The conventional framing of tech tax avoidance goes something like this: companies earn money, then scramble to hide it before the IRS comes knocking. That framing is wrong in an important way. The largest tech companies don’t move profits after the fact. They build the disappearing act into their corporate structure before the first dollar of revenue arrives. Transfer pricing isn’t a tax strategy layered on top of a business. For many tech giants, it is the business model.
What Transfer Pricing Actually Does
Transfer pricing is the practice of setting prices for transactions between subsidiaries of the same parent company. When Google’s Irish subsidiary licenses intellectual property from a Bermuda entity and pays a royalty for it, both sides of that transaction belong to Alphabet. The price they agree on determines where profit appears on the books, and therefore where it gets taxed.
This is legal. It is also enormously consequential. The OECD estimates that profit shifting by multinationals costs governments between $100 billion and $240 billion in lost tax revenue annually. Tech companies account for a disproportionate share of that figure because their core assets, software and intellectual property, are easy to move across borders and nearly impossible to value with precision.
Apple’s structure, examined in detail during a 2013 U.S. Senate subcommittee hearing, routed profits through Irish subsidiaries that were tax resident in neither Ireland nor the United States. The committee found that Apple had paid an effective tax rate of less than 2 percent on $74 billion in offshore income over four years. The money didn’t hide itself. Apple built a structure specifically designed to put it somewhere no government could easily reach.
Why Software Makes This Uniquely Easy
Manufacturing companies shift profits too, but they face a practical constraint: physical goods have to cross borders, generating customs records and observable trade flows. Software has no such problem. A patent can be “relocated” to a low-tax jurisdiction through a paper transaction. A license agreement between two subsidiaries can be priced at whatever rate survives an audit challenge. The marginal cost of delivering another copy of a software product is effectively zero, which means almost all of the revenue above that cost is profit, and almost all of that profit can be assigned wherever the structure allows.
The critical mechanism is intellectual property ownership. When a U.S. tech company transfers the rights to its IP to a foreign subsidiary early in its growth (a “cost-sharing agreement” in tax terminology), future profits attributable to that IP accumulate outside the U.S. The IRS gets to tax what was earned developing the IP when it was still cheap. The foreign entity captures all the appreciation. Microsoft, Amazon, and Google have all used variations of this structure. It is not an edge case. It is standard operating procedure for any well-advised multinational tech company.
As we’ve covered previously, understanding how profits move across borders before tax season is essential context for reading any tech company’s effective tax rate with clear eyes.
The Arm’s Length Fiction
Tax law requires that transactions between related parties be priced as if they were between independent parties dealing at arm’s length. In theory, this prevents a company from charging its Cayman subsidiary one dollar for a patent worth ten billion. In practice, arm’s length pricing for unique intellectual property is a fiction that both companies and tax authorities largely agree to maintain because neither side can prove the other wrong.
There is no market price for the algorithm that powers Google Search or the software stack underlying AWS. Expert witnesses for the taxpayer will produce one valuation. Expert witnesses for the government will produce another. Courts split the difference, often decades after the original transaction. The penalty for aggressive pricing is, in the worst case, paying some of what you owed anyway, plus interest. The benefit is years of deferral and often permanent avoidance if the structure holds.
This asymmetry is not an accident in the law. It is the predictable result of writing rules that require market comparables for assets that have no market.
The Counterargument
The strongest defense of current transfer pricing rules is that companies are doing exactly what legislators and courts have sanctioned. Apple, Google, and Microsoft employ armies of tax counsel who know the rules precisely and operate within them. If the outcome looks wrong, the problem is the law, not the company. Shareholders have a legitimate interest in minimizing tax expense, and executives have a fiduciary duty to pursue it.
There is something to this. The 2017 Tax Cuts and Jobs Act introduced GILTI (Global Intangible Low-Taxed Income) specifically to claw back some of what transfer pricing gives away, and the OECD’s Pillar Two agreement, targeting a 15 percent global minimum tax, represents a genuine attempt to put a floor under the race to zero. Reform is possible and has happened.
But the defense proves too much. The fact that a practice is legal tells you something about lobbying effectiveness, not about whether the outcome is defensible. Tech companies have a well-documented habit of publicly supporting regulations they are privately paying to gut, and transfer pricing reform is no exception. The companies that built these structures also spend heavily to keep the rules ambiguous enough to exploit.
The Honest Conclusion
Transfer pricing works because tech companies build their corporate architecture around it before revenue scales. It is not tax planning applied to a mature business. It is a structural choice made at formation, refined during growth rounds, and defended during audits. The billions that disappear before tax season were never going to appear in a high-tax jurisdiction in the first place.
Calling this a loophole misses the point. A loophole is a gap that someone found. Transfer pricing in tech is a door that was designed to be left open, maintained carefully, and walked through deliberately by everyone large enough to afford the key.
