AI Is Creating the Biggest Wealth Transfer in History. No One Is Taxing It.

Let us talk about money. Specifically, about who is making it, and who is paying the price.

In the first quarter of 2026, Microsoft reported profits of $32 billion, Google’s parent Alphabet made $62 billion - up 81 percent from the same quarter a year earlier - and Meta cleared $27 billion. These are not revenues but profits, after costs, after taxes, after everything - quarterly figures - and a significant and growing share of them is being driven by one thing: the deployment of artificial intelligence at scale across their products and services, reducing headcount, automating tasks, and extracting value from users and workers at a ratio that would have been unimaginable a decade ago.

At the same time, across Europe and North America, hundreds of thousands of workers in sectors from legal services to financial administration to content production are losing jobs, seeing their hours cut, or watching their wages driven down by the same technology generating those quarterly numbers - and they are not being compensated for this, nor retrained at any meaningful scale. They are, in the language of economics, bearing the negative externalities of a transformation whose positive returns are flowing almost entirely to a small number of companies and the people who own them. This is not a market failure in the technical sense but a policy choice - specifically, the choice not to tax the winners of this transformation and use the proceeds to protect those bearing its costs, a choice that is becoming harder to justify with every quarterly earnings report.

The intellectual case for an AI levy is not complicated, and it does not require any particularly radical assumptions about the role of the state in the economy. It rests on a principle that is already embedded in the legal frameworks of most democratic societies: when an economic activity generates significant costs that are borne by parties other than those profiting from it, the state is entitled to intervene to ensure those costs are fairly distributed.

We apply this logic to carbon emissions, to financial transactions, to alcohol and tobacco whose social costs in healthcare and law enforcement are partially recovered through taxation. The principle is not controversial - what is controversial, and what the technology industry has spent enormous resources lobbying against, is applying it to AI.

The case for doing so is straightforward. When Microsoft deploys Copilot across its enterprise software and a law firm subsequently lays off twelve paralegals, those twelve people do not simply disappear into a statistical abstraction: they lose income, draw on unemployment insurance funded by taxpayers, may need retraining funded by public budgets, and leave their communities with less spending power. The costs are real, they are measurable, and they are being borne by people and institutions that had nothing to do with the decision to automate those jobs, while Microsoft books the productivity gain. That is an externality - and externalities, in a functioning political economy, get taxed.

What would a serious AI tax regime actually look like? There are three distinct instruments worth combining.

1. The first is a levy on large-scale AI model usage - essentially a per-query or per-API-call tax on commercial AI deployment above a certain threshold of scale, targeting the activity most directly responsible for labour displacement: the industrial-scale use of AI to replace human cognitive work in commercial settings. It would apply to OpenAI’s API, to Google’s Gemini deployment, to Anthropic’s enterprise contracts, to every company using these tools at the scale that generates meaningful displacement, while leaving small businesses and individual users entirely exempt.

2. The second is a windfall tax on the extraordinary profits being generated by AI-driven productivity gains in the technology sector more broadly. Microsoft, Alphabet, Meta, Amazon, and Apple have collectively added trillions of dollars in market capitalisation over the past two years, driven substantially by AI-related earnings growth and investor expectations, and a portion of those profits - not all of them, not even most of them, but a meaningful fraction - should be redirected toward the workers and communities whose displacement is generating them. This is not redistribution as ideology but redistribution as basic accounting: the people creating the value being captured by these companies are entitled to a share of it.

3. The third instrument is a wealth tax on the individuals who have benefited most dramatically from this concentration. Elon Musk’s net worth stands at over $700 billion, making him the first person in history to cross that threshold - a fortune driven substantially by AI and automation plays across Tesla, xAI, and his other ventures. Jensen Huang of Nvidia - whose chips power virtually every significant AI system in the world - has seen his personal wealth grow from $4.7 billion in 2020 to over $160 billion today, a thirtyfold increase built almost entirely on the AI boom. Jeff Bezos, Mark Zuckerberg, Satya Nadella: the list of individuals whose personal fortunes have been transformed by this technological moment is short, their gains are extraordinary, and the social contribution they are currently making through taxation is, relative to those gains, embarrassingly small.

The objections to these proposals are predictable, and worth addressing directly.

The first is competitiveness: tax AI too heavily in Europe (or in any other major economy) and the companies will simply move their operations elsewhere. This argument has been made against every attempt to regulate or tax the technology sector for the past thirty years, and it has consistently proven more useful as lobbying rhetoric than as economic analysis - the companies generating these profits need access to European markets, European talent, and European data, and they are not going to abandon a market of 450 million relatively wealthy consumers because they have to pay a levy on their API calls. The appropriate response to the risk of regulatory arbitrage is international coordination, not unilateral surrender, which is precisely the kind of policy agenda that requires political organisation to pursue.

The second objection is innovation: tax the technology and you slow the development of AI, depriving society of its benefits. This conflates the taxation of extraordinary profits with the suppression of innovation, which are not the same thing - Google is not going to stop developing AI because its quarterly profit is $32 billion instead of $34 billion, and the margin available for taxation without any meaningful impact on investment decisions is vast. Anyone arguing otherwise is either innumerate or arguing in bad faith.

The third objection is the most honest one: it is politically difficult. The technology industry is among the most powerful lobbying forces in both Washington and Brussels, and the individuals whose wealth would be affected by a serious wealth tax have the resources to fund political campaigns, think tanks, and media operations that will push back against any such agenda with considerable force. This is true - and it is also precisely why the political response to AI cannot be left to the existing parties, which are too embedded in existing donor relationships to take these proposals seriously.

The money exists, and the legal and intellectual frameworks for taxing it exist too. What is missing is the political will to do it - and the political organisation to create that will where it does not currently exist. An AI levy, a windfall tax on technology profits, and a serious wealth tax on the individuals most enriched by this transformation would generate hundreds of billions of euros annually at the European level alone - enough to fund a meaningful AI Transition Fund, serious retraining programmes, and the beginnings of the income support infrastructure that a just response to this disruption requires.

The question, as always, is not whether this is possible. It is whether anyone is willing to fight for it.