In autumn 2025, air quality in Tashkent once again became a major public concern. In November, PM2.5 concentrations in the capital repeatedly exceeded the established limits.
On November 20, PM2.5 levels in Tashkent reached 202 µg/m³, according to monitoring data. This was around 3.4 times higher than the permissible hourly level of 60 µg/m³ reported by Uzbekistan’s environmental authorities. Residents were advised to limit outdoor activities. The problem was not limited to a single day. On November 11, PM2.5 concentrations reached 143 µg/m³, or approximately 2.4 times the same threshold.
IQAir data also showed that November was Tashkent’s most polluted month in 2025, with an average PM2.5 concentration of 75.5 µg/m³. Against this backdrop, Uzbekistan is introducing a new approach to urban greening: using artificial intelligence to determine not only how many trees should be planted, but where and how they should be planted.
Under a new presidential decree, Uzbekistan plans to launch an AI assistant called Yashil AI by March 2027. The system is designed to combine data on soil conditions, land quality, water availability, climate, satellite imagery and field observations to identify areas most suitable for greening. It will also help determine:
- which tree and shrub species are suitable for a particular area;
- how many seedlings should be planted;
- how they should be arranged;
- when they should be planted;
- how they should be irrigated and maintained.
The system is also expected to monitor seedlings after planting and identify risks such as water shortages, diseases and the potential loss of plants.
In other words, the focus is shifting from simply counting planted trees to determining whether those trees can survive and produce a measurable environmental effect.
Every year, up to 100 billion soums will be allocated to 14 cities and districts with the lowest levels of greenery. The funding will be used for planting trees as well as maintaining them.
This distinction matters because the number of seedlings planted does not necessarily reflect the actual result. A tree that dies shortly after planting contributes little to long-term urban greening. Yashil AI is therefore intended to support not only the initial selection of planting sites but also subsequent monitoring.
The “Green Lot” model
From January 2027, individuals and organizations will also be able to acquire land plots through the “Green Lot” system to create gardens. Construction will be prohibited on these plots, while the resulting gardens may be named after their sponsors. The idea creates another channel for businesses and individuals to participate in expanding green areas without turning such land into conventional development projects.
Uzbekistan has set a broader target of increasing the country’s overall greenery level from 14.2% to 30% by 2030. The government also aims to increase green space per capita and reduce concentrations of pollutants such as PM2.5 and PM10.
This puts the role of Yashil AI into perspective. Its value will not necessarily come from the number of recommendations it generates or trees it helps plant. The more important question is whether data-driven planning can improve where trees are planted, which species are selected, how effectively they are maintained and what environmental impact they ultimately produce.
Tashkent’s air-quality episodes in 2025 demonstrated the scale of the urban environmental challenge. The next stage of Uzbekistan’s greening policy is therefore moving beyond the question of how many trees are planted. The focus is increasingly on where they are planted, whether they survive and what they change.
