Monopoly

Artificial intelligence, not taxes but concessions to curb Big Tech

With regard to large digital companies, Europe should undertake a different strategy than taxation

by Andrea Bonaccorsi

26 May 2025

Aggiungi Il Sole 24 Ore
ai preferiti su Google

4' min read

Key points

How is the value calculated?
Is taxation the right way?
The concession route
The Internet lesson

4' min read

In the context of the upcoming tariff negotiations with the United States, the topic of taxation of large digital companies, so-called Big Tech, has reopened. As is well known, the top seven digital companies have a market capitalisation greater than the GDP of many countries but pay little tax on their profits. The debate immediately took a political turn (pro- or anti-taxation, pro- or anti-US), which does nothing to help address the problem.

As is now clear, digital enterprises are structurally different from the rest of the production system, both on the demand and supply side.

How is the value calculated?

The demand for digital services is in fact subject to very strong network externality effects: the value that users derive from services does not only depend on their quality, but on the overall number of those who have adopted them. The value of Facebook is not only given by its technology, but by the fact that we find all our friends there, just as on Linkedin we find all our colleagues. This effect means that once a digital service has reached a critical mass of adopters, it attracts the whole market to it.

On the supply side, digital firms are not subject to what economists call the law of diminishing returns, generated by the saturation of plant capacity. Digital firms have incurred huge fixed costs to create the initial technological infrastructure (and many in fact have gone bankrupt), but once these costs have been absorbed they can expand production without limit, at negligible marginal costs. As Hal Varian (in the meantime become Google's chief economist) explained back in 1999, in the digital world the cost is incurred to produce the first unit, but the reproduction cost from the second unit onwards is negligible and tends to zero.

The combination of these supply and demand conditions generates an unprecedented situation: there is no upper limit to the growth in size of digital companies, which therefore inevitably tend towards monopoly.

Is taxation the right way forward?

Can taxation be the tool to regulate the market power of Big Tech? It seems that there is an economic problem that comes first and is completely overlooked in the debate. Foreword: at the birth of the Internet (mid-1990s) it was said that in the knowledge economyy there is a new production factor, in addition to capital and labour, namely an intangible factor represented by knowledge obtainable from data.

Thirty years after the birth of the Internet and twenty years after the entry of Big Tech, it must be recognised that digital enterprises use all three factors of production, capital, labour and knowledge, but only remunerate the first two.

Why do they not pay for the knowledge embedded in the data? My explanation is as follows. When digital companies developed their business model, they offered their services for free or almost free, in exchange for the right to use all the data produced by users. All our purchases, Internet searches, movements, content posted on social networks - all this is acquired by digital companies at zero cost. It must be admitted that at the beginning of the Internet era, this exchange was fair and reasonable: a contract accepted by all.

Over the last decade, the picture has changed dramatically. Once the initial critical mass has been exceeded, the economic value of data no longer grows linearly but exponentially. According to Metcalfe's law, once the initial threshold is exceeded, each new piece of data enters a pre-existing data network to which it brings increasing value, which is appropriated by companies through digital business models. This blows up the initial contract, because users now surrender data whose value has increased enormously, but to which no payment from Big Tech corresponds.

The concession way

The legal paradigm must therefore be changed: data are not a matter of private law, but of public law. They are a public and indivisible good, because they contribute in a systemic and integrated way to the generation of value. If we accept this idea, then the contract between individuals and Big Tech must be renegotiated. Data are a production factor that must be remunerated.

But how to do this, given that the data are produced in a fragmented manner and in real time? How to attribute them to individual producers? I do not think there is any point in defining property rights, which would only increase transaction costs without any real benefit. Rather, it is a question of activating the publicistic legal scheme of the concession. Data is a public good whose owner, on behalf of all citizens, is the state. The state recognises the usefulness of the public good being used by private parties, as happens with the concession of telecommunications frequencies, or airspace, or kilometres of beaches. But private parties, in competition with each other, pay for the concession of the public good. It is not a question of taxation of profits, as the concession must be paid for well before the profits are generated.

The Lesson of the Internet

If European governments were to put a regulatory innovation of this magnitude on the table, the balance of power with Big Tech would change. Not least with a view to preventing monopoly positions from being repeated today in artificial intelligence applications. Europe did not make the same economic policy mistakes during Covid as it did in the 2008 crisis: it learned from its mistakes. In the regulation of artificial intelligence, it should learn from the mistakes it made in the days of the Internet, when a lack of regulation left the field open to US-made business models, placing Europe in a subordinate position.

It must be said that the legal and political reflection on data has not helped to bring the problem into focus. It has focused, with a private law and individualistic perspective, on the protection of certain rights that may be violated by the use of data, such as the right to privacy. There has been a lack of ontological and legal reflection on the nature of data. Perhaps there is still time to change perspective. A new wave of monopolies on artificial intelligence can be avoided.