Cross-border expansion into the United States is one of the most common growth moves for Canadian fintech and digital asset businesses. But the boundary between “serving US customers” and “carrying on business in the US” is thinner than many founders realise — and crossing it without planning can trigger US federal tax filings, state nexus exposure, and a transfer pricing audit on the Canadian side. With the IRS sharpening its focus on digital businesses and the CRA increasing transfer pricing scrutiny under section 247 of the Income Tax Act, 2026 is the year to revisit how the Canada-US tax treaty actually applies to your structure.

Why the Treaty Matters More Than You Think

The Canada-US Tax Convention exists to prevent double taxation, but it does not exempt Canadian companies from US tax simply because they are tax-resident in Canada. The treaty allocates taxing rights based on whether a Canadian enterprise has a permanent establishment (PE) in the United States. If a PE exists, the US can tax the business profits attributable to it; if it does not, those profits remain taxable only in Canada.

For traditional businesses, PE analysis was straightforward: a US office, a US warehouse, or a dependent agent meant a PE. For digital businesses, the analysis is messier. Canadian fintechs often operate with US-based contractors, US-domiciled cloud infrastructure, US payment partners, and remote-first teams scattered across both countries. Each of these arrangements can create — or avoid — a PE depending on how the relationships are structured and documented.

Three PE Triggers Digital Businesses Underestimate

1. The Dependent Agent PE

Under Article V(5) of the treaty, a Canadian company has a US PE if a person in the United States habitually exercises authority to conclude contracts in the company’s name. The 2007 Fifth Protocol expanded this to include agents who play a principal role leading to contract conclusion, even if signing happens elsewhere.

For fintechs, the risk often hides in US-based sales staff, business development hires, or even “contractors” whose day-to-day work involves closing customer contracts. Calling someone an independent contractor in a services agreement does not, on its own, neutralise PE risk. The CRA and IRS both look at the economic substance of the relationship.

2. The Services PE

The Fifth Protocol added Article V(9), a services PE rule unique to the Canada-US relationship. A Canadian enterprise is deemed to have a US PE if either:

• A single individual is present in the US for 183 days or more in any 12-month period and earns more than 50% of the enterprise’s active business revenue from US-sourced services, or
• Services are provided in the US for 183 days or more in any 12-month period in connection with the same or connected project for US customers.

Remote-first teams routinely breach this threshold without realising it. A Canadian engineering lead who spends four months a year in San Francisco visiting a key US client can singlehandedly trigger services PE exposure for the entire Canadian parent.

3. The “Server PE” Question

The OECD’s commentary on PE recognises that a server — if at the enterprise’s disposal, fixed in location, and used for core business functions — can constitute a PE. Canada and the US generally follow this view. For digital asset platforms, the practical question is whether your US-hosted infrastructure (a custody node, a matching engine, a settlement layer) is being run by your Canadian entity or for your Canadian entity through a third-party cloud provider. The distinction matters, and it should be documented in your cloud and infrastructure agreements.

Transfer Pricing: The Other Half of the Story

Even when a Canadian group avoids PE status, it may still operate through a US subsidiary. That brings section 247 of the Income Tax Act and the corresponding US section 482 rules into play. Intercompany flows — licensing of IP, shared services, cost-sharing for product development, intercompany loans — must be priced at arm’s length and supported by contemporaneous documentation.

The CRA has signalled that fintech and crypto groups will receive increased transfer pricing attention through 2026 and 2027, particularly where intangibles (algorithms, customer data, brand) are developed in one jurisdiction and exploited in another. Penalties under section 247(3) apply to adjustments exceeding the lesser of $5 million or 10% of gross revenue.

Practical Steps for Canadian Groups in 2026

• Map your US footprint. List every US-based employee, contractor, sales partner, server, and physical asset. Note who has authority to negotiate or conclude contracts.
• Track US presence days. Implement a simple travel log for any individual who spends meaningful time in the US on business. The 183-day threshold is cumulative across any 12-month window.
• Review intercompany agreements. If you operate through a US subsidiary, confirm that services, IP licensing, and cost-sharing arrangements are documented, signed, and priced consistently with the functional analysis.
• Consider Form 8833 protective filings. Where treaty-based positions are taken to override the default US tax treatment, IRS Form 8833 protects against the failure-to-disclose penalty under section 6712.
• Coordinate with state nexus analysis. The treaty does not bind US states. California, New York, Texas, and Washington each have their own economic nexus thresholds that can apply even when no federal PE exists.

The Bottom Line

The Canada-US treaty is a powerful tool, but it rewards groups that plan ahead and punishes those that don’t. A clean cross-border structure with documented intercompany arrangements, intentional staffing patterns, and clear treaty positions is straightforward to build before you grow into US revenue — and expensive to retrofit once the IRS or CRA asks questions. With both tax authorities sharpening their focus on digital businesses through 2026, the cost of a structural review now is small compared to the cost of an inbound audit later.

Need guidance? Reach out to our team — no pressure, no jargon.