A Mid-Year Network Expansion: New Cities, New Countries, Bigger Compute
Cloudflare’s infrastructure build-out has continued at a rapid pace through the first half of this year, despite ongoing global supply-chain and semiconductor challenges. The company has added ten new cities across four countries not previously on its network, and reports that its total computational capacity has doubled since the onset of pandemic-related lockdowns.
The expansion spans every continent except Antarctica, and includes several first-time deployments in new territories. These additions are part of a broader push that also encompasses experimental work with ARM processors, ASICs, and Nvidia accelerators at the edge.
The New Points of Presence
The ten new Cloudflare cities are:
- Tbilisi, Georgia
- San José, Costa Rica
- Tunis, Tunisia
- Yangon, Myanmar
- Nairobi, Kenya
- Jashore, Bangladesh
- Canberra, Australia
- Palermo, Italy
- Salvador and Campinas, Brazil
Several of these deployments fill notable geographic gaps. Tbilisi marks Cloudflare’s first presence in the Caucasus country of Georgia, while San José — brought online in partnership with NIC.CR, run by the Academia Nacional de Ciencias — is the company’s first point of presence in Costa Rica. Tunis extends Cloudflare’s reach in North Africa, and Yangon is the company’s first city in Myanmar.
Nairobi gives Cloudflare a second Kenyan city and brings the total number of cities in sub-Saharan Africa to fifteen. In Bangladesh, Jashore becomes the third Cloudflare city after Chittagong and Dhaka. Canberra adds a presence in Australia’s capital territory, and Palermo extends coverage to the island of Sicily, where it is already yielding measurable latency improvements across Italy. Brazil gains two new locations: Salvador, the capital of Bahia state, and Campinas — the only non-capital city in this batch. These are early steps in a larger planned Brazilian expansion.
This announcement is in addition to dozens of new cities added in Mainland China through Cloudflare’s partnership with JD Cloud, which the company says has been actively testing and deploying new locations since last year.
Measuring the Performance Payoff
Deploying a new city is just the beginning of the work. Because互联网 routing is unpredictable, a new point of presence doesn’t automatically translate to better user experience. Cloudflare tracks performance using the 25th percentile latency of non-bot traffic, a metric that approximates latency affected primarily by physical distance, filtering out the noise of automated traffic.

The results are visible in countries like Kenya, where the Nairobi deployment produced a clear before-and-after improvement in real user latency:

While internet latency variations are common — particularly in regions with heavy mobile traffic from non-fixed connections — the chart shows that the new point of presence not only reduced overall latency but also smoothed out many of the frustrating fluctuations users experience. Lower jitter, combined with shorter physical distance to the server, contributes to more consistent page loads.
These differences accumulate quickly. A typical site requires hundreds of individual asset requests, some of which happen serially, so the gap between 25 milliseconds and 5 milliseconds of latency can mean the difference between a page load measured in fractions of a second versus multiple seconds.
For users in or near these new cities, the benefits apply across Cloudflare’s products, from the free 1.1.1.1 DNS resolver to the millions of internet properties the company serves.



