As we navigate the summer of 2026, the global economy stands at a pivotal crossroads. After three years of breakneck developments in Generative AI, the initial market euphoria is beginning to give way to a more sober—and perhaps unsettling—reality. Ludovic Subran, Chief Economist of the insurance giant Allianz, recently voiced strong reservations, speaking of an 'exuberance' that may not align with the actual macroeconomic data of the coming years.
The Solow Paradox 2.0
Subran’s warning is not merely a pessimistic forecast but an analysis rooted in historical precedent. He frequently references the 'Solow Paradox,' the observation by Nobel laureate Robert Solow in 1987, who famously stated: 'You can see the computer age everywhere but in the productivity statistics.' According to Subran, the same phenomenon is repeating today with Artificial Intelligence. While tech companies report record profits and investments in data centers skyrocket, the overall productivity of national economies remains stagnant or shows only marginal improvement.
Subran argues that the market has priced in a 'golden age' of efficiency that may take decades to fully materialize. 'The transition from adopting a technology to substantially altering GDP is a painful and uneven process,' he notes. The complexity of integrating AI into traditional workflows, the cost of energy, and the need for a radical re-skilling of the workforce are hurdles that Wall Street analysts tend to overlook in their optimistic reports.
An Uneven Distribution of Benefits
One of the central points of Subran's critique concerns the inequality driven by AI. Unlike previous industrial revolutions which, despite hardships, tended to diffuse their benefits across broader segments of the economy, AI seems to disproportionately favor early adopters and those with massive capital. Large multinationals have the capacity to fund the expensive infrastructure required, while small and medium-sized enterprises (SMEs)—the backbone of the European economy—are being left behind.
This 'productivity asymmetry' could lead to a two-tier economy. On one side, a technological core with extremely high profit margins, and on the other, a broad base of sectors (such as healthcare, education, and manufacturing) struggling with implementation costs and regulatory hurdles. Subran warns that if productivity gains do not diffuse throughout the entire economy, the risk of social unrest and economic instability increases sharply.
The Infrastructure Barrier and Energy Crisis
Beyond theoretical economic frameworks, practical obstacles remain. Allianz highlights that the demand for computing power has exceeded all forecasts, leading to a new form of inflation: 'chip and energy inflation.' The data centers required to train and run large language models consume vast amounts of electricity, at a time when Europe and the US are striving to transition to green energy sources.
- The shortage of specialized personnel to manage AI systems.
- The rising cost of GPUs and specialized semiconductors.
- Regulatory uncertainty (EU AI Act) delaying long-term investments.
- Geopolitical tensions affecting the technology supply chain.
Subran concludes that markets must prepare for a 'correction of expectations.' Artificial Intelligence will change the world, but not as quickly or as easily as investors would like. The real challenge for 2026 and beyond is not creating more powerful models, but integrating them in a way that generates real value for society as a whole, rather than just for Big Tech shareholders.
'We do not question the revolutionary nature of AI, but the speed at which it translates into prosperity for the average citizen. Exuberance is useful for innovation, but dangerous for monetary policy,' Subran states.
