The Three-Percent Illusion

The official picture of the world economy appears reassuring at first glance. A global recession seems avoidable, with growth remaining around 3% in 2026 and potentially accelerating again in 2027. Closer inspection, however, reveals a distinctly different structure. Growth is sustained by a few Asian economies and an American AI investment wave, while the improvement expected for 2027 presupposes a geopolitical normalization that currently lacks convincing visible foundations.

In a Reuters survey encompassing nearly five hundred economists, growth expectations were lowered in thirty-two of the fifty economies studied, while inflation forecasts for 2026 were raised in thirty-nine. The global growth consensus nevertheless remains at 2.9%, largely due to investments linked to AI. Meanwhile, growth expectations in the eurozone have fallen to 0.5%.

The apparent contradiction is resolved by the fact that the global average does not indicate a general upturn. It is a weighted outcome of diverging economic regions: a 4–6% growth zone in Asia and Africa, Latin America hovering around 2%, and a transatlantic region growing by 0–2% all fall into the same metric.

The Improvement Built into the Base Case

In July 2026, the IMF expected global growth of 3% for 2026 and 3.4% for 2027. This acceleration, however, does not simply follow from the continuation of current processes. The forecast assumes that the Strait of Hormuz gradually reopens, energy supply returns to pre-war levels by March 2027, oil prices moderate, inflation declines, and financial conditions loosen later on. Meanwhile, the positive demand effect of AI investments remains part of the baseline.

The OECD’s scenario, which accounts for a permanent disruption in energy supply, yields entirely different results. In this scenario, global growth is 2.1% in 2026 and only 1.8% in 2027. If we additionally remove the American AI-CAPEX contribution to the world economy on an accounting basis, global growth in 2027 could fall to approximately 1.6–1.7%. This is already below the generally accepted threshold for a global recession.

The difference is not merely a forecast error. The path above 3% incorporates geopolitical improvement and enduring technological investment demand; the path between 1.5–2% starts from the persistence of current disruptions and does not count on an AI dividend.

The Concentration of American Growth

US GDP grew at an annualized quarter-on-quarter rate of 2.1% in the first quarter of 2026. According to Federal Reserve calculations, investments related to data centers, software, IT equipment, and energy infrastructure accounted for 1.36 percentage points of this figure. After deducting imported hardware, the estimated net contribution is approximately 0.73 percentage points.

Without AI investments, American growth would not necessarily disappear, but its exceptional nature would largely vanish. Calculations corrected for imports show roughly 1.4% growth in the first quarter, and without the entire broad technological investment block, only 0.7%. In this picture, the core American economy moves much closer to weak European performance.

The investment itself is undoubtedly real. However, broad-based productivity gains remain unproven. Therefore, AI is not merely a potential positive effect but also a concentration risk. If corporate revenues do not justify the scale of investments, the consequences will not stop at data centers: they could affect technology stocks, consumption, credit markets, and East Asian exporters alike.

What Remains Without AI

The most important structural finding is that removing the AI boom would not affect the world economy uniformly. Growth models in the USA and exports from Taiwan and South Korea would weaken primarily. The fundamental growth mechanisms of India, Africa, Latin America, and the Middle East, however, would change little. These economies would continue to be driven by demographics, urbanization, infrastructure, domestic consumption, industrialization, regional trade, and raw material production.

China’s situation also differs from America’s. The country is not only building data centers but also developing semiconductor, electronics, energy, telecommunications, and industrial production capacities. Therefore, when evaluating investments, it is insufficient to apply only Western corporate return metrics. Technological learning, import substitution, supply security, and export market position are also yields for the Chinese state.

Furthermore, the Global South likely does not need to copy the American frontier model. For agricultural, healthcare, educational, administrative, or corporate tasks, lower cost, local language support, energy efficiency, and operability on everyday devices are far more important. China is already capable of offering such smaller models, hardware, network infrastructure, financing, and local adaptation as a single system.

Thus, a world without AI would not necessarily weaken the Global South’s growth most severely. Rather, it would reveal that the center of gravity for world economic growth has already shifted, while the performance of the transatlantic economy is propped up by a narrow investment cycle with yet-unproven returns.

Therefore, global growth around 3% is not necessarily a false number. It is, however, an average that hides more than it reveals.

The full analysis can be read after registration.

The purpose of Unus Multorum is not to tell readers what to think about a particular event. Its purpose is to reveal the connections that allow every reader to form their own conclusions.