The debate over where “income originates” in the digital economy has been challenged for years by an uncomfortable reality: a cloud service can be sold in one country, operated from another, and rely on intellectual property developed elsewhere. To organize this puzzle, the U.S. Internal Revenue Service (IRS) and the U.S. Treasury approved a technical reform with direct impact on multinational tech companies, infrastructure providers, and businesses selling SaaS to international clients.
It’s important to clarify a key point, as there is some confusion circulating: the final regulation has been in effect since January 14, 2025 (not 2026), while the most disruptive part — how to “localize” cloud income — is still in the proposal phase and does not apply until it becomes a final rule.
1) Cloud isn’t “digital content”: the new technical frontier
The final regulation reinforces the distinction between two types of transactions:
- Cloud transaction: when the customer gains on-demand access, via the internet, to resources such as hardware and digital content, among others. In other words: it’s not about “taking” software but about consuming capacity/functionality remotely.
- Digital content transaction: when the primary aspect is the transfer of digital content (for example, certain downloads or content/software deliveries under a scheme closer to “copy” or “assignment of rights”).
This distinction is important because the tax characterization changes: the Treasury/IRS states that cloud transactions are treated as service provision under the applicable regulations.
2) Goodbye to “creative” splitting: the “predominant character” test arrives
One of the points most interest to international finance and tax specialists is the so-called predominant character test. Translated into business language: if a product combines elements (for example, a SaaS that also allows offline downloads), the transaction is classified by the main benefit the customer receives, and that classification “drives” the overall treatment.
This reduces the scope for income “fragmentation” strategies (separating part as a license, another as a service, etc.) when, in practice, the buyer is paying for an integrated experience.
3) The most controversial aspect is not final: the formula for locating cloud income
The area where the IRS genuinely shifts the landscape (pending approval) is in the proposal to determine which part of cloud income is considered U.S. source. Instead of focusing on the location of the customer or where the contract is signed, the proposal points to a formula based on three factors:
- Intangible assets (intellectual property and related R&D activities)
- Personnel directly involved in providing the cloud service
- Tangible assets (physical infrastructure such as servers and equipment)
The idea is to calculate the “U.S. share” of these three factors and apply that proportion to the gross revenue of cloud transactions, with additional anti-abuse measures included.
4) What does this mean for international tech companies
Although the text is technical, the practical consequence is quickly understood: the taxation of cloud would tend to follow the people, infrastructure, and IP, not the “market” where the customer is located.
For a European or Latin American company selling cloud services to the U.S., the analysis could change if:
- Part of the operational or R&D team is based in the U.S.
- There are servers, equipment, or colocation contracts in the U.S.
- The “real” value chain (development/operations) relies on assets or staff located there.
And for a U.S.-based company operating cloud services abroad, the mirror effect is also relevant: the source of income influences issues like interactions with foreign taxes and certain procedures within the U.S. tax system.
Additionally, an often-overlooked detail is that the IRS has opened the door to discussing whether these characterization rules (cloud vs. digital content and “predominant character”) should apply beyond the specific international scenarios for which they are currently designed.
5) 2026: why it’s back on the radar now
As of January 5, 2026, two key aspects are relevant:
- What’s already in effect (2025): definitions, classification of cloud transactions as services, and the predominant character test.
- What could redefine planning and structure (pending): the “sourcing” formula based on intangible assets, personnel, and tangible assets, which is still proposed and may be adjusted if finalized.
For many companies, the real work isn’t just about “paying more or less,” but about being able to demonstrate with traceability where services are created and operated: which teams do what, from where, with what infrastructure, and which intangible assets are involved.
Frequently Asked Questions
What does the IRS consider a “cloud transaction” versus a “digital content transaction”?
Cloud focuses on on-demand online access to resources (a service). Digital content is more about transfer/delivery of content or software in digital format.
If my SaaS includes a downloadable app, can I separate income for different tax treatments?
With the “predominant character” rule, the operation is classified based on the primary benefit to the customer, reducing attempts to artificially fragment the transaction.
Is the new formula for determining where cloud income is generated already in effect?
No. The three-factor formula (intangibles, personnel, tangibles) appears in proposed regulations and is not applicable until approved as a final rule.
What should a company selling cloud services internationally review?
Primarily, its actual operational footprint: where the personnel providing the service are located, where infrastructure is situated, and how IP/R&D are attributed to the cloud service, along with maintaining consistent documentation.
via: GreenbergTraurig