Artificial Intelligence: an Asset for Democracies on the International Stage?
What We Ought to Understand
The image depicts the hand of a robot creating a human figure, in reference to Michelangelo’s painting in the Sistine Chapel. May 8, 2023 Source: https://capirossi.org/wp/wp-content/uploads/2023/05/technology-human-touch-background-modern-remake-creation-adam-scaled.jpg Licence CC 4.0-BY
We will be offering over the next two weeks a series of essays on the challenges posed by artificial intelligence at the international level. What follows is a brief general introduction.
This series of in-depth articles was written with Constantin Vaillant Tenzer, a researcher in mathematics applied to cognitive neuroscience and machine learning at the École normale supérieure Ulm (Paris Sciences et Lettres University). For more than four years, he has been working on improving training methods for artificial intelligence algorithms, in collaboration with French and American companies.
Some of our pieces will be paywall-free, but others will be exclusive to full subscribers of Tenzer Strategics.
Our aim is not to follow some passing trend, but to attempt to acculturate Tenzer Strategics readers who are not necessarily familiar with issues related to artificial intelligence. We do not intend to be exhaustive or provide a general overview of AI, but we felt it was necessary to better understand the subject of “artificial intelligence” in several of its components in order to grasp what was at stake internationally. It is not possible to address the intersections between the technical dimension of AI and geostrategic issues without understanding the very nature of AI. Readers familiar with the subject may wish to skip ahead to certain pages.
Fundamentally, AI is reshuffling the deck in economic terms and, consequently, in terms of power relations. It is shifting the balance of power between nations with strengths in this area and those with little or no such assets, with major consequences for the rest of their economies. AI is also changing the game in several areas of international relations: military strategy, influence politics or propaganda, standard setting, structuring of thought, etc. This does not mean, however, that it is bringing about a complete upheaval in this field. Not all of the emerging trends can be fully grasped at this stage, and we are still limited to making conjectures. Finally, AI is entering the confrontation between democracies and revisionist powers.
It is also necessary to assess the dangers and benefits of AI as accurately as possible today. Between the fantasies of a total revolution that would change humanity itself and the denial of the problem, there is room for an analysis of the reality of real risks.
The economic weight of artificial intelligence
To gauge the economic reality of artificial intelligence, it is worth recalling a few key figures.
The global AI market in 2025 was worth $371.71 billion. According to current projections, it is expected to be worth $2,407.02 billion by 2032, with a growth rate of 30.6%. Today, AI accounts for 4% of global GDP. Overall, digital technology accounts for 22% of global GDP. In Europe, the size of the AI market in 2025 is estimated at between $60 and $65 billion, with annual growth expected to exceed 25% to reach approximately $338 billion by 2032.
Although some AI companies are currently operating at a loss, the sector’s revenues represent $41 billion in the largest market, the US, which is double that of the People’s Republic of China, its main competitor, which comes second with $24 billion.
But AI is not some kind of economic monad: it has a major knock-on effect on other sectors. It is estimated that the potential contribution of AI to the global economy by 2030 will be $15.7 trillion, more than the current combined output of China and India. Of this amount, $6.6 trillion would come from productivity gains and $9.1 trillion from effects on consumption. Already, the share of US growth attributable to AI is estimated at 92%.
This is also reflected in the size of companies. For example, Nvidia’s market capitalization on February 2, 2026, was $4.653 trillion, or about one and a half times the combined market capitalization of French or German companies and nearly four times the combined market valuation of Italian companies. This manufacturer of chips needed for AI is the world’s largest market capitalization. Next in order of market capitalization are Alphabet ($4.086 trillion, Google, Gemini), Apple ($3.834 trillion), Microsoft ($3.195 trillion, Azure, OpenAI), Amazon ($2.558 trillion, AWS), Meta Platforms ($1,812 billion, Facebook, Llama, MetaAI), Taiwanese semiconductor giant TSMC ($1,714 billion), and only in eighth place is Saudi Aramco ($1,663 billion).
A few European companies are attracting attention. For example, the valuation of French gem Mistral AI in September 2025 is $13.8 billion (compared to $350 billion for Anthropic and $830 billion for OpenAI). Its annualized revenue exceeded $100 million in early 2025, marking rapid growth (revenues multiplied by 25 in one year). The Dutch company ASML, which specializes in the manufacture of nanometer-scale etching machines, achieved a historic profit of $11.5 billion in 2025 on revenues of $38.3 billion. This result is directly fueled by AI-related demand for its advanced lithography machines, which are essential for the manufacture of AI chips. Its market capitalization has nearly doubled in nine months to $544 billion (as of February 3, 2026), by far the highest market capitalization in the European Union. Finally, SAP (Germany) has set a cloud revenue target of €21.6 to €21.9 billion for 2025, an increase of 26-28%. This growth is driven by its AI-focused restructuring strategy, which includes the elimination of 8,000 jobs to reallocate resources to AI.
Nevertheless, revenues directly related to the sale of AI products are low but growing rapidly: OpenAI’s projected revenues for 2025 are expected to be $12.7 billion, an increase of 343% compared to 2024 ($3.7 billion). Gemini’s revenues are only a few hundred million, but have tripled in a year, and Alphabet has invested more than $70 billion in the company. The cloud is growing by about 35% per year in terms of revenue.
Additionally, AI is entering the lives of individuals as well as businesses. The number of direct individual users of generative AI is 1.8 billion, including 600 million daily users. There are now 90,904 AI companies worldwide, including 29,618 in the United States, 8,178 in India, and 6,270 in the United Kingdom. 78% of global companies will use AI in at least one business function by 2025, up from 55% the previous year. In Europe, adoption in 2024 was very uneven: it rose to 70% among large companies in Finland, compared to around 33% for large companies in France and Italy.
Finally, in financial terms, the share of venture capital investment in AI is 50%. The World Economic Forum estimates that AI will create 170 million jobs and destroy 92 million by 2030. AI is currently destroying 78,000 jobs per year. According to an analysis by the International Monetary Funds (IMF), around 50.2 million Europeans are in jobs considered “at risk of replacement” by AI, representing 32% of the working population. However, the European unemployment rate remains stable at 5.8% in 2025, showing no signs yet of a wave of mass layoffs at the macroeconomic level.
The “cloud” is a key element
Cloud computing (the infrastructure on which almost all modern AI is based) is a colossal market in 2025, approaching $1 trillion, driven by “re-accelerated” growth thanks to insatiable demand for generative AI.
Global end-user spending on public cloud services (Infrastructure as a Service: IaaS, Platform as a Service: PaaS, Software as a Service: SaaS) would be about $723 billion. This represents a massive 42% increase over 2024 ($595.7 billion), with projections exceeding $1 trillion by 2026-2027. The specific growth of cloud services related to generative AI (GenAI) in the second quarter of 2025 was 160%. This is the main driver pulling the entire sector upward.
In this market, the top three—the “hyperscalers”—dominate everything. The pure infrastructure market (IaaS/PaaS, the “factories” of the cloud) generated approximately $400 billion in revenue in 2025. Three players share 63% of this pie.
The first is Amazon Web Services (AWS), whose market share is 30-31%. Its annual revenue for 2025 was approximately $117 billion. While it remains the historical leader, as cloud services represent most of Amazon incomes that it was one the first actors to provide cloud services in 2002, its market share is eroding slightly in the face of the rise of AI among its rivals. The second is Microsoft Azure (created in 2008), with a 20% increase in market share (it even increases by 26% if we include more PaaS/SaaS services). Its annual revenue exceeds $120 billion (for the global “Intelligent Cloud” segment). This company has the strongest growth among the leaders, directly linked to its partnership with OpenAI and the integration of Copilot. Finally, the third is Google Cloud (created in 2008), which holds a 13% market share. Its annual revenue in 2025 was approximately $61 billion. It has exceeded the break-even point and is accelerating (+34% growth in the third quarter of 2025), consolidating its third place far ahead of Alibaba or Oracle.
It is also worth looking at the breakdown of the cloud by type of service. SaaS (Software) represents $390 billion in 2025. It is still the largest segment in terms of volume (CRM, collaborative tools), but its growth (around 19%) is now slower than that of infrastructure.
IaaS (Infrastructure) represents $180 to $190 billion. This is the fastest-growing segment (+25-26%), as it is where companies rent raw computing power (GPU) to train their AI models.
Finally, PaaS (Platform) represents $208 billion. This segment includes developer tools and databases, which are essential for building AI applications.
It is important to note the key trend: AI as the “savior” of cloud growth. While the cloud experienced a slight slowdown in cost optimization in 2023-2024, AI reversed the trend in 2025. Cloud infrastructure is no longer used just to store data (which is inexpensive), but to compute AI models (which is very expensive and generates much more revenue for providers).
More figures can be found at https://epoch.ai/trends—a website we recommend exploring to better understand the extremely rapid expansion of the field and the scale of the AI race.
It is also crucial to understand that the industry is constantly evolving and that some things that were true six months ago are no longer true today and will no longer be true in six months. For example, between September 2025, when we started working on this series, and now, we have had to double some figures or, for example, reconsider the undisputed supremacy of OpenAI in terms of model quality and user traction, which has been seriously damaged. Nevertheless, most of the ideas we develop in this article are likely to survive for a long time: they were already true before the advent of generative AI in late 2022 and should remain so. The AI trend has only amplified, sometimes radically, realities and issues that have already existed in the digital world for many years. It is therefore likely that a reissue of this series in a year or two, given a similar geostrategic situation, would only require updating certain figures and perhaps adding a few proper names.
What we will be studying: outline of the series of articles
Our series of six articles, to which others may be added later, will be structured as follows.
In the first piece, we will present the economic, environmental, and geostrategic issues, but we will also endeavor to explain in simple terms what artificial intelligence is in its various components.
A second part is devoted to the material economy of AI. Some people often think that artificial intelligence is primarily a matter of the immaterial and service economy, whereas the reality of AI is first and foremost one of massive investment and heavy infrastructure. There is thus an AI “industry” that is every bit as significant as the heavy industries of the first two industrial revolutions.
A third part deals with what is known, mainly in Europe, although it also concerns other countries, as sovereign AI. This concern for sovereignty, which can be defined as emancipation from the uncontrollable influence of the largest countries, mainly the United States and China, is at the heart of the public policies that will need to be put in place in the coming years. It concerns hardware, chips, and data centers as much as it does access to the resources necessary for the functioning and development of artificial intelligence.
Our fourth part presents the issues related to the uses, but also the misuses, of artificial intelligence. After clarifying, in order to dispel certain myths, what AI makes possible and what should remain impossible, we show that some uses of AI are virtuous, while others pose considerable risks to democratic societies. It is from this perspective that it is essential to consider implementing regulations for AI systems.
The fifth part offers a general overview of the risks posed by artificial intelligence to societies and political life. Indeed, these internal perspectives cannot be separated from the geostrategic dimension of the subject, given the strong links that exist today between the cohesion and unity of societies and their ability to cope with threats and wars.
Finally, the sixth article in the series is directly devoted to how artificial intelligence affects international relations in the broad sense and security strategies. AI is changing the relationships between powers, influencing power games, and may also lead to increased vulnerabilities for democracies. It is also changing some of the core elements of military strategy. But AI in international relations should not be seen as a new utopia or a looming catastrophe—regardless of what the coming wars already portend.