The rise of AI has highlighted how the global tax system is ill-suited to capture the profits derived from such technology. While creating an equitable AI economy ultimately requires a durable and robust multilateral tax framework, digital-service taxes can serve as an important stepping stone to that outcome.
WASHINGTON, DC—AI’s rapid advance and the resulting popular backlash have lent new urgency to an old question: How can we promote equitable economic growth from technological innovation while mitigating the negative social and environmental consequences? Both domestically and globally, the answer starts with corporate tax policy.
Anxieties about AI have less to do with any philosophical unease than with the societal effects. Financial gains are increasingly concentrated among a handful of firms and their shareholders, while the costs—from job displacement to rising energy prices—are borne by everyone else. In democratic societies, taxation is the primary mechanism for translating private gains into public benefits. When that mechanism breaks down, a backlash is inevitable.
Distributive justice is an equally salient concern at the global level. While the Global North is preoccupied with the significant public pushback against AI, the Global South faces a pressing need to mobilize domestic tax flows from multinational corporations, particularly tech firms, in the wake of drastic aid cuts. These problems stem from the same source: when companies shift profits out of the countries where they are generated to tax havens, that reduces public revenues for all governments.
The digitalization of the global economy has demonstrated the insufficiency of the international tax system. But the need to shatter the status quo has never been clearer. AI has supercharged the trend of taxable value being decoupled from physical corporate presence—long a prerequisite for a country’s right to tax. Training data, compute, and intellectual property can be scattered across many jurisdictions, far from where workers or users are located, fueling debate about where value is created.
US-based multinational tech giants, some of which are poised to win big from the AI revolution, have created a playbook that companies like OpenAI and Anthropic could follow. Because their profits are mostly generated from intangible assets such as software, they can shift them to tax havens, manipulating decades-old norms to pay rock-bottom rates. It is no wonder that so many Big Tech firms have been embroiled in tax controversies.
For example, a 2013 US Senate investigation and a subsequent 2016 European Commission investigation revealed that Apple had shifted a large portion of its billions of dollars in global earnings to its Irish subsidiaries, which then paid a tax rate on Apple’s European profits as low as 0.005%. The uproar culminated in European Union authorities ordering Apple to pay Ireland a staggering €13 billion ($15 billion) in back taxes and sparked a major international tax overhaul at the OECD.
The inability to tax the profits of digital companies demands nothing less than an overhaul of the international tax system. Policymakers have attempted to arrest the global tax race to the bottom by introducing minimum taxes at both the national and global levels. The United States enacted a corporate alternative minimum tax in 2022, but it has since been diluted. The same is true of the global minimum corporate tax adopted under the OECD/G20 Inclusive Framework on Base Erosion and Profit Shifting, whose planned reallocation of taxing rights from headquarters to market countries has stalled.
Tax-justice groups have criticized the Apple court decision and the OECD reforms for failing to consider developing countries’ interests. To create a fairer and more predictable system, African governments have proposed a Framework Convention on International Tax Cooperation under the auspices of the United Nations, with negotiations currently underway in New York.
The world ultimately needs a robust and durable multilateral tax framework (whether at the UN, the OECD, or elsewhere) to capture the value of AI-driven economic activity where it occurs. But a global minimum corporate tax and the reallocation of taxing rights, while both necessary, would not help governments tax AI companies in the near term, as many have yet to turn a taxable profit. That is why a growing number of jurisdictions are looking beyond taxes on net income and introducing digital-service taxes (DSTs) on gross revenues.
More than 20 countries are considering or have already implemented DSTs, prompting threats of retaliatory US tariffs. Moreover, different versions of such taxes are currently in effect or have been proposed in several US states. In all these cases, policymakers recognize that DSTs are a valuable release valve, be it globally or at the subnational level.
While not a long-term substitute for reforming an obsolete global tax system, DSTs are at least a stopgap. They provide proof of concept by normalizing the once-radical idea that taxing rights should align with where users are located, not just where companies choose to incorporate or play accounting games. DSTs may not be the final destination, but they are an important and necessary first step toward a fairer international tax order.
DSTs, like broader multilateral tax reforms, have come under fire from short-sighted politicians and experts who insist on defending the status quo. But these leaders fail to recognize that the status quo is fiscally and politically untenable. This was the case before AI arrived but has come into sharper focus as the technology spreads.
If AI-driven profits are subject to the same asymmetric global tax architecture that has allowed digital activities to be undertaxed, there is a good chance that Big Tech’s outsize power will only grow. There is also the risk of resistance to AI hardening into a legitimacy crisis, which means that society may miss out on its potential benefits. Corporate tax policies are the battleground where public trust in this new technology, and our governing institutions, will be won or lost.
