22.5
3	20.0
2	15.0
1	10.0
0	5.0
A linear interpolation is calculated for scores that fall between categories.

Because liquidating a property takes time, issuers need to begin the process before the shortest-dated liability matures (legal final maturity). In our analysis, we expect the process to begin at the sooner of our assumed time to liquidation (see table 2) or the start of our estimated disposition period (see table 3). The length of the disposition period in our model is typically one to three years and depends on the rating on the shortest-dated liability. For example, if the liability is rated 'A', our model assumes that all assets are sold at least 36 months before the first legal final maturity date. If the disposition period in the transaction documents does not start early enough (that is, the time before the legal final maturity is shorter than that shown in table 3), we may apply additional stresses while considering the maturity profile of outstanding liabilities. Where disposition periods are longer than those in table 3, we consider whether the property is likely to remain competitive until it is disposed of, given our time to liquidation assumptions (see table 2).

Table 3

Our estimated disposition period
Rating category	Months to maturity of shortest-dated liabilities
A	36
BBB	26
BB	18
B	12

Step 3: Revenue Assumptions

We account for cash flows from the in-place leases in our analysis. When a contract includes rent escalations over time, we typically step-up the initial lease rate--as defined in the contract--while considering inflation expectations. For usage-based contracts, we will consider historical utilization while accounting for contractual minimums and the transaction's exposure to such leases when determining lease revenues.

We stress cash flow from leases to account for potential delinquencies or vacancies and assume a recession that increases the stress and begins at the time of the analysis (see "Utilization stress" below). In addition, we assume that in-place leases mature no more than 20 years from the time of analysis for all utility properties, or 15 years if the property has speculative-grade tenants and a utility score of 2 or below.

We assume most properties are subject to new lease contracts before liquidation, although this is dependent on the remaining term of the initial lease. We assume that the term of each new lease will be the equivalent to the shorter of:

• The weighted-average original lease term for the pool, or
• Five years.

We may adjust these renewal assumptions to the extent we see significantly longer original leases terms or to address higher lease renewal terms.

Where a property has a utility score of 0, we assume it will not enter into replacement leases when the existing contract expires.

Each time capacity is leased to the same or a replacement tenant, we apply a haircut to the contractual rate. We consider the terms of the initial contracts alongside historical lease trends and the outlook for the relevant sector when determining rent escalations on future contracts related to the property portfolio.

At the time of analysis, if we expect the lease to be renewed only once more, we apply the whole haircut at that renewal. If we model two or more renewals, we interpolate intervening haircuts so that the lease rate gradually declines until the property is liquidated. The number of renewals modeled for a given capacity, and thus the size of each haircut, depends on our assumptions regarding the:

• Term of the new contract;
• The time to liquidation, which is influenced by the utility score.

At the final lease renewal, the cumulative effect of the lease rate haircuts is equal to the corresponding property value haircut for the relevant jurisdiction based upon the liability rating (see step 2), including concentration-related adjustments. Economic stress increases at higher rating levels--therefore, we project an increased decline in the contracted lease rate at higher rating categories.

In the case of properties that have significant exposure to a single tenant, we may size the potential for renewal risk by assuming a waiting period after the initial lease. In determining the prevalence of such risk, we will review the structure of the lease exposure. For example, we may view renewal risk as mitigated where a tenant exposure is represented by staggering lease expiration over a multi-year time horizon or a lease has sufficient contractual termination notification provisions. The length of a lag period will vary above the 'B' level based on the liability rating and the following factors (see table 4):

• Size of the exposure; and
• Location of the property.

We would apply assumptions at the lower end of the range for smaller exposures or those located in stronger market locations. Conversely, we would apply assumptions at the higher end of the range for larger properties or weaker market locations. We will account for other relevant attributes where relevant, which could lead us to adjust our assumptions beyond the below ranges. Structural mitigants, such as liquidity reserves and letters of credit, may be used to address the renewal risk.

Table 4

Lag period
Rating category	Waiting period (months)
A	12-15
BBB	9-12
BB	6-9
B	6
A linear interpolation is calculated for ratings that fall between categories.

Additional revenue

Although most data center operators view the leasing of capacity as their main revenue source, they have other revenue streams. These can include interconnection fees, managed services, marking up the cost of power, or "remote hands" support.

Because we view most of these additional revenue sources as more volatile than lease-related income, we may exclude some of them from our analysis or apply additional stresses. When determining the appropriate treatment, we consider the:

• Historical performance and longevity of each revenue source; and
• Potential impact of replacing the existing manager on the generation of additional revenue.

In our view, interconnection fees (cross-connections) are the most stable of the alternative revenue sources. Cross-connections offer tenants added value and contribute to the existing data center ecosystem, thus promoting tenant inertia.

Step 4: Utilization Stress

To account for fluctuations in delinquency and occupancy over time, we apply a utilization stress to each transaction's lease-related cash flows throughout its life. The stress increases during recessions, when we expect lease performance to deteriorate.

Base utilization stress

The base utilization stress incorporates the diversity of the tenant pool and lease tenor while considering the portfolio's historical performance and sector outlook. For usage-based contracts, we may consider a higher utilization stress to recognize the potential for increased volatility in lease payments over time. We determine base utilization at the property level and account for potential migration in tenant quality and concentration over time. For example, a property that has a single investment-grade tenant that has agreed a long-term lease is expected to exhibit lower cash flow volatility based upon contractual performance. That said, we recognize the renewal risk and may increase the utilization stress later in the transaction's life. Our experience to date suggests a range of 1%-3% is typical for wholesale business models, which experience higher utilization rates than retail business models. Without portfolio-specific performance data, the base utilization stress would depend on our expectations for the relevant sector. In transactions that do not feature advancing, we may increase our base utilization stress to address the potential effect of increased expenses and other costs on transaction liquidity. Finally, we apply rating-specific multipliers (see table 5) to represent the increase in cash flow volatility that we expect higher-rated liabilities to be able to withstand.

Table 5

Cash flow stresses by liability rating
Rating category	Utilization stress multiplier (x)
A	3.0
BBB	2.0
BB	1.5
B	1.0
A linear interpolation is calculated for ratings that fall between categories.

For pools where the majority of the exposure is to a single tenant, we increase the base utilization stress level.

Recessionary utilization stress

The stress reflects our expectations for cyclicality within the data center leasing industry. Our benchmark assumption is that there will be a four-year recession every 10 years, and that the first begins at the time of the analysis. To determine the total number of recessionary periods, we compare the assumed time to liquidation for each property (shown in table 2) with the disposition period and legal final maturity of the earliest-dated liability (see table 3) to establish when each property is likely to be sold. For example, if our analysis indicates that:

• The assets in the pool have 20 years to liquidation, and
• The shortest-dated liability matures in 25 years and is rated 'A' (implying that the disposition process should begin at or before year 22), then
• We model the transaction as lasting 20 years.

The example given above would have two recessionary periods: one beginning immediately (at the time of the analysis) and the second in year 11.

In a recession, we increase the utilization stress and further reduce collections. The base utilization stress serves as a floor during the recessionary period but is not applied alongside the recessionary utilization stress. The time taken for cash flows to revert to their earlier level (known as the downtime) varies by rating category (see table 6). We derive the appropriate utilization stress for a recessionary period using S&P Global Ratings' CDO Evaluator and the following inputs:

• The tenants' credit quality (see "Credit quality of the tenants" below);
• The tenants' industry;
• Tenants' domicile or that of guarantors, where relevant;
• The value of the lease contract; and
• The lease term.

Our initial recessionary analysis accounts for the credit quality of the in-place tenants. In our analysis, we model a reduction in tenant creditworthiness at each subsequent recessionary period to address the potential for tenant migration.

The utilization stress during a recession varies by rating category. We use the CDO Evaluator outputs, including the scenario default rate and typically the largest obligor supplemental test, to estimate the reduction in utilization during the recession. The CDO Evaluator output serves as a proxy for tenant performance during recessionary periods. We allocate the utilization stress across the recessionary period, typically by applying a front-loaded 40/30/20/10 annual curve and the downtime specified in table 6. For example, if the utilization stress is 50% for a 'A' level stress scenario, then we would reduce the utilization rate by:

• 20% (1.67% monthly) in year 1 for 12 months,
• 15% (1.25% monthly) in year 2 for 12 months,
• 10% (0.83% monthly) in year 3 for 12 months, and
• 5% (0.42% monthly) in year 4 for 12 months.

We may apply additional curves to address shifting economic conditions.

Table 6

Recessionary period downtime
Rating category	Months
A	12
BBB	9
BB	6
B	3

Tenant concentration:  For pools exposed to significant tenant concentration (for example, when there's event risk related to a low number of tenants), the recessionary utilization stress has a floor, which is based on the weighted average rating on the tenants in the pool (or another measure of their creditworthiness) and the rating category of the liability (see table 7). The stronger the average credit quality of the in-place tenants, the lower the utilization stress floor across all rating categories.

Table 7

Utilization stress floor for pools that have significant tenant concentration
	--Rating category--
Average tenant rating	'A' stress (%)	'B' stress (%)
AAA	15.0	5.0
AA	22.5	7.5
A	30.0	10.0
BBB	45.0	15.0
BB	60.0	20.0
B	75.0	25.0
Linear interpolation is used between the 'A' and 'B' stresses.

Credit quality of the tenants:  In our analysis of the utilization stress during a recession, tenant creditworthiness is typically a material driver of our credit analysis. We may consider various measures of creditworthiness in relation to the tenants. For example, we may use our ratings on the tenants or another measure of creditworthiness as a proxy for lease performance, while considering the presence of lease guarantees. Our surveillance includes rerunning this analysis to incorporate an up-to-date view of the tenants' creditworthiness.

For diversified pools of tenants, ratings from third-party credit agencies may also be considered as rating inputs, to a limited extent, if these are made available to us and updated on a timely basis. We do not use third-party ratings if tenant exposures are a substantial driver of our ratings analysis.

If we have not assigned a rating or other measure of creditworthiness to a given tenant and the pool is diversified (and no third-party ratings have been made available to us), we use a 'CCC' rating input in CDO Evaluator.

Step 5: Expenses

The transaction usually takes responsibility for expenses related to data center operations, maintenance, taxes, insurance, and electricity (wholesale facilities usually pass the cost of electricity through to the tenant). We estimate these expenses based upon historical performance data, asset level reports and appraisals, and the outlook for the relevant sectors. Typically, we model these expenses as ranking senior to liability interest in the waterfall.

In some cases, tenants sign a triple-net lease for turnkey facilities. Unless the transaction documents include specific eligibility parameters that require all future tenants to sign triple-net leases, we gross up both the revenue and expenses and model the leases as gross lease contracts, using our gross lease assumptions. This captures the risk that when the initial triple-net lease expires, the issuer may become responsible for the property's maintenance expenses because the new lease is a gross lease.

Given how integral the property manager's role is to the transaction's ongoing operations, we consider manager fees to be a senior expense, irrespective of its place in the transaction waterfall.

Where usage is reduced, we use the assumptions that underpin our base utilization stress to reduce variable expenses. Power efficiency may suffer at lower usage levels, which may affect our estimated variable expenses.

In addition, we use the utility score to estimate discretionary capital spending on upgrading data center facilities to maintain their relative competitiveness in the market (see table 8). For example, operators may improve the cooling infrastructure so that the facility can accommodate higher-density computation. Where the utility score is lower, we expect that keeping the facility competitive requires a larger investment. The landlord of a lower-utility property also has less ability to increase rental rates to pass through the cost of investments to the tenants. We assume that landlords would not make capital investments in properties that have a utility score of 1 or below, given that we assume a shorter term for such properties (see table 2). The assumptions in table 8 may be modified to accommodate regional or portfolio considerations and may be updated to address evolution in the data center industry.

Table 8

Capital expenditure on improvements
	Monthly expenditure per kilowatt ($)
Utility 5	3.0
Utility 4	4.5
Utility 3	7.0
Utility 2	11.0
Utility 1	0.0
A linear interpolation is calculated for scores that fall between categories.

Step 6: Broader Considerations

Construction risk

Properties backing data center securitizations have a fully constructed shell and core, and the existing data halls are typically occupied, so that the transaction generates a cash flow from at least one lease contract. If the shell and core are still under construction or a significant percentage of the data halls have yet to be completed, we consider what structural mitigants are in place. In the past, examples of mitigants have included letters of credit from financial institutions, sized to pay down the liabilities if construction is not completed within a specified timeframe. Mitigants characterized as nonfinancial obligations (defined in appendix 2) are not eligible to offset construction risk. For example, construction-related repurchase obligations from related parties, including the tenants, would not be eligible because our view of the creditworthiness of these entities does not capture the performance risk associated with the obligations. Similarly, hell-or-high-water leases would not mitigate construction risk.

For unmitigated, or insufficiently mitigated, construction risk, we exclude any cash flows from properties that are still under construction from our analysis, while including the securitization's exposure to the costs and potential liabilities of construction.

Forward starting leases

Data center facilities are often built out in stages, so that additional data halls come online in response to tenant demand. In many cases, properties use forward-dated leases to reserve the space for tenants. To determine whether to give credit to future sources of income such as forward-dated leases, we consider the estimated time to occupation, the proportion of the facility that is fully complete, and the tenants' credit strength. Typically, we would not include future sources of income if it exceeded 15% of the income from a specific property or 10% from the entire pool at closing. In addition, we would expect to see structural mitigants, such as reserves, to offset the risk of delay and the potential for tenants to walk away.

Capacity limitations

Data centers may lease out more power capacity than is available at the property level, based on the manager's expectations regarding usage levels. It is not uncommon to find overcommitments that significantly exceed the maximum power allotment for a given property. In such cases, we examine the nature of the underlying contracts when determining credit for such exposures, including whether they are fixed or variable (that is, whether the revenue is usage-based). We also consider historical information when determining the level of credit, if any, to allocate to such commitments.

Property insurance

In our view, sufficient property insurance to protect against losses implies a market-standard policy from a rated property/casualty (P/C) insurer. Such a policy typically includes general liability coverage and protection against damage to the property by fire. The policyholder should be able to claim at least the cost of fully rebuilding the property, in case of extensive damage. General liability coverage also includes the cost of replacing equipment and reinstating any physical improvements that had been made to the building, as well as costs associated with interruptions to the operation of the data center. Generally, we recognize insurance cover only if the provider--or all providers within a syndicate--is rated at least 'BBB-'. In limited cases, self-insurance will be considered for highly rated tenants when there's evidence of structural mitigants to negative ratings migration alongside consequences for failure to replace.

We may assign lower ratings if we consider the insurance coverage and related deductibles or the creditworthiness of insurance providers to be insufficient, and the transaction also lacks sufficient excess credit enhancement to offset the risk.

Issuers may maintain an umbrella policy to compensate for any insurance deficiencies at the individual property level. When evaluating a transaction's overall insurance provision, we analyze the minimum coverage and rating requirements for both umbrella- and property-level policies. Where the transaction does not own the real estate, our analysis focuses on the general liability and business interruption insurance contracted by the transaction.

Where transaction documents include minimum coverage and rating requirements for insurance policies and providers, we analyze those standards, rather than the policy in effect and the rating on the insurer providing coverage. Our approach assumes that issuers commit to replacing any insurance provider that no longer meets the minimum rating specified in the transaction documents with a provider that meets our minimum rating requirements.

In general, data center transactions are concentrated by both property and tenant. For portfolios where the risk associated with insufficient insurance coverage is greater because of concentration by property or tenant, we typically give credit for insurance coverage only where we rate the provider or can assess its creditworthiness. This applies to coverage for all the risks we consider appropriate. However, if property portfolios are not concentrated by property and tenant, whereby insurance risk would typically not be a material driver of the credit analysis, we may consider ratings from third-party credit rating agencies (as long as these are made available to us and updated on a timely basis), even when considering the creditworthiness of providers of umbrella insurance policies.

Specialty insurance:  To protect properties and cash flows, some transactions may require additional cover. For example:

• Environmental insurance, or an equivalent mitigant, may be required, depending on the nature of any environmental conditions (including those identified through Phase I or Phase II environmental site assessments) and the estimated cost of remediation.
• Business interruption insurance covering at least 12 months, or its equivalent, is generally required.
• Policies that cover natural catastrophes, such as floods and earthquakes, are generally required at the property level, depending on location.

Variable funding notes (VFNs)

Liability structures that allow the outstanding principal balance to be increased through future drawdowns can increase the total outstanding debt of a master trust. In evaluating such structures, we use the maximum amount of debt allowed for a given program, unless a structural mitigant is present.

Contractual considerations

When applying our utilization assumptions, we consider the risks related to clauses that allow the termination of the lease, such as a change to the property manager. We may adjust utilization stress to address variations in contractual risk, including breakage fees, especially for concentrated tenant exposures.

Leased data centers

Where the data center operator is itself a tenant--that is, it has leased some or all of the buildings in the portfolio from a third party--it assigns to the issuer the real estate leases that enable it to provide space and power to its own tenants. To incorporate this lease exposure, we typically assume that the issuer is able to renew the initial lease on the property based on in-place contractual renewal terms up to the utility-informed time to liquidation in table 2. When determining property lease rates upon renewal, we'll consider the extension terms under lease contracts. We also apply our legal analysis to determine whether the leases would survive a sale of the underlying building or buildings.

Ground leases

Where the issuer owns the land but not the data center and has a sufficiently long ground lease term (including renewals) that extends beyond the final maturity of the transaction, we will use the utility score associated with the related data center to analyze both the proceeds arising from the sale of the land (the liquidation value) and the ground lease payments.For the avoidance of doubt, in this criteria framework we will not ascribe value to land where there is exposure to construction risk. For utility scores of 3 or above, we will assign a liquidation value using the approach in our CMBS property evaluation methodology and the associated guidance (see related criteria and other related publications) while applying the liquidation timing in table 2. At the same time, we will give credit to the ground lease payments while considering lease rate decline assumptions and utilization stresses in-line with those used to analyze data centers in Steps 3 and 4. For utility scores below 3, we will not ascribe liquidation value or consider credit to ground lease payments.

Analytical adjustments

Where we see heightened sensitivity to collateral characteristics such as geographic exposure, we may add additional stresses or modify some of our assumptions--for example, the disposition period in table 3. Our approach to assigning and monitoring ratings also includes additional qualitative analysis based on transaction-specific factors, such as structural subordination or the amortization profile. We may adjust the ratings implied by our cash flow results, depending on these qualitative factors, by up to two notches. For instance, the rating may be adjusted downward for structural subordination. Any such adjustment would result from our overall analysis and an assessment of factors such as:

• Leverage at a given rating level (based on our stressed valuations);
• Available cushion, relative to peers; and
• Our forward-looking outlook for the sector.

For master trusts, we consider our expectation of the vehicle's all-in leverage. In such cases, our adjustments may exceed two notches downward.

Where a transaction's portfolio has low property diversification (property count) and very few tenants, our view of the creditworthiness of those tenants, or related lease guarantors, act as a cap on our rating on the transaction when the weighted average utility score for the property portfolio below 3. In such cases, we consider the applicability of the largest obligor test when determining the recessionary period utilization stress.

The maximum potential rating under these criteria is typically limited to 'A+'. Given the rapid changes to the data center industry, properties within existing portfolios could be exposed to obsolescence risk. As a result, it is difficult to project certain key variables (such as property value and contractual rates) over longer periods with the level of certainty that we would require for any ratings higher than 'A+'. Nevertheless, where transactions exhibit lower exposure to such long-term risks--including shorter-dated legal final maturity profiles--our ratings may exceed 'A+'.

Surveillance  In our surveillance analysis, we may adjust our stress scenarios to incorporate transaction performance. As transactions become more seasoned, we may gain data on how variables such as contractual and utilization rates have performed under recessionary conditions or other stresses. After the ARD, transactions typically accelerate amortization, which reduces leverage across the master trust. We may limit upgrades in such cases, to recognize the structural subordination of existing classes of notes. In addition, our review of property valuations is informed by the external assessments that are provided to us periodically. We will consider additional market value decline assumptions based upon the provisioning of such assessments.

Step 7: Application Of Other Criteria

In our article "Principles Of Credit Ratings," published Feb. 16, 2011, we set out our analytical framework for rating structured finance transactions, which has the following five key areas:

• The credit quality of the securitized assets;
• Payment structure and cash flow mechanics;
• Operational and administrative risk;
• Counterparty risk; and
• Legal and regulatory risk.

The prior steps focus on the credit quality of the securitized assets and aspects of the payment structure and cash flow mechanics that are specific to this type of transaction; further information about our approach to cash flow analysis can be found in our global cash flow criteria, including our approach to foreign exchange risk. We consider legal risk within our wider analytical framework (see "Legal Criteria: Structured Finance: Asset Isolation And Special-Purpose Entity Methodology," published March 29, 2017). In addition, for single- and multi-jurisdictional structured finance transactions that are rated above the relevant sovereign(s), we apply "Incorporating Sovereign Risk In Rating Structured Finance Securities: Methodology And Assumptions," published Jan. 30, 2019.

Here, we consider our analysis of the third and fourth areas in our overall framework.

Operational risk in data center transactions

Under our "Global Framework For Assessing Operational Risk In Structured Finance Transactions," published Oct. 9, 2014, we assess whether the structure explicitly describes and assigns responsibilities that may weigh on the rating if they are not performed as agreed. Table 9 includes some examples of how we apply our operational risk criteria to specific aspects of data center transactions and the key transaction participants within these securitizations.

Table 9

Applying our operational risk criteria
Role	Factor	Assessment
Manager	Responsible for marketing data center space, collecting lease cash flows, operating facilities, maintaining sufficient insurance, paying property taxes. In some transactions, the manager is also responsible for advancing.	We assess whether the manager can comply with its obligations according to the property management agreement.
Servicer	Responsible for reviewing and approving operating expense budget, foreclosing on mortgage lien when needed. In some transactions, the servicer is also responsible for advancing.	We assess whether the servicer can comply with its obligations according to the servicing agreement.

The issuer, as landlord, needs to remarket properties to find new tenants at least once and probably multiple times throughout the transaction's life. It is also responsible for disposing of the assets it owns. This aspect introduces several types of operational risk.

Our analysis of data center transactions focuses on the performance risk associated with remarketing the properties to subsequent tenants, and how future cash flows may vary. Additionally, our analysis assumes that the sale of properties is required to generate sufficient cash flows to pay off the rated debt in a timely manner when the properties are owned by the issuer.

Counterparty risk in data center transactions

We apply "Counterparty Risk Framework: Methodology And Assumptions," published March 8, 2019, to incorporate this risk in data center transactions (see table 10 for examples).

Table 10

Counterparty risk in data center transactions
Role	Factor	Assessment
Bank account provider	Bank account exposures and the commingling of funds may pose a risk, depending on which account is used for the deposit of lease payments and proceeds related to the sale of properties.	We review account ownership to confirm that relevant amounts are held in the trustee's name for the investors' benefit. We consider whether any other funds could be commingled in the accounts. We also review replacement provisions, where applicable.
Trustee	Acts as a back-up to the advancing entity.	We review the terms under which the trustee may advance funds to cover interest shortfalls and property maintenance, including how quickly it pays the advances.

APPENDIXES

Appendix 1: Scope Of The Criteria

The criteria apply to all new and outstanding data center securitizations where:

• The issuing special-purpose entity (SPE) i) owns the data center in fee (that is, owns it outright and in full, without limitations or restrictions), regardless of whether it owns or leases the land, ii) doesn't own the data center or land but generates cashflows by providing space and power to data center tenants (see section on leased data centers), or iii) in limited circumstances and when the total exposure to the transaction is minimal, owns the land but not the corresponding data center (acts as a lessor to the owner of a data center; see section on ground leases);
• Either construction is completed, or any construction risk present is fully mitigated (where there is unmitigated construction risk, our analysis excludes any benefits from the properties under construction while including the securitization's exposure to the costs and potential liabilities of construction); and
• The properties have a fully constructed shell and core, and the data halls are typically occupied and are generating cash flows under existing lease contracts.

We exclude from the scope of these criteria:

• Commercial mortgage loans backed by data centers. Such exposures are analyzed in line with our approach in the applicable CMBS criteria;
• Transactions that have unmitigated construction risk, including those that are being built out in stages and have yet to complete a significant proportion of the facility's data halls (such exposures will be analyzed in line with our project finance criteria); and
• Transactions exposed to greater, and more-specialized, operational risk than we would typically expect for an SPE issuer--regardless of whether those risks are managed by the SPE directly or outsourced to a key transaction party--will be analyzed in line with our project finance criteria. In specialized cases, there would likely be no replacement counterparty able to manage the operational risk.

We may use these criteria as a starting point for our analysis of transactions involving other digital infrastructure, such as cell towers. In such cases, we would make adjustments or include additional assumptions to reflect the assets' particular features. Considerations may include those regarding the utility of the assets, the transaction's revenue streams, and asset liquidation values.

Appendix 2: Glossary

The following glossary comprises terms frequently used in S&P Global Ratings' analyses of data center securitizations. The definitions are limited to how these terms apply when reviewing these types of securities.

Glossary
Term	Definition
Advancing	Any advances the manager or trustee may make with respect to the properties to cover payments, including lease payments, property taxes, and property insurance.
Anticipated repayment date (ARD)	In data center securitizations, the payment date when the issuer expects to repay the liabilities, typically before the legal final maturity date. This can be different for each tranche. After the ARD, additional interest may accrue and all available cash flows after fees and expenses are usually used to amortize the liabilities sequentially.
Carrier hotel	Data centers that exhibit significant physical interconnectivity and operate as internet exchange points.
Construction-related repurchase obligations	These obligations generally require construction counterparties or related transaction parties to pay a specified value if construction is not completed.
Dual power feed	Two independent power paths to each piece of computing equipment.
Fee or fee title	Ownership of real property without limitations or restrictions.
Gross lease	A lease agreement between a landlord and a tenant where the tenant is not responsible for all of the costs related to the leased asset. Exclusions typically include the payment of real estate taxes, as well as the cost of insuring and maintaining the leased asset.
Ground lease	A lease of land to a third party that allows them to construct, improve, or operate facilities on that land for a specified period.
Hell-or-high-water lease	A lease that does not allow the tenant to cancel the agreement, irrespective of events such as condemnation and casualty.
Hyperscale data center	A data center facility that typically caters to tenants in sectors such as cloud computing or big data.
N	An industry measure of redundancy, defined as the capacity needed to power, backup, and cool a facility at full IT load, without redundancy.
N+1	Adds sufficient redundancy to allow for a single component to fail or be shut down for required maintenance without causing a service interruption.
Nonfinancial obligations	These include settlements for a commercial dispute, employment benefit obligations and other labor obligations, lease payments (both finance and operating leases), and vendor or supplier obligations (see "S&P Global Ratings Definitions," Nov. 10, 2021).
Remote hands support	Enables tenants to delegate IT management and maintenance tasks to on-site technicians at the facility.
Retail colocation data center	A data center facility that typically caters to a diversified pool of enterprise clients.
Triple-net lease	A lease agreement between a landlord and a tenant where the tenant is responsible for all of the costs related to the leased asset, in addition to the rent paid under the lease. These responsibilities may include the payment of real estate taxes, as well as the cost of insuring and maintaining the leased asset.
Uninterruptible power supply (UPS)	A device that allows a computer to temporarily continue running when the primary power source is interrupted.

Appendix 3: Sector And Industry Variables

The sector and industry variables and associated details in this appendix are expected to be periodically updated and republished as market conditions warrant.

The assumptions listed below apply to the U.S. and Canada markets only, reflecting our current universe of rated transactions in the U.S. and Canada.

Table 11

Examples of market locations
Subscore	Region
5	Northern Virginia
5	Silicon Valley (California)
4	Atlanta
4	Austin, Texas
4	Central Washington State
4	Chicago
4	Dallas
4	Houston
4	New York City Metro Area
4	Montreal
4	Phoenix
4	Portland, Oregon
4	Seattle
4	Toronto
3	Columbus, Ohio
3	Las Vegas
3	Minneapolis
3	Quebec City, Canada
3	Raleigh and Durham, North Carolina
3	Sacramento
3	Salt Lake City
3	San Antonio, Texas
2	Cleveland
2	Nashville, Tennessee
2	Pittsburgh

Table 12

Age of build
Utility score	Building age adjusted for upgrades (years)
1	>20
2	>15 and <=20
3	>10 and <=15
4	>5 and <=10
5	<=5

Table 13

Cost of power
Utility score	Cost of power (Cents/kWh)
1	>25
2	>20 and <=25
3	>15 and <=20
4	>10 and <=15
5	<=10

Table 14

Power efficiency score
Utility score	Power usage effectiveness*
1	>2
2	>1.7 and <=2
3	>1.4 and <=1.7
4	>1.2 and <=1.4
5	<=1.2

Table 15

Redundancy
Utility score
1	Basic, non-redundant data center infrastructure
2	Redundancy on some components of data center infrastructure data center infrastructure, including power generation, UPS, and cooling
3	Concurrently maintainable data center infrastructure; redundancy on some components of data center infrastructure, including power generation, UPS, and cooling
4	Concurrently maintainable data center infrastructure; N+1 or above redundancy on critical data center infrastructure components, including power generation, UPS, and cooling
5	Concurrently maintainable data center infrastructure; dual power feed to the racks; N+1 or above redundancy on critical data center infrastructure components, including power generation, UPS, and cooling

Weighting the subscores

Table 16 shows the weighting used to determine the property-level utility score from the subscores in tables 11 to 15.

Table 16

Weighting used to determine the property-level utility score
Utility attribute	Weighting
Location	35
Age of build	25
Cost of power	15
Power efficiency	15
Redundancy	10

Appendix 4: Worked Example Using The Criteria

Hypothetical portfolio: a worked example

Here, to show how the credit and cash flow assumptions addressed our criteria would apply in practice, we have applied the criteria to a hypothetical data center securitization.

Table 17

Property input
Property	Owned/leased	Location	Age (year)	Cost of power ($/kWh)	Redundancy	PUE	S&P Value (Mil.$)
Data center 1	Owned	Northern Virginia	7	0.05	N+1 on most critical components	1.3	75
Data center 2	Owned	Silicon Valley	12	0.16	N+1 on most critical components	1.4	75
Data center 3	Owned	Salt Lake City	3	0.06	N+1 on most critical components	1.6	75
Data center 4	Leased	Cleveland	19	0.08	Dual power feed to the racks, and N+1 on all components	1.8	0*
*No S&P Value is assigned if the issuer does not own the real estate of the data center. PUE--Power usage effectiveness.

Table 18

Lease input
Property	Lease	Obligor	Tenant credit	Monthly revenue ($)	Tenor (months)	Contract value (Mil. $)	Annual escalation (%)
Data center 1	Lease 1	Obligor 1	AAA	250,000	120	30.0	1
Data center 1	Lease 2	Obligor 1	AAA	250,000	120	30.0	1
Data center 2	Lease 3	Obligor 2	AA	250,000	84	21.0	1
Data center 2	Lease 4	Obligor 3	A	250,000	60	15.0	1
Data center 3	Lease 5	Obligor 4	AA	100,000	72	7.2	1
Data center 3	Lease 6	Obligor 5	A	100,000	60	6.0	1
Data center 3	Lease 7	Obligor 6	A	100,000	48	4.8	1
Data center 3	Lease 8	Obligor 7	BBB	100,000	36	3.6	1
Data center 3	Lease 9	Obligor 8	BBB	100,000	24	2.4	1
Data center 4	Lease 10	Obligor 9	BBB	50,000	24	1.2	1
Data center 4	Lease 11	Obligor 10	BB	50,000	18	0.9	1
Data center 4	Lease 12	Obligor 11	BB	50,000	18	0.9	1
Data center 4	Lease 13	Obligor 12	BB	50,000	18	0.9	1
Data center 4	Lease 14	Obligor 13	BB	50,000	18	0.9	1
Data center 4	Lease 15	Obligor 14	BB	50,000	18	0.9	1
Data center 4	Lease 16	Obligor 15	CCC	50,000	12	0.6	1
Data center 4	Lease 17	Obligor 16	CCC	50,000	12	0.6	1
Data center 4	Lease 18	Obligor 17	CCC	50,000	12	0.6	1
Data center 4	Lease 19	Obligor 18	CCC	50,000	12	0.6	1
Data center 4	Lease 20	Obligor 19	CCC	50,000	12	0.6	1
Total				2,050,000		128.7

Step 1: Utility score

We derive property-level utility scores based on the weighted average of the attribute subscores. Appendix 3 includes our benchmark assumptions for the five standard utility attributes within Canada and the U.S.

Table 19

Deriving property-level utility scores
Property	Location (35%)	Age of build (25%)	Cost of power (15%)	Power efficiency (15%)	Redundancy (10%)	Utility Score
Data center 1	5	4	5	4	4	4.5
Data center 2	5	3	3	4	4	4.0
Data center 3	3	5	5	3	4	3.9
Data center 4	2	2	5	2	5	2.8

Step 2: Liquidation assumptions

We use the property level utility score in step 1 to inform our view on the assumed time to liquidation interpolated from table 2, subject to the transaction's maturity profile. Property value haircut is interpolated from table 1, or the relevant region-specific CMBS criteria.

Table 20

Assumed time to liquidation
Property	Utility score	Time to liquidation (years)	Property value haircut (%) ('A')	Liquidation recovery (Mil. $)
Data center 1	4.5	23.8	37.5	46.875
Data center 2	4.0	22.4	37.5	46.875
Data center 3	3.9	22.3	37.5	46.875
Data center 4	2.8	18.8	N/A	0
N/A--Not applicable.

Step 3: Base revenue

As shown in the lease input table above, the monthly aggregate monthly revenue is expected to be $2.05 million before applying utilization stress.

The space in each property in the pool is assumed to be rented out again, at a haircut to the initial lease rate and for a shorter lease term. For example, our hypothetical data center 1 has time to liquidation of 23.8 years as shown in step 2 and a liability tenor of 25 years. Because lease 1 has a remaining term of 10 years at closing, we assume the space is rented out three more times (to the same or different tenants) during the life of the transaction, each time with an assumed term of five years.

We align the reduction in lease rate with the property value haircut, or 37.5% for 'A' level stress. The starting revenue of $250,000/month in lease 1 would increase by 1% annually from contractual escalation but decline by approximately 14.5% upon expiration at year 10 and each subsequent renewal at years 15 and 20 (see table 21). Property is liquidated on month 283 based on a utility score of 4.5.

Table 21

Reduction in lease rate
Period	Monthly rent including escalation ($)	Cumulative lease rate decline (%)	Monthly rent after lease rate decline ($)
Month 1	250,000	0	250,000
Month 121	276,156	14.5	236,110
Month 181	290,242	26.9	212,168
Month 241	305,048	37.5	190,655

Step 4: Utilization stress

Assuming a 3% base utilization stress rate, the 'A' base utilization stress of 9% is applied in this example. We used CDO Evaluator to estimate utilization stress during recessionary periods. In this example, the 'A' scenario default rate is 16.32% considering the largest obligor supplemental test. We apply an annual default timing curve of 40/30/20/10 during the recession, spread across the 12 months evenly. The time taken for cash flows to revert to their earlier level is given in table 6; for a 'A' rated liability, it would be 12 months. Recessions are assumed to start in month 1, year 11, and year 21. Because the base utilization stress exceeds the recessionary utilization stress in this example, we apply a flat 9% haircut to the base revenue in step 3.

Step 5: Expenses

We use the property level utility score in step 1 to inform our view on the assumed capital improvement costs from table 8. The expense is subtracted from the revenue monthly after applying the utilization stress in step 4. This is in addition to the ordinary operating expense and maintenance capital expenditure budgeted for the transaction.

Table 22

Capital expenditure on improvements
Property	Utility score	Capital expenditure on improvement ($/kW/month)
Data center 1	4.5	3.75
Data center 2	4.0	4.63
Data center 3	3.9	4.75
Data center 4	2.8	8.00

Additional risk analysis

This example demonstrates how applying our credit and cash flow assumptions affects the timing and amount of cash flows from leases and property liquidations. These include assumptions regarding liquidation timelines, lease rates, lease tenor, expenses, and utilization.

To derive the rating, we then assess the likelihood of timely payment of the rated debt, based on the transaction structure alongside the application of our global cash flow criteria. Finally, we consider the legal, counterparty, and operational risks; these are covered in step 7.

KEY FEATURES

The key features are:

• To introduce our property utility metric, which is used to determine the credit risk associated with the pool of data center-related assets. This concept informs our view of the timing of asset liquidations and the cost of maintaining and upgrading facilities during the life of a transaction.
• To more-closely align our property value haircuts with those in the commercial mortgage-backed securitizations (CMBS) criteria listed in the Related Criteria section. By applying these assumptions, we address the similarity in characteristics between the properties backing CMBS and those backing data center transactions.
• To derive lease rate assumptions from our CMBS property value haircuts, thus linking our estimated income from property sales to our analysis of how lease income is likely to change over the longer term.
• To deploy a base utilization stress to leases to capture the impact of any cash flow reductions during the term of the initial lease, or beyond. This stress increases during recessionary periods, when cash flow is more likely to come under pressure; for example, because of reduced occupancy or late-paying tenants.
• To incorporate our analysis of the insurance risk pertaining to the underlying property portfolios into our assessment. Our analysis includes both minimum coverage levels and minimum rating thresholds for insurance providers.

Given the evolving nature of the data center industry, any long-term predictions about property value and contractual rates are too uncertain to rely on above a 'A' category stress; therefore, we typically cap our rating under these criteria at 'A+'. Nevertheless, when transactions exhibit lower exposure to such long-term risks--including shorter-dated legal final maturity profiles--our ratings may exceed the 'A' category.

IMPACT ON OUTSTANDING RATINGS

We anticipate that:

• Approximately 75% of outstanding ratings would be unaffected by the proposal; and
• Approximately 25% of outstanding ratings could be lowered by up to two notches.

The effect of the criteria depends on the interplay of multiple factors, including the:

• Utility of the properties within the pool;
• Changes to the haircuts that apply to cash flow and property values;
• Transaction's asset disposition period; and
• Timing of property liquidations.

This impact analysis is intended to serve as a broad, directional guide to the possible rating outcomes.

RELATED PUBLICATIONS

Fully superseded criteria

• Criteria | Structured Finance | ABS: Methodology And Assumptions For Rating North American Single-Tenant Real Estate Triple-Net Lease-Backed Securitizations, March 31, 2016

Related criteria

• Environmental, Social, And Governance Principles In Credit Ratings, Oct. 10, 2021
• Global Framework For Payment Structure And Cash Flow Analysis Of Structured Finance Securities, Dec. 22, 2020
• Global Methodology And Assumptions For CLOs And Corporate CDOs, June 21, 2019
• U.S. Structured Finance Asset Isolation And Special-Purpose Entity Criteria, May 15, 2019
• Counterparty Risk Framework: Methodology And Assumptions, March 8, 2019
• Incorporating Sovereign Risk In Rating Structured Finance Securities: Methodology And Assumptions, Jan. 30, 2019
• Foreign Exchange Risk In Structured Finance--Methodology And Assumptions, April 21, 2017
• Structured Finance: Asset Isolation And Special-Purpose Entity Methodology, March 29, 2017
• Rating Methodology And Assumptions For Global CMBS, April 6, 2015
• Global Framework For Assessing Operational Risk In Structured Finance Transactions, Oct. 9, 2014
• Rating Methodology And Assumptions For Japanese CMBS, Jan. 22, 2014
• European CMBS Methodology And Assumptions, Nov. 7, 2012
• General Criteria: Criteria For Assigning 'CCC+', 'CCC', 'CCC-', And 'CC' Ratings, Oct. 1, 2012
• Rating Methodology And Assumptions For U.S. And Canadian CMBS, Sept. 5, 2012
• CMBS Global Property Evaluation Methodology, Sept. 5, 2012
• Principles Of Credit Ratings, Feb. 16, 2011
• Methodology For Servicer Risk Assessment, May 28, 2009

Other related publications

• S&P Global Ratings Definitions, updated from time to time
• Default, Transition, and Recovery: 2021 Annual U.S. Corporate Default And Rating Transition Study, May 11, 2022
• Credit Rating Model: CDO Evaluator, May 8, 2020
• Guidance: CMBS Global Property Evaluation Methodology, March 13, 2019

This article is a Criteria article. Criteria are the published analytic framework for determining Credit Ratings. Criteria include fundamental factors, analytical principles, methodologies, and/or key assumptions that we use in the ratings process to produce our Credit Ratings. Criteria, like our Credit Ratings, are forward-looking in nature. Criteria are intended to help users of our Credit Ratings understand how S&P Global Ratings analysts generally approach the analysis of Issuers or Issues in a given sector. Criteria include those material methodological elements identified by S&P Global Ratings as being relevant to credit analysis. However, S&P Global Ratings recognizes that there are many unique factors / facts and circumstances that may potentially apply to the analysis of a given Issuer or Issue. Accordingly, S&P Global Ratings Criteria is not designed to provide an exhaustive list of all factors applied in our rating analyses. Analysts exercise analytic judgement in the application of Criteria through the Rating Committee process to arrive at rating determinations.

Analytical Contacts:	Jie Liang, CFA, New York + 1 (212) 438 8654; jie.liang@spglobal.com
	Ryan Butler, New York + 1 (212) 438 2122; ryan.butler@spglobal.com
Methodology Contacts:	Eric Gretch, New York + 44 20 7176 3464; eric.gretch@spglobal.com
	Nik Khakee, New York + 1 (212) 438 2473; nik.khakee@spglobal.com