AI-driven profits concentrated at top despite widespread global AI adoption

Artificial intelligence (AI) production spans the globe, but most of the value it generates remains concentrated among firms that control technology design and intellectual property (IP).

The profits from the AI boom are bottlenecked at high-value stages of the supply chain, according to the UN Conference on Trade and Development (UNCTAD).

AI chips and servers made up around one-sixth of global goods in trade in 2025, contributing 42% of the sector’s value-based growth, the World Trade Organization (WTO) projects.

AI server production involves a broad supply chain, from mining to semiconductors and from energy equipment to cooling systems.

Each country involved in the process derives varying economic gains based on the specific production activities it can participate in.

Countries like Chile and Peru supply raw materials for AI hardware, while South Korea and Taiwan dominate semiconductor manufacturing, and countries like Malaysia and Mexico lead in the final assembly.

For instance, the total value of the silicon, copper, and tin used to produce advanced server systems running AI models accounted for only 0.2% of the final sales price, according to UNCTAD.

Countries supplying raw materials critical to AI tech receive only a small share of the final product’s economic value.

UNCTAD analyzed the production process of the Nvidia GB200 NVL72 server system to reveal the disparity. The critical piece of AI data infrastructure is valued at around $3.3 million.

The research tracked which activities and supplier groups accounted for more than 87% of the total added value; of the around $2.92 million tracked, 68% was profit after deducting operational costs and taxes.

The research showed that the vast majority of the profit margin flowed directly to Nvidia, which designs advanced chips but outsources physical manufacturing.

The same report found that only 4.3% of the traceable value went to workers involved in physically producing the server system.

Meanwhile, workers not involved in physical production of the system, such as corporate managers and design engineers, captured a 9.9% share, bringing the labor share of the final product to a mere 14.2%.

The report found that while the AI boom increased mineral exports from developing countries, it has yet to boost their share of the wealth the supply chain generates.

Emerging markets that contribute to assembling and manufacturing the systems create jobs, but the economic returns from these activities lag behind those generated by high-value technology design.

The UNCTAD published its “Trade and Development Report 2026” on Oct. 9, urging developing countries to shift focus from attracting foreign investment to strengthening local suppliers and cultivating skilled labor.

It noted that penetrating the upper tiers of AI component production is very difficult for developing countries because the sector requires massive capital and faces geopolitical barriers.