Foreign media commentators believe that various proposals have emerged within the US policy circle regarding whether AI requires major tax reform, including separate taxes on AI elements such as computing power and tokens, or the establishment of a special fund to address employment impacts. However, the article argues that current assessments of the labor market remain highly uncertain and insufficient to support rewriting tax rules specifically for AI.
Oppose establishing a special tax for AI
The article mentions that recently, several public figures, including Sam Altman, Vinod Khosla, Mark Cuban, and U.S. Senator Elizabeth Warren, have put forward different ideas on the tax system in the AI era, ranging from reducing labor tax burden to levying wealth tax.
However, the author argues that the fundamental principles of tax policy should not be disrupted by the emergence of new technologies. Whether it was the Industrial Revolution or the widespread adoption of personal computers, the more prudent approach has always been to maintain simple rules, lower tax rates, and a broader tax base, rather than imposing new taxes specifically on any particular technology.
The article argues that taxing AI-specific elements such as computing power or tokens could increase the cost of using new technologies, thereby discouraging companies from adopting AI.
Existing tax categories can still cover AI revenue.
The commentary points out that the wealth growth brought about by AI does not mean that the tax system is ineffective. Taking the rise in the stock price of AI companies as an example, the increase in book value itself may not immediately generate taxable profits, but investors may still incur capital gains tax after selling the shares.
The article also cites the example that the expansion of data centers has not caused local tax revenue to fail. Loudoun County, Virginia, benefited in part from property tax revenue generated by data centers in 2025, thereby reducing the tax burden on residents.
Based on this, the author argues that the policy focus should not be on designing additional incentives or special terms for the data center or AI industry, but rather on continuing to handle new revenue through existing tax categories and using the unexpected tax revenue to address real-world issues such as deficits and local fiscal pressures.
If employment is impacted, first repair unemployment insurance.
The article also holds reservations regarding the discussion on "whether an AI transformation fund should be established." Supporters believe that if AI leads to the loss of some jobs, a portion of the wealth created by AI should be used to compensate the affected workers.
However, the author points out that the United States has had similar experience. The Trade Adjustment Assistance Program (TAA) was originally intended to help workers affected by trade disruptions, but it has long suffered from low claim rates, partly because it is difficult to pinpoint which policy directly caused a particular layoff.
The article argues that identifying AI-driven layoffs faces similar difficulties. Current cases of so-called AI-driven layoffs are often accompanied by business pressures or are related to over-hiring in 2021 and 2022.
Therefore, the commentary argues that if a more significant employment shock does occur in the future, a more realistic approach would be to improve the existing unemployment insurance system rather than to establish a separate compensation mechanism specifically for AI.
