Apple, Microsoft, IBM and other American technology companies have closed or sharply reduced R&D operations in China. The usual explanation is geopolitical tension and decoupling. I think a technical change in U.S. tax law deserves far more attention.
The change: immediate deductions became long amortization
Before 2022, Section 174 of the Internal Revenue Code gave technology companies a powerful advantage. Research expenses could be deducted in the year they were incurred. In cash-flow terms, it worked like a tax credit card: spend on research now and receive the tax benefit now.
That treatment dated to 1954 and had supported American high-tech companies for almost seventy years. It also gave management some flexibility. A company having a strong year could increase research and take the deduction immediately. In a weaker year, it could pull back.
The concealed cost of the 2017 tax cut
The 2017 Tax Cuts and Jobs Act looked like a large tax reduction, especially after it cut the corporate rate from 35 percent to 21 percent. But it also rewrote Section 174. From 2022 onward, domestic research expenses had to be amortized over five years. Research conducted outside the United States had to be amortized over fifteen.
The tax benefit no longer arrived with the expenditure. Companies spent the cash today and recovered the deduction slowly.
What this does to a China-based research center
Consider an American company spending 00 million a year at an R&D center in Shanghai.
Before 2022:
00 million was deductible in the same year.
At a 21 percent rate, the tax saving was about
1 million.
The after-tax cost was about $79 million.
For a technology company that must fund research continuously, this is not an accounting footnote. It is a cash-flow shock. The company pays for the work now but waits fifteen years to receive the full deduction.
Since 2023, the pattern has become difficult to ignore. An American chip company closed its Shanghai research center and cut hundreds of jobs in 2023. Microsoft was reported to be planning a large China workforce reduction of nearly 2,000 positions in April 2025. IBM announced the closure of its China research operation in 2024, affecting more than 1,800 employees.
People inside the industry have pointed to tax pressure as an important part of these decisions. The issue is not simply whether a company wants to innovate in China. The location has become much more expensive to carry on an American tax return.
Research locations used to be chosen mainly for talent, market access and operating cost. Now the tax strategist's spreadsheet can overrule the engineer's judgment. A small section of the tax code is helping redraw the map of global innovation.
The policy also serves several American goals at once.
It raises revenue by expanding the taxable base.
It pushes high-value research back toward the United States by treating domestic and foreign expenditure differently.
It lets politicians retain the headline claim that they cut the corporate rate.
Seen from farther away, this is industrial policy written in tax language. The United States is using the tax code to alter the division of labor in the global technology supply chain and pull valuable research activity home.
Researchers may have to move to follow the work. International technology partnerships may be reorganized. Concentrating research geographically may also reduce the efficiency of cross-border collaboration.
Other countries should pay attention. Tax policy is not merely a tool for collecting revenue. It can be industrial policy and a weapon in international competition. Governments will have to decide how to attract foreign investment without hollowing out their own industries.
The closure of American R&D centers in China looks, on the surface, like another consequence of geopolitics. I think the tax change is a more direct cause, and one planted during Trump's first term. A dry amendment to Section 174 is changing where technological innovation happens. In this contest, tax codes may matter as much as speeches about decoupling.
