plate 216

Baghdad's phones now download faster than Erbil's

Average mobile download speed in Speedtest results, Mbps, 25 km around each city, 2019 vs Q4 2025–Q3 2026.

Rising cities
  1. 1. Jakarta, not Tokyo, is the world's largest city
  2. 2. Kinshasa will soon be bigger than Paris and London combined
  3. 3. Six of the ten fastest-growing big cities are African
  4. 4. Sub-Saharan Africa's cities are now bigger than Latin America's
  5. 5. Nigeria's cities now hold more people than Japan's
  6. 6. Town dwellers are now the majority in 19 sub-Saharan countries
  7. 7. Lusaka grew bigger than Zambia's 13 other old cities combined
  8. 8. Lagos took about a tenth of Nigeria's city growth
  9. 9. Karachi doubled to become the world's 11th-biggest city
  10. 10. The Mekong Delta got 3% of Viet Nam's population growth
  11. 11. Iraq went from one city of a million to eight
  12. 12. Nearly half of Mongolia now lives in its capital
  13. 13. Most of East Asia's big cities lost people since 2020
  14. 14. Nearly half of America's metros have more over-65s than children
  15. 15. Tijuana added almost as many domestic passengers as Mexico City
  16. 16. China's third air hub is now Chengdu, not Guangzhou
  17. 17. India's smaller airports took almost half of its air growth
  18. 18. Mexico's million-domestic-passenger airports more than tripled to 24
  19. 19. Baghdad's phones now download faster than Erbil's
  20. 20. Algiers now has North Africa's fastest home internet
  21. 21. Nouakchott's mobile internet sped from Africa's slowest to top 30
  22. 22. Caracas home internet has caught up with Mexico City's
  23. 23. Three in four Bangkok residents now live on a floodplain
  24. 24. Nearly 20 million Shanghai residents live in a flood zone
  25. 25. Six bigger cities of 2000 now trail Cairo
  26. 26. Brazil's fourth-biggest local economy is an oil town of 197,000
  27. 27. Mexico City took 2% of Mexico's new international flyers
  28. 28. Hyderabad is now India's fourth-busiest airport, ahead of Chennai
74.4 vs 46.0

Baghdad's average mobile download now beats Erbil's, after trailing it more than fourfold in 2019 (3.2 vs 14.4 Mbps).

23×

Baghdad's rise since 2019. Six of the ten fastest risers among 899 cities tracked are Iraqi; the median MENA city outside Iraq rose 4.6×.

Q4 2024

Baghdad has led Erbil every quarter since, and Mosul, Kirkuk and Basra now beat it too: Iraq's mobile edge has moved south.

the numbers & method
Baghdad's phones now download faster than Erbil's. Source: Ookla, Speedtest Global Fixed and Mobile Network Performance Maps (open data), 2019–2026.
Download speed (Mbps)2019 (grey bar) (Mbps)Times faster (×)
Amman96.516.35.9
Baghdad74.43.223.1
Mosul69.62.428.5
Al Hillah68.51.936.1
Kirkuk68.12.626.3
Najaf58.81.930.5
Basra57.42.622.3
Cairo51.316.13.2
Sulaymaniyah51.218.22.8
Erbil46.014.43.2

Claim types Fact · Interpretation

Speedtest® by Ookla® Global Fixed and Mobile Network Performance Maps, mobile quarterly tile files (Apache Parquet, s3://ookla-open-data; documentation github.com/teamookla/ookla-open-data). Platform-reported, crowdsourced Speedtest app results with GPS-quality location only; not a measure of network coverage. Read from the progrss vault (table city_ookla), which loads Ookla's own quarterly files. Checked: the four 2019 and the four latest quarterly files (2025 Q4, 2026 Q1–Q3) are listed in Ookla's public bucket s3://ookla-open-data, and the vault's file keys match them; two city-quarters were recomputed straight from Ookla's parquet files and match the vault exactly (Nouakchott mobile 2026 Q3: 458 tiles, 3,466 tests, 122.78 Mbps; Caracas fixed 2019 Q1: 940 tiles, 42,348 tests, 4.25 Mbps). Ookla defines avg_d_kbps as the average download speed of all tests in the tile; progrss weights tiles by their tests, so each figure is the mean of all tests within 25 km. Method (progrss calculation, same for all four plates in this set): a city is every zoom-16 tile (about 610 m) whose centre lies within 25 km of the city point; each period value is the tests-weighted mean of avg_d_kbps over all tiles and all four quarters of the period ('2019' = Q1–Q4 2019; 'latest four quarters' = Q4 2025–Q3 2026). These are means of tests, not the medians Ookla publishes in its Speedtest Global Index, so they will not match that index. Mobile, 2019 → latest (tests): Baghdad 3.21 (5,040) → 74.37 (44,683), quarterly 2.7, 2.5, 2.6, 7.0 → 71.5, 70.1, 76.1, 77.6; Erbil 14.36 (9,470) → 46.04 (17,940), quarterly 10.9, 10.3, 16.6, 18.1 → 42.4, 43.1, 46.8, 51.2. Baghdad first passed Erbil for one quarter in Q2 2022 (74.4 vs 41.4), fell back, and has led every quarter since Q4 2024 (46.7 vs 46.2). Others: Mosul 2.44 → 69.62; Al Hillah 1.90 → 68.47; Kirkuk 2.59 → 68.10; Najaf 1.93 → 58.80; Basra 2.58 → 57.44; Sulaymaniyah 18.19 → 51.24; Amman 16.34 → 96.52; Cairo 16.06 → 51.33 (Riyadh, 285.7, left off the chart for scale). 2019 test counts in Iraqi cities are small (Mosul 1,216, Najaf 997), and Q4 2019 fell during protests and internet cuts. Derived by progrss: multiples = latest ÷ 2019 (Baghdad 23.1×; Erbil 14.36 ÷ 3.21 = 4.5× Baghdad in 2019). 'Six of the ten': among 899 cities in the vault with at least 400 tests in 2019 and in the latest four quarters and tests in all four latest quarters; Iraqi cities rank 1, 2, 4, 6, 9, 10 (Al Hillah 36.1×, Najaf 30.5×, Mosul 28.5×, Kirkuk 26.3×, Baghdad 23.1×, Basra 22.3×); median of 55 MENA cities outside Iraq 4.63×; all 899, 4.71×. Iranian cities drop out: Ookla's 2026 files hold no Iranian tests. Context, not checked by progrss: Zain Iraq says 4G was approved for federal Iraq in early 2021 (company-reported); Kurdistan's operators ran 4G earlier. 'The edge has moved south' is an interpretation. Attribution: Speedtest® by Ookla® Global Fixed and Mobile Network Performance Maps, accessed from https://registry.opendata.aws/speedtest-global-performance. Based on progrss analysis of Speedtest® by Ookla® Global Fixed and Mobile Network Performance Maps for Q1 2019–Q3 2026. Ookla trademarks used under license and reprinted with permission. Licence: CC BY-NC-SA 4.0, non-commercial only: allowed on plates, never in a paid pack, licence or data product.

license this plate