The new economy is already here. Has anyone thought about the tax?

This is a publication that is part of RSM's Voice of SCM. On a monthly basis, RSM issues the Voice of Supply Chain Management (SCM). Our SCM consultants constantly follow global SCM developments and translate their impact into practical considerations for internationally active companies. 

Taxing the new economy is not yet on most business agendas, even though that economy has already arrived. Tax rules were built around people, owners and places, and each of those assumptions is loosening as machines, software and subscriptions take over work that people and owners used to do. A car that drives itself, a robot that never takes a day off, a subscription instead of a purchase and an AI that does the work of a team: each of them is already part of everyday business. Most of us are still finding out how these technologies work, what they will cost and which of them will last. Few step back to see that together they are changing how the economy works, and fewer still ask what this means for taxation, which has always been built around the way the economy worked yesterday.

This article raises the tax question now, while we are still figuring out how the new economy works, and not when the answers are due. It first describes, in economic terms and without the detail of the daily news, what is going on around us: what is already here, what is coming and what is not yet visible. It then turns to what this could mean for taxation, without predicting how taxation will develop. Tax experts have argued for more than a decade about how to tax the digital economy, and that debate is not settled. The new economy reopens it and adds new questions, and the businesses that set up contracts and structures today will meet those questions in their tax returns long before the rules have caught up.

This article was written by Mario van den Broek (mvdbroek@rsmnl.nl) and Dick Brinkhof (dbrinkhof@rsmnl.nl). Mario and Dick are tax consultants with RSM Netherlands with a focus on Tax Governance and international taxation.

What is happening?

Platforms now carry a large part of daily mobility, and many of the people who drive for them get their rides from an app and have no employer. Industrial robots are installed at a pace that was once reserved for consumer goods. Data centers, the buildings in which AI runs, take a growing share of the electricity supply. Subscriptions are replacing purchases in more places and sharing turns owners into providers of access. Behind these stories sits a contrast. The economy that the statistics measure grows slowly, while a second economy underneath it moves at a very different speed.

The examples reach well beyond taxis. A small but growing share of the working population earns its income through a platform, and Europe has agreed rules on when such workers count as employees. Robots are moving from the factories of large companies to the shelves of small ones: most of them are built in Asia, and renting a robot as a service grows faster than buying one. The largest online retailers already run their warehouses with fleets of robots.

Energy and mobility are growing together. Batteries on the grid are expanding faster than planners expected, and electric cars have started to feed power back: a car-sharing service that does so is already live in a Dutch city. Ownership is coming apart from use as well. For cars, private lease and subscriptions now sit next to outright purchase, which is still the norm but no longer the only option.

Artificial intelligence changes who does the work inside a company. Companies with a single founder are becoming more common, and some of them reach in months a value that used to take a team years. Pricing follows. A growing number of AI services charge for the result they deliver, where software used to charge per seat or per hour. All this runs on computing power, and the electricity use of data centers is expected to keep rising steeply for the rest of the decade.

Governments are customers too. Exports of dual-use goods, which have both civilian and military uses, have grown strongly, and the European Union has set up large funds to support defense spending. Money itself is changing form: stablecoins, digital tokens tied to a currency, have become a sizeable market, although the euro plays only a small part in it. At the local level, energy cooperatives and shared cars show that new forms of shared ownership grow alongside the commercial ones.

Behind the individual stories a few patterns return. Bottlenecks such as grid connections, nitrogen rules and a shortage of workers limit growth more than demand does. Capital moves from customers to providers, because a growing share of computing power, batteries, robots and products is rented instead of bought. Pricing moves from hours and seats to usage and results. Technologies depend on one another, since autonomous vehicles, AI, robots and energy storage reinforce each other, and much of the supply comes from outside Europe, with China leading in robots and the United States in computing power, while Europe is mostly the one that sets the rules. And output needs fewer people: one founder with AI, or one supervisor for a fleet of vehicles or robots, changes how many people a business needs.

Other changes have a direction but not yet a market. Driverless taxis already carry passengers in a few American cities, and manufacturers have started to take orders for driverless cars from fleet operators before private buyers can order one at all. Humanoid robots, built to work in places designed for people, are still mostly a matter of forecasts. And an individual who buys a car that earns its own income, as a small business, is not yet possible. Even the people who follow humanoid robots most closely disagree on their pace: forecasts of how many will be in use by the end of the decade differ several-fold, depending on who makes them. That spread shows how much of this future still rests on expectation, with little yet to measure.

Who owns a driverless car matters too. In today's services the technology owner typically keeps the software and the app, a fleet operator runs, charges, cleans and maintains the vehicles, and the rider books through the technology owner's app. Three parties, and perhaps three countries, share one ride, which makes the example of Anna below less theoretical than it looks.

Then there is a layer that still sounds like fiction. Tesla has shown a Roadster that uses thrusters to launch itself off a ramp, turn in the air and land on its wheels, and says production could start within a few years. Whether it ever becomes a flying car in daily use is uncertain, and its maker has said that it may not be legal everywhere. Travel to Mars belongs in the same group: a promise without a date.

A further layer is a legal one. Argentina's president has proposed giving companies that are run by AI agents their own legal status, with little or no human needed at the helm, and services already exist that let anyone set up a company within minutes through a digital wallet and choose whether a person or an AI agent runs it. Whether such proposals succeed is uncertain, but they show where the question could lead: a business with profits and no person who decides.

What is the potential impact?

What does the tax system assume about the economy?

Taxation is rarely the first thing people think of when a new technology arrives, yet every tax system rests on a picture of how the economy works. Three parts of that picture matter here. People do the work, so wages and the taxes on them follow the worker. Owners own the assets, so a company buys a machine and writes its cost off over time. And value arises in a place, so a factory, an office or a shop shows where a business is, and which country may tax it.

Each of these is loosened by what we have just described. When a robot or a piece of software does the work, the wage that was taxed may disappear while the profit of the work remains. When a car is subscribed to instead of bought, use and ownership are no longer the same thing. When a car crosses a border, or a model runs in a data centre abroad, it becomes less clear where the business is.

Take Anna, who in theory buys a driverless car to start her own company. She buys the car, perhaps abroad. The driving software is a subscription from a foreign company. A platform sends her the rides and keeps a share of every fare. Between rides the car charges at cheap hours and sells power back to the grid, and then it crosses a border to serve customers in a second country.

Every step raises a question for a tax authority. Who earns what from a single ride: Anna, the platform, the software maker or the energy company? In which country is that profit taxed when the car, the software and the platform each sit in a different one? And who reports Anna's income, the platform or Anna herself? Anna is still a thought experiment, but each of these questions can already be asked about businesses that exist today.

Does taxing the new economy need new rules?

Tax systems have always adapted to the economy they serve, from taxes on land and trade, to taxes on wages and profits when industry grew, to new rules on digital services when the internet arrived. It would be remarkable if an economy in which machines and software do a growing share of the work left the tax system unchanged, although today's rules may well stretch further than expected. Many of these questions are old questions in new clothes. A driverless taxi company still has customers, contracts, costs and a profit, and tax rules that rest on principles instead of technology have been stretched before, for franchises, for online platforms and for the internet. What is new is the pace, the lower threshold, since an AI agent is far easier to deploy than a factory or a subsidiary, and the number of questions that arrive at the same time.

Work: who pays when a machine does the job?

The first question concerns work. In most countries a large part of public revenue comes from wages, as income tax and social contributions that are paid because a person does a job. If a robot, a driverless car or an AI agent does that job, the wage disappears from the tax base while the work, and the profit it generates, remain. Because the cost of a machine can be written off over time while employing a person brings payroll taxes, some argue that the balance tilts towards machines. This is why people talk about taxing robots or the use of AI. The idea is almost a decade old, easy to state and hard to design: it is not clear what exactly counts as a robot, the work can be moved to another company or country, a tax on a robot is in practice a tax on its owner, and nobody yet knows how much work will disappear and how much will simply change. Research on robot taxation therefore tends to conclude that a robot tax works better as a reason to rethink the taxation of business income as a whole than as a separate levy on machines.

Other ideas move the focus from taxing work to taxing capital or data: a tax on the data a company uses, a tax on AI tokens, a levy on the energy that data centers use, a public stake in AI companies or a basic income financed from capital. Each has drawbacks, since taxes on capital can slow down investment and innovation, and none of them is ready to use.

A related question is the status of the person who works through a platform. Whether a driver or courier is an employee or an independent contractor decides who withholds tax and pays social contributions, and with margins that are not large the answer can matter a great deal for the economics of the platform. The European rules that address this must be put into national law by 2 December 2026, the Netherlands has already tightened its enforcement of the rules on self-employment, and recent Dutch case law on platform drivers shows that the answer may differ from driver to driver. Platforms also report the income of their sellers to tax authorities under European rules, and Member States may, and from 2030 must, make platforms responsible for the value added tax (VAT) on rides and short stays.

Ownership: bought, rented or subscribed?

The second question concerns ownership. Tax rules treat buying a machine and renting a service differently: a purchase is written off over time, a lease or a service fee is deducted, and VAT and customs can differ as well. Subscriptions and as-a-service contracts blur these lines. When a bakery pays a monthly fee for a baking robot that the supplier maintains, updates and replaces, is it leasing a machine or buying a service? The answer depends on who carries the risks of ownership, and it is usually settled in a commercial contract before anyone has thought about tax. Both readings can be honest descriptions of the same arrangement, and corporate income tax, VAT and customs do not always read the same contract in the same way, so one arrangement can produce different answers. Supply chain management and international trade add a further layer. Because much of the supply of robots comes from outside Europe, imported robots raise customs questions: how a machine is classified, what it is worth when software and service are bundled with the hardware, and how VAT applies to a machine that arrives together with a subscription.

A recent example shows how much a label can decide. Prediction markets, where people trade contracts on the outcome of events, were ordered to stop in the Netherlands because the gambling regulator regards them as gambling, while in the United States the same contract is argued to be gambling, a capital investment, a futures contract or a swap, each with a different tax result. The instrument is the same, and the tax depends on a label that regulators are still disputing.  

Place: where is the business?

The third question concerns place. A country normally taxes a foreign company only if it has a real presence there, such as an office, a factory or a person who concludes contracts on its behalf. A robot in a customer's warehouse, a fleet of vehicles driving through a country or an AI agent that closes deals without a person involved do not fit comfortably in that picture. Existing rules do accept that equipment without staff can sometimes be a presence, but usually only if the company has it at its own disposal, which a rented cloud server often does not. The same applies to transfer pricing, the rules that divide profit between countries. The rules ask where people take the important decisions and carry the risks, and if software takes a growing share of those decisions, that test becomes harder to apply. The decision that matters also tends to move earlier in time, to those who design the system and set its limits, and that is where the rules will have to look. Some writers already ask whether a 'significant robot function' should sit next to the familiar significant people function.

Then comes the most basic question. If a company is run by an AI agent and serves customers in twenty countries without a single employee, who is the taxpayer? The profit must belong to someone, so tax authorities will look to the people or entities behind the system, but that requires that they can find them. As far as we can see, such a company still has a legal vehicle and people or entities behind it somewhere, so the questions that follow are more likely to concern residence and ownership than a new kind of taxpayer. Dutch case law already points that way: a court of appeal treated the profits of a self-built crypto trading bot as income of the person who built and ran it.

Energy adds its own version of the question. A battery that buys power when it is cheap and sells it when it is expensive, or a car that feeds power back to the grid, turns a user of electricity into a seller. Is that a separate business, a service or a side effect of owning the asset? Tax rules on income, energy and VAT were not written with a car in mind that is also a small power station, and in the Netherlands the question becomes practical on 1 January 2027, when net metering for households with solar panels ends.

Borders, competition and cooperation: can international tax keep up?

Behind all of this lies the challenge of borders. Technology moves across them freely while taxation is tied to them. When countries react individually, companies that work in several countries can face double taxation and legal uncertainty, and countries that stand aside risk losing revenue. This is not a new problem in international tax. Since 2013, the Organisation for Economic Co-operation and Development (OECD) has tried to agree on where the profits of the digital economy should be taxed, and the plan to give market countries a share of the profits of the largest groups (Pillar One) has stalled. Artificial intelligence widens this old question in two directions: it reaches physical goods and machines, not only digital services, and it reaches any company that uses AI, not only the largest groups that Pillar One was meant for. Countries are negotiating a framework convention at the United Nations, but the outcome is open, and the rules that exist today still rest largely on residence and physical location, and much less on data, algorithms or the creation of value.

Digital services taxes show what this means in practice. In the absence of an international solution, around a dozen European countries levy a tax of a few per cent on the revenue, not the profit, that large groups earn from online advertising, online marketplaces and the sale of user data; the Netherlands has none. Because the tax is due on revenue, it is also due in loss years and usually cannot be credited under a tax treaty, so it adds to the total tax burden, and platforms have passed it on through higher seller and advertising fees, so smaller businesses pay part of a tax that was never meant for them. It has also become an international trade issue, since the United States regards these taxes as aimed at American companies and has threatened tariffs in return. For the new economy the question is which services they will reach next. An app that matches riders with driverless cars, a marketplace for robots as a service or an AI agent that places advertising can look very much like the services these taxes already cover.

Countries also compete. Some try to attract AI investment with friendly tax regimes or new legal forms for companies, and Europe has proposed a simple, digital European company form that so far contains almost no tax rules. The global minimum tax of 15% (Pillar Two) limits how far such regimes work for large groups, although it leaves some room for incentives tied to real activity. For the Netherlands, the opportunity lies in clarity, in the form of a compliance regime designed for AI companies, and not in a lower rate.

Visibility, fairness and timing

Tax authorities face a problem of visibility. Using an AI agent is so easy that registering every agent a company uses would be impractical, and it is not always clear which party should report what. Income earned through platforms and crypto transactions is already reported under European rules, but nothing shows how much work an AI agent did, or where. The same technology also offers authorities new ways to collect tax in real time, as the European plans for digital reporting of cross-border invoices from 2030 show, but only if it is used with care, because nobody wants a repeat of earlier cases in which automated decisions wrongly harmed citizens. And there is the question of who carries the cost. The use of AI draws on energy, water and grid capacity, while the benefits may flow to a limited number of companies and owners, which makes the tax question also a question about fairness.

Timing makes this more pressing. Tax law is usually made slowly, after a development has become visible, while the contracts and structures that companies set up today will still appear in their tax returns years from now. Some dates are already fixed, such as the platform work rules and the end of net metering. Positions taken without thinking about tax may have to be revisited later, when the rules have been clarified or changed.

What does this mean in practice?

Taken together these questions do not point to one answer. Some models will turn out to be ordinary businesses that existing rules can handle, perhaps with some stretching, as happened with the internet. Others will probably put pressure on the foundations: how public revenue is raised when less of it comes from wages, how to decide where a business is when it moves, and how countries cooperate when cooperation is difficult. Looking at the three assumptions, ownership appears the most manageable under existing rules, whereas the question of place appears the hardest, which is also the question on which countries have failed to agree for more than a decade. Which of the two applies will differ per model, and finding out requires looking at each of them instead of assuming one answer for all. The outcome will lie somewhere between two extremes: a world in which transactions are invisible, and tax is hard to collect, and one in which tax is collected automatically, in real time and fairly.

The practical point is modest: the economy is changing in ways that many businesses see only in part, and the tax consequences are not yet on most agendas. A useful first step for any company is to look at its own business through these changes, as a user, a supplier or an investor, and to ask where machines, platforms or subscriptions already do work that people or owners used to do. That picture makes it easier to notice when a tax question appears, and to ask it early. If the economy of the next decade is built on machines, access and software, what should a tax system designed for people, owners and places look like? Nobody can answer that today, which is why it is worth thinking about now.  

RSM is a thought leader in the field of supply chain management including ESG. We provide frequent insights through training and thought leadership, based on detailed knowledge of industry developments and practical application gained from working with our clients. RSM helps companies connect individual geopolitical and regulatory developments to the broader changes affecting their supply chains. By doing so, we support businesses in distinguishing short-term disruption from structural change and translating that assessment into coordinated decisions on sourcing, origin, landed cost, logistics and supply chain design. If you would like to discuss what these developments mean for your business, please contact one of our consultants.