Global AI Adoption in 2025 - A Widening Digital Divide

The Global AI Adoption in 2025 report by Microsoft analyzes the diffusion of generative artificial intelligence tools during the second half of 2025. Data comes from aggregated and anonymized telemetry, adjusted for operating system market share, internet penetration, and population.

Global adoption reached 16.3% of the world population, increasing by 1.2 percentage points. The report identifies a gap between the Global North (24.7%) and the Global South (14.1%), with a difference of 10.6 percentage points.

The United Arab Emirates leads with 64%, followed by Singapore (60.9%), Norway (46.4%), Ireland (44.6%), and France (44%). The document analyzes 130 economies. The United States, despite leading in infrastructure and frontier model development, registers 28.3% and dropped from 23rd to 24th place, showing that technological leadership does not guarantee mass adoption.

South Korea rose from 25th to 18th place, reaching 30.7%. Growth is attributed to three factors: government policies (reconstitution of the National AI Strategy Committee and AI Basic Act in September 2025); improvements in Korean language models (GPT-4o and GPT-5); and a viral phenomenon in April 2025 when Ghibli-style images generated with ChatGPT-4o achieved massive diffusion on social platforms.

The document examines DeepSeek, an open-source platform under MIT license offering services at no cost. It registers adoption 2 to 4 times higher in Africa, with market shares of 89% in China, 56% in Belarus, and 49% in Cuba. This diffusion is attributed to elimination of economic barriers and promotion through technology partnerships.

The analysis includes data from the Edelman Trust Barometer 2025, placing AI trust at 67% in UAE versus 32% in the United States. The report examines determining factors: digital infrastructure, government investment, AI training, and availability of models in local languages.

Key points

• Analysis of 130 economies based on aggregated and anonymized Microsoft telemetry, adjusted for internet penetration and population
• Global generative AI adoption: 16.3% of world population
• North-South gap: 10.6 percentage points (24.7% vs 14.1%)
• UAE leads with 64%, followed by Singapore (60.9%) and Norway (46.4%)
• United States drops from 23rd to 24th place despite leading in infrastructure
• South Korea rises 7 positions (25th to 18th place) with 30.7% adoption
• Korean success: government policies, language model improvements, and social virality
• DeepSeek: adoption 2-4 times higher in Africa than other regions
• DeepSeek reaches 89% in China, 56% in Belarus, and 49% in Cuba
• AI trust: 67% in UAE vs 32% in United States
• Key factors: digital infrastructure, government investment, and models in local languages