THE SURGE OF ARTIFICIAL INTELLIGENCE Dominique Marchese, 2025-11-05

Key words: AI, financial markets, GAFAM, productivity.

Version PDF

The first quarterly earnings reports from listed companies propelled US stock indices to new record highs, driven by the continued enthusiasm for artificial intelligence (AI). So far, the ongoing closure of numerous US federal agencies has had no impact on investor sentiment, reassured by the temporary trade agreement reached between Donald Trump and Xi Jinping. On the monetary policy front, the Federal Reserve cut its key interest rates by a quarter of a point (0.25%), while the European Central Bank (ECB) maintained its current rate.

Everything now revolves around AI.

In our previous monthly reports, we emphasized the fundamental importance of AI in the performance of financial markets. The main New York Stock Exchange indices, built on market capitalization, have become highly concentrated in a handful of technology stocks driven by a new industrial revolution with incalculable consequences for future economic growth and corporate profits. The market capitalization of the ten largest American companies is approaching 40% of the total market capitalization of large companies listed in the United States—a level unseen in forty years—far exceeding that observed before the bursting of the dot-com bubble (25%) in the early 2000s. The financial results of Nvidia, Microsoft, Amazon.com, Meta Platforms, Alphabet, and Apple now have an overwhelming influence on market behavior. The GAFAM companies, joined by Nvidia, the world leader in GPU chips designed specifically for AI-powered servers, seem to be single-handedly dictating investor sentiment. Investors have grown accustomed to trade tensions and abrupt policy reversals from the White House. Government shutdowns are barely on their minds. This undisputed dominance of the technology leaders is entirely deserved, given their profit growth in recent years and, above all, their contribution to the overall earnings of publicly traded companies worldwide. Following the investment themes linked to the digitalization of the economy, cloud computing, and big data, which captivated financial markets in the wake of the 2020 pandemic, the enormous progress in artificial intelligence achieved in just a few years has accelerated the revenue growth of the GAFAM companies and initiated a powerful investment cycle in technological and energy infrastructure that will extend over several years. The considerable need for computing power and hosting of AI-enhanced applications requires the deployment of numerous data centers which are currently struggling to meet the exponential demand from their customers.

One of the key stock market arguments in favor of tech giants in the 2010s was their ability to expand with little additional investment – a phenomenon known as scalability. Unlike many economic sectors, tech stocks experienced rapid growth without requiring significant capital investment, allowing them to generate substantial cash reserves and, in turn, fuel large share buyback programs. For example, acquiring a new user on a social network like those offered by Meta, or on software developed by Microsoft, requires virtually no additional investment or cost. AI seems to be radically changing the game. Heavy investments by technology companies are transforming the risk profile of a sector long praised for its low capital intensity, strong balance sheets, and high free cash flow generation. For the moment, investors remain oblivious to the risk of a bubble—a phenomenon that is not at all unusual during periods of industrial revolution that absorb available savings—and seem to be setting aside the crucial question of return on investment. The AI wave is too powerful to discourage them. The figures are indeed impressive. While the average annualized revenue growth rates expected for the four leading companies – Microsoft, Alphabet, Amazon.com and Meta Platforms – over the period 2024/2029 are 15%, 10%, 11% and 16% respectively (Bloomberg consensus), capital expenditures measured as a percentage of EBITDA will increase in 2026 to 49%, 57%, 72% and 90% respectively, compared to 26%, 34%, 101% and 66% in 2022. The efforts undertaken are in fact unprecedented in recent history. For these four AI leaders, total investments are expected to reach around $480 billion in 2026, a tripling compared to 2022, the year of launch of the ChatGPT chatbot, whose parent company OpenAI (unlisted) plans to invest several hundred billion dollars in the coming years.

Certainly, stock markets are showing high valuations, regardless of the valuation method used. Risk premiums appear unattractive from a long-term perspective. The yield on free cash flows of the New York Stock Exchange has returned to its level just before the 2008 subprime mortgage crisis. Except for emerging markets and the European stock market—despite a robust performance since January 1st whose valuation multiples are in line with their historical averages, the main global indices are forecasting lower long-term returns than those seen in recent years. But this does not take into account the artificial intelligence revolution, from which technology and industrial stocks exposed to infrastructure spending are fully benefiting as evidenced by the excellent quarterly results recently published. However, this revolution is still far from having its full effects permeating the rest of the economy. Optimism about productivity gains, potential economic growth and therefore the future trajectory of corporate profits rests almost exclusively on a technological revolution which it is essential to better understand in order to move away from irrational fears.

The hopes for productivity gains brought about by AI

Artificial intelligence (AI), already rich in history, is experiencing a dramatic acceleration thanks to major advances in deep learning, particularly convolutional neural networks (2010s), and the development of generative AI such as diffusion models, generative adversarial networks (GANs), and large language models ( ChatGPT launched in November 2022). These latter models, capable of understanding and producing natural language, are revolutionizing the relationship between humans and machines by making powerful capabilities accessible without in-depth technical expertise. Cloud-based platforms now offer tools for automatically creating models adapted to user-provided data samples (e.g., Microsoft's Azure Machine Learning and SageMaker). Amazon.com's Autopilot). AI developments will not stop at large language models, which currently only imperfectly meet the needs of machine planning and automation. New architectures under development aim to enhance the reasoning and planning capabilities of AI.

Without a doubt, AI is emerging as a powerful driver of productivity gains and potential economic growth worldwide. It is accelerating the digitalization and automation of tasks, far beyond the manufacturing sector. Industry, which represents only 10 to 17% of the gross domestic product (GDP) of OECD countries, has long benefited from a wave of Robotics, with varying levels of adoption depending on countries' capacity to integrate technological innovations – the world leaders in this area being in Asia (Japan, Taiwan, South Korea… without underestimating China's considerable efforts). AI now enables the rapid expansion of digitalization into services, which represent approximately 70% of the GDP of developed economies and have long experienced insufficient productivity gains. Furthermore, the development and deployment of AI agents, combined with big data and cloud computing, could offer an effective response to the challenges of demographic aging and labor shortages already affecting many sectors.

However, this revolution raises two major challenges: public education and access to abundant and affordable energy. In our view, the main obstacle to the widespread adoption of AI lies in the public's lack of scientific and technical knowledge, which fuels both Hollywood fantasies and irrational anxieties. Regarding the second challenge, we would like to remind you that every major industrial revolution in history has been accompanied by significant advances in the energy sector. The first industrial revolution, which occurred between the late 18th and early 19th century (the steam engine), was made possible by the abundance of coal. The second industrial revolution, which took place between the late 19th and early 20th century, benefited from a broadening of energy sources. Coal remained dominant, but oil and natural gas began to play a major role as primary fossil fuels. At the same time, the use of electricity (a secondary energy source) gradually spread. The third industrial revolution, which began around the 1970s (with the emergence of information and communication technologies), was characterized by a significant diversification of energy sources. In addition to fossil fuels, renewable energies, particularly hydroelectricity, have gained importance. Nuclear power generation took off. What will be the next energy revolution capable of meeting the enormous needs of AI while accelerating the essential decarbonization of electron production?

Conclusion

Faced with one of the greatest technological revolutions in history, investors cannot afford to stand idly by and watch the train go by. In our view, the question of whether a bubble exists is not the most relevant. The investment cycle in artificial intelligence infrastructure will extend over many years, while the digitalization of the global economy is only just beginning. However, in a global environment still marked by uncertainty—whether commercial or geopolitical, the high valuation of indices calls for greater discernment. The risk of new entrants challenging economic models, supported by the impressive capabilities of artificial intelligence, weakens stock market indices heavily concentrated around a few near monopolies. In this context, active management and diversification should be prioritized.