Tech Industry Council Seeks Simplification of Cloud Transaction Regs

November 11, 2019

CC:PA:LPD:PR (REG-130700-14)
Room 5203
Internal Revenue Service
P.O. Box 7604
Ben Franklin Station
Washington, D.C. 20044

Dear Commissioner Rettig:

We appreciate the opportunity to comment on proposed regulations REG-130700-14, “Classification of Cloud Transactions and Transactions Involving Digital Content,” which update existing regulations classifying transactions involving computer programs and propose new rules for classifying cloud transactions, with the goal of providing taxpayers with additional guidance on classifying and sourcing cloud transactions and transactions involving digital content.

These regulations are particularly welcome given the rapid growth in recent years of digital content and cloud services, which represent an increasingly important part of many of our companies' business models, as well as of our economy as a whole. Indeed, worldwide revenue from cloud services in 2018 totaled almost $183 billion, representing 27.4% growth from the previous year.¹ Revenue from digital content has also exploded in terms of both worldwide revenue and importance for companies of all kinds, from more traditional companies to the most innovative tech companies' business models.

We believe these to be the first regulations by a taxing authority to specifically address how cloud transactions should be classified, so we view this as a particularly exciting opportunity to weigh in. Accordingly, we respectfully recommend that Treasury consider the following revisions to improve the regulations and provide additional certainty while streamlining administrability.

§1.861-18 Classification of Transactions Involving Digital Content

Sourcing

The proposed regulations provide that electronic sales of copyrighted articles are considered to occur at the location of the download or installation onto an end-user's device used to access the content (“Download Location Approach”). In the absence of download or installation information, a sale is deemed to occur at the location of the customer based on taxpayer's sales data (“Sales Data Approach”). Treasury specifically requested comments about sourcing issues related to cloud and digital content transactions, including whether information about the user's location is practically available in a reliable form, and what administrable rules could be crafted to implement these approaches.

We believe that the Download Location Approach laid out in the proposed regulations would introduce unnecessary ambiguity, and would be too burdensome to administer in light of the information realistically available and the other, easier to administer options that could be used. Additionally, we believe that this approach may intersect unfavorably with various data privacy laws being enacted and proposed both inside and outside the United States.

With respect to creating ambiguity, the most readily available information that could be used to establish the customer's location at the time of the download — the geolocation corresponding to the device's IP address — may or may not be reliable in terms of either the customer's actual location or the location where the content will be accessed or used, and may be subject to interpretation. Indeed, a customer's IP address may not represent their actual location or the location where the copyrighted articles will actually be used for practical reasons — such as when the customer downloads content while traveling — or because the customer intentionally obscures their IP address and location using a virtual private network (VPN) or software to “spoof” their IP address and cause them to appear to be in a different city, state, or even halfway across the world. Additionally, if a customer is logged on through a business network, the data often only shows the location of the business's gateway, not the actual location of the user themselves. In light of these technical limitations, the most readily available information — the customer's IP address — may or may not be reliable and may or may not represent where the content is likely to be used.

Applying the Download Location Approach creates further complexity beyond just the difficulty of determining where the customer was actually located when the download took place, because as part of the same sale, the customer may download the same content multiple times, onto multiple devices, at different times that may extend even across multiple tax years. This presents the taxpayer with the difficult situation of reconciling conflicting information about locations where a download or multiple downloads of the same material may have taken place, and having to show which location is more accurate or reliable, or should be controlling. Moreover, as discussed elsewhere in these comments, a sale may not require the customer to download the content at all, and they may specifically choose not to do so in light of technical limitations or for reasons of convenience. In those circumstances, it is not clear how the seller should characterize the customer's location under the Download Location Approach.

With respect to taxpayer burden, we do not believe any taxpayers currently base the source of their sales on the geolocation of the IP address of the device used for download or streaming due to the difficulty in obtaining accurate information about the customer's actual location as described above, and the privacy issues discussed in the next paragraph. Updating systems to track that information in a way that is usable for tax reporting and/or income sourcing purposes would be both challenging and, as discussed above, not necessarily particularly meaningful or applicable to many sales. Additionally, tracking this information for tax reporting and/or income sourcing purposes may conflict with reporting for sales tax/transaction tax purposes, adding additional complication.

Finally, as mentioned above, there may be conflicts with data privacy laws created by collecting and using this information in this way. Many jurisdictions (both foreign and in the United States at the federal or state level) are enacting or proposing data privacy laws that may limit taxpayers' ability to retain IP geolocation data — whether or not that data is reliable — which would make auditing of this data by taxing authorities difficult or impossible.

Given all of these factors, we recommend revising the final rules such that taxpayers are not required to rely on the Download Location Approach as their first resort. Instead, we recommend using one of two alternatives.

As one option, Treasury could permit taxpayers to choose between the Download Location Approach and the Sales Data Approach rather than only allowing use of the Sales Data Approach if the Download Location Approach is not an option. As long as taxpayers use a consistent approach, we believe this would allow Treasury to significantly reduce the administrative burden for both the government and taxpayers without undermining the government's interests or creating loopholes.

In the alternative, we would recommend that Treasury revise the rules to allow sales of digital content to be sourced based on the customer's billing address (which we believe corresponds to the Sales Data Approach, assuming this information would be acceptable proof of the customer's location under that rubric). Where the customer is an individual, the customer's billing address is the most likely indicator of where they reside at the time of the sale and are likely to use the digital content. Additionally, the Sales Data Approach does not require sellers to collect additional information and allows for resort to information that is readily available, consistent, and not subject to interpretation. Additionally, there is minimal potential for abuse because the commercial reality is that it is difficult for a taxpayer to secure the result of a customer changing their billing address to create a more favorable outcome for the seller. As an unrelated party, the customer has few incentives to try to help the seller get a better tax outcome, and must also manage their own tax liabilities.

Allowing taxpayers to rely primarily on the address of the customer would be consistent with other parts of U.S. law, which already employ the purchaser's address as an indicator of where they are located. For example, the proposed regulations implementing the Foreign Derived Intangible Income (FDII) provision of the Tax Cuts and Jobs Act allow taxpayers to rely on the customer's shipping address to establish foreign use in small transactions (Prop. Reg. §1.250(b)-4(c)(2)(ii)). Moreover, in the information and reporting context, payors are able to use a payee's address under certain circumstances to substantiate the status of the payee.²

Limiting Taxpayer's Analysis to the First Sale to an Unrelated Party

Another point that we believe it would be helpful to clarify is that the taxpayer's analysis of the place of download or installation, or the customer's location, should not extend beyond the first sale to an unrelated party. It would be overly burdensome to require the taxpayer to engage in further efforts to obtain information about the location of the ultimate end-user, and that information is not likely to be readily available — and may not be available at all. Additionally, retention of such data may violate data privacy laws, making it difficult or impossible to perform audits.

Accordingly, we recommend amending §1.861-18(f)(2)(ii) to more clearly state that for purposes of §1.861-7(c) only, copyrighted articles transferred through an electronic medium are sourced to the billing location of the first unrelated purchasing entity. We would, however, recommend including an election for transfers to instead be sourced to the primary location of the individual user of the copyrighted article (for example, the location of an employee using a license purchased by their employer), if the taxpayer has access to that information and chooses to use this sourcing approach. We believe that supporting information that, if accurate, would allow the taxpayer to reasonably conclude the primary location of the individual user should be considered sufficient under this election.

The following examples show how the proposed clarification and election would be applied.

Example 1

Facts.

Corp A is located in Country X and produces software in Country X. Based on the facts and circumstances, each software license produced by Corp A is inventory property. The software licenses are copyrighted articles under § 1.861-18, and the copyrighted articles are delivered electronically. Corp A sells software licenses to Distributor D. Distributor D, located in Country X, sells software licenses to Reseller R for resale to end customers. Reseller R is located in Country Y. Corp B purchases Corp A software licenses from Reseller R. Corp B is located in Country X, with 50% of its employees located in Country X and 50% located in Country Z. Corp A, Distributor D, Reseller R, and Corp B are unrelated.

Analysis.

Corp A. Because Corp A produces the software in Country X, and each copyrighted article of the software is treated as inventory property for the purposes of section 863, Corp A sources the income from the sale of these copyrighted articles under section 863(b) to Country X.

Distributor D. Because Distributor D sells copyrighted articles and delivers them electronically, D must source its income from the sale of copyrighted articles transferred digitally under the amended rule above. Accordingly, Distributor D sources its income by reference to the billing location of Reseller R. Reseller R's billing location is in Country Y, so Distributor D sources its income to Country Y. Distributor D does not make an election to source its income based on the location of the ultimate end user.

Reseller R. Similar to Distributor D, R sources its sale of copyrighted articles transferred digitally under the amended rules. Reseller R properly sources all of its income from the sale to Corp B to Country X.

Example 2

Facts.

The facts are the same as in Example 1, except that Reseller R makes an election to source its income based on the location of the ultimate end users of the copyrighted articles it sells.

Analysis.

The results are the same as in Example 1, except that Reseller R sources 50% of its income from Corp B to Country X and 50% to Country Z.

Example 3

Facts.

The facts are the same as in Example 1, except that Corp A does not produce software. Corp A purchases software from an unrelated party located in Country X. Corp A and Distributor D are related parties.

Analysis.

The results are the same as in Example 1, except that Corp A must look to the location of the first unrelated purchaser for the purposes of sourcing its income. Because Corp A and Distributor D are related parties, and Corp A does not make a sourcing election, Corp A sources its income to Country Y based on the billing location of Reseller R.

Example 4

Facts.

The facts are the same as Example 1, except that Corp B has two billing locations: Corp B's billing office in Country X purchases software for employees located in Country X; and Corp B's billing office in Country Z purchases software for employees located in Country Z.

Analysis.

The results are the same as in Example 1, except that Reseller R sources its income from its sales to Corp B according to the location of the Corp B billing office. Because the Country X billing office purchases software for the employees located in Country X (50% of Corp B's total employees), Reseller R sources 100% of the income from its sales to the Country X billing office to Country X. Because the Country Z billing office purchases software for the employees located in Country Z (50% of Corp B's total employees), Reseller R sources 100% of the income from its sales to the Country Z billing office to Country Z.

Example 5

Facts.

The facts are the same as Example 4, except that Reseller R makes a sourcing election under the proposed rule above.

Analysis.

The results are the same as in Example 4. Because the Corp B billing offices in Country X and Country Z purchase software only for the employees located in those countries, Reseller R's sourcing under the election is the same as in Example 4.

§1.861-19 Classification of Cloud Transactions

Classification of Cloud as “Services”

Cloud transactions are, as defined in the regulations, transactions through which a person obtains on-demand network access to computer hardware, digital content, or other similar resources. The proposed regulations are designed to provide rules for classifying cloud transactions as either a “lease” or “services” based on the facts and circumstances. In response to Treasury's direct request for comments, we agree with the proposed regulations that cloud transactions are not appropriately classified as a “license.”

We believe that the examination of the facts and circumstances of these transactions will lead to the conclusion that they are services in every case, as is true of the examples provided in the proposed regulations. Accordingly, to improve both administrability and consistency, we would recommend that Treasury revise the final regulations to conclude that based on the facts and circumstances framework, all cloud transactions are services transactions.

Treating all cloud transactions as “services” would be simpler for taxpayers and for the government, and would maximize consistency and predictability without creating avenues for abuse. Additionally, given that Treasury is the first taxing authority worldwide that we are aware of to provide guidance on the character of cloud transactions and is thereby in a position to lead on establishing an international consensus on the treatment of cloud transactions, we believe that defining cloud transactions as “services” would set an excellent precedent globally, thereby creating additional certainty for U.S. multinationals.

Alternatively, we recommend updating the relevant factors based on the realities of the context in which cloud transactions exist. Specifically, we would recommend the following adjustments.

Integrated Nature of the Operation: Given that a number of the elements of the analysis of whether a cloud transaction is a lease or provision of services focus on the physical property itself and where it is located — which is not the essence of a cloud transaction — we believe that the element examining whether the property is a component of an integrated operation should be expanded upon and given appropriate weight. While the regulations seek to analyze the nature of cloud transactions, the nature of cloud services offerings is underpinned by the fact that they are frequently part of an ecosystem made up of hardware and software, with interrelated services offerings that complement each other and form an infrastructure that can also provide access to the offerings of other vendors. Service offerings including compute capacity, applications, and infrastructure are frequently dependent on each other, both from a seller and a customer perspective. A customer might decide to use a particular seller's services because they offer compatibility with other offerings, or in fact, a particular offering might serve as a portal to the overall ecosystem and may not be compatible with another provider's offerings at all. In light of this interrelatedness, we recommend expanding this particular factor to more fully acknowledge the range of responsibilities a provider might have beyond simply maintaining and updating the property.

Physical Possession by Customers: Cloud computing may include multiple infrastructures in different locations, which are connected to provide the desired service to the customer. For instance, hybrid cloud — one type of cloud computing that is growing rapidly — involves linking one cloud provided and administered by a third party with a private cloud accessible only by the customer to allow for added privacy and security of data and resources. The infrastructure supporting the private cloud might be physically located on the customer's premises, but owned, operated, and managed by the third-party provider, with the customer being not permitted to move, alter, or maintain the equipment. We recommend clarifying that in these scenarios, the customer is not considered to be in physical possession of the equipment, or that their physical possession of equipment is not considered to be significant in comparison to other factors.

Comparing Contract Price to Rental Value: We would suggest that this factor is not always a relevant consideration for how certain cloud transactions — primarily those relating to computing resources — should be classified. The very nature of those cloud services is more of a utility model — with computing power fluctuating or being added or subtracted based on the customer's needs of the moment — and is very different from a rental business model. Additionally, cloud services are predicated on providing a cohesive ecosystem of infrastructure and services. For these reasons, it is difficult and not particularly meaningful to compare rental value to the contract price in these circumstances. Indeed, the comparable rental value will likely be difficult or impossible to determine in many cases. For most cloud providers, there will be no similar property that is being rented by the taxpayer to use as a proxy for determining the rental value of equipment, as the taxpayer may operate a purely services-based business and may have specially designed or proprietary equipment that is not otherwise available on the rental market.

Mere Passage of Time: We would recommend that the final regulations clarify that where there are multiple factors that contribute to how a customer is billed for access to cloud services, the “mere passage of time” element is not considered met where the billing is based on other usage factors, such as level of use and number of employees with access, in addition to an element of time. Additionally, clarification would be helpful to address situations where a customer reserves and pays for capacity for a certain amount of compute power over a particular term, regardless of whether that entire capacity is used (which sometimes offers more favorable pricing). We would recommend an additional example clarifying that in these cases, for the purposes of analysis of the “mere passage of time” factor, the fee is considered primarily based on a measure of work being performed or the level of the customer's use, rather than the mere passage of time, such that analysis of this factor supports the overall conclusion that the transaction is services.

Moreover, we would request clarification of how the element of “mere passage of time” applies to subscription services for digital content. In these subscription models, customers are typically billed monthly for access to digital content ranging from books to movies or television to music, with customers often receiving access to unlimited content. That said, we believe that qualitative analysis of the other factors otherwise weighs strongly in favor of a determination that these subscription services for digital content are provision of services, not leases. Accordingly, we would appreciate clarification that this element is not decisive or heavily weighted in analysis of the character of subscription services, as it seems less applicable to these services in general.

Transaction-by-Transaction Characterization Rules and De Minimis Standard

Under §1.861-19(c)(3), an arrangement that involves multiple transactions and includes at least one cloud transaction requires a separate classification of each component transaction (except any transaction that is de minimis). We anticipate that in practice, this will actually be one of the more complex parts of implementing these regulations, so we appreciate the opportunity to offer some recommendations to improve administrability.

We appreciate that the examples frequently reference “core” functions, “primary benefit,” or similar, for purposes of determining whether transactions are separate and whether an element is “de minimis.” Language in the regulations themselves clarifying that analysis should examine the “primary benefit,” “core function,” or “predominant character” based on the facts and circumstances would help to further clarify that analysis should be qualitative in nature and does not require a numerical comparison of the notional cost or fee assigned to a particular component of a transaction (which may not exist) in order to arrive at the conclusion that a component of the transaction is “de minimis.” “De minimis” might also be replaced with “ancillary” to further make clear that the focus is the predominant purpose of the transaction, and that the use of “de minimis” does not suggest an element of the transaction must be truly minimal.

Adding an elective numerical safe harbor that taxpayers can use when the cost or fee for a component is known, such as 25 percent or less of the overall fee, to the qualitative test might further improve certainty and administrability while clarifying the expectations for when an element of a transaction is “de minimis.”

In the absence of adopting a predominant character analysis, it would be helpful for these regulations to further clarify the distinction between what may be considered a separate transaction that is part of an arrangement, and what should be considered a component or an ancillary feature that is included in a transaction. For example, where a transaction is billed as one sales transaction and merely contains ancillary features that are not billed separately and would not customarily be, the regulations could provide a presumption that there is only one transaction. As one example, a monthly subscription for streaming services might include temporary downloads subject to an activation lock, but these features are not billed separately and would not customarily be. These features should be considered components of one transaction.

Sourcing

Beyond assisting taxpayers in determining the character of a cloud transaction as a service or a lease, the regulations do not provide additional guidance on how cloud transactions should be sourced. Sourcing certain cloud transactions is likely to prove challenging, given that the operations and employees involved in providing the services — as well as the customers using them — may be dispersed. Determining the location of Software as a Service (SaaS) transactions between businesses can be particularly difficult. In an effort to provide additional certainty, taxpayers could be given the option to use the location of contracting as set forth in the contract itself. We believe this approach would allow taxpayers the option to significantly simplify the sourcing of these transactions, and that there is minimal potential for abuse as long as taxpayers use a consistent method.

Treatment of Downloaded Content

In the case of digital content that does not represent the sale of a copyrighted article under §1.861-18, the regulations are structured in a way that assumes a binary option between digital content that is downloaded to the user's device (which is deemed to be a lease under §1.861-18), as compared to digital content that is provided online or otherwise streamed to the user (which is considered a cloud transaction and therefore likely a service under §1.861-18). We are concerned that this will not be a meaningful or helpful distinction to draw for several reasons.

First, technological and practical limitations, rather than differences in the type of transaction or business model, often drive whether the user downloads digital content to their computer or device, or streams the content instead. For instance, the type of media may play an outsize role in that decision, despite the transaction being otherwise similar, due simply to the technological limitations. Books obtained as part of a subscription may be downloaded instead of streamed because their file sizes are relatively small, meaning a short download time and minimal storage capacity required on the user's part. On the other hand, videos or movies a user enjoys as part of a subscription are more likely to be streamed because very large file sizes mean that they would take longer to download and require more storage space.

Indeed, users often have the option to either download or stream the same content based on their preferences and needs. Users of a streaming music service may prefer to download some songs so they can still be accessed when internet or cellular data service is not available, but might otherwise stream content through the same service. Even content that is downloaded may be downloaded by the same user multiple different times, either to different devices or in different versions.

Ultimately, we believe that whether a user downloads or streams content — a decision that will be made largely based on technological or practical limitations in what is fundamentally the same underlying transaction for time limited access to digital content — should not drive the tax treatment. Instead, we believe that transactions involving access to a catalog of digital content where content may be, subject to restriction, downloaded to the user's device on a temporary basis should be treated as “services” in accordance with the characterization of cloud transactions provided in §1.861-19.

Interaction with Foreign-Derived Intangible Income (FDII) (§250)

Given that the proposed regulations would apply to the international provisions of the Code, including several provisions enacted as part of the Tax Cuts and Jobs Act (e.g. Sections 59A, 245A, 250, and 267A), the determination of whether a cloud transaction is a service or a lease of property can have a significant impact.

Specifically, in the case of §250, treatment of a cloud transaction as a lease or service would impact eligibility for the FDII deduction, given that there are different rules for establishing “foreign use” and providing acceptable documentation for services and lease transactions. The proposed regulations implementing §250 are also unclear as to whether transfer of a copyrighted article is a transaction in tangible or intangible property for purposes of applying the rules. Notably, in our comments on the proposed regulations implementing §250, ITI expressed that we believe that the rule on foreign use for intangible property should replicate the rule governing foreign use of general property, which would eliminate these concerns.

Treatment of Intercompany Transactions

It would be helpful to have clarity in the context of intercompany transactions, such as data hosting services. In a global structure, a corporation may obtain data hosting services from a related party, which may be a subsidiary or may be under common control. The related party obtains space, purchases equipment, purchases telecommunications services, and incurs other costs, and is responsible for ensuring the equipment is operational, including providing for physical maintenance, repair, and security. The corporation may be the sole customer or one of multiple customers within the affiliated group. As a key customer, the corporation typically has certain specification requirements for the equipment used to provide the services, and the related party receives an arms-length fee calculated based on costs incurred plus an appropriate mark-up, which may influence the data hosting services provider's overall procurement process. We would expect that this would overall be treated as a services transaction, and would not anticipate this would change based on the proposed regulations.

However, it would be helpful to clarify that the related party aspects of the transaction do not change the conclusion that the customer corporation does not control the equipment or have a significant economic or possessory interest in the equipment, thus allowing the transaction to be treated as services. Specifically, it would also be helpful for the regulations to affirm that the arms-length cost-plus charge is primarily based on a measure of work performed or the level of the customer's use rather than the mere passage of time, and does not change a conclusion that a provider otherwise bears any risk of substantially diminished receipts or increased expenditures in case of non-performance (or concludes that these factors are not relevant to the analysis in this instance).

Once again, we appreciate the opportunity to comment on these regulations, and are more than happy to discuss our input further.

Sincerely,

Sarah Shive
Senior Director, Government Affairs
ITI
Washington, DC