Pakistan’s mobile Internet blackout, seen from the network edge

Pakistan’s mobile broadband networks were taken offline after authorities ordered a nationwide shutdown on May 9, following the arrest of former Prime Minister Imran Khan. Cloudflare’s traffic data shows the scale of the resulting disruption — and the workarounds Pakistanis adopted. Peak country-level HTTP request volume fell by nearly 30% in the first several days after the shutdowns began, while the lowest daily traffic levels dropped by about a third. The composition of that traffic also shifted: request volume from mobile devices declined 60%, while desktop-driven traffic stayed roughly flat. Mobile’s share of peak traffic fell from 70% to 43%. Human traffic, as classified by Cloudflare’s bot score, dropped 40% at peak, while bot requests held steady.

Cloudflare’s view of Internet disruptions in Pakistan

Which networks went dark

On May 10, the Pakistan Telecommunication Authority (PTA) said Internet services would stay suspended for an “indefinite” period following a directive from the Ministry of the Interior to block mobile broadband. Cloudflare observed outages across the country’s four major cellular providers, with shutdowns beginning at various times on May 9:

  • Telenor Pakistan (AS24499) at 17:00 UTC (22:00 local)
  • Zong/China Mobile Pakistan (AS59257) at 17:00 UTC (22:00 local)
  • Jazz/Mobilink (AS45669) at 13:00 UTC (18:00 local)
  • Ufone/PTML (AS56167) at 20:00 UTC (01:00 local, May 10)
BLOG-1857 Embedded Image - 9og6CB

Although mobile use dominates Pakistani Internet access, fixed broadband appears to have absorbed much of the displaced traffic. Nationwide providers PTCL (AS17557) and Cybernet (AS9541) saw higher peak volumes compared to a week earlier starting at 17:00 UTC on May 9. Smaller local networks such as Nayatel (AS23674) and Wateen Telecom (AS38264) also recorded elevated peak traffic levels starting around 16:00 UTC the same day.

BLOG-1857 Embedded Image - LUiJ0l

The shift to fixed connections may explain a side effect visible in the data: median latency within Pakistan actually dropped after the mobile shutdowns, from a prior range of roughly 90-100ms to about 75ms, suggesting users moved to lower-latency wired links.

BLOG-1857 Embedded Image - G0JblB

Regional and city impact

Because the affected mobile operators serve all of Pakistan, no single administrative unit went completely offline — but peak traffic fell across the board. The largest loss was in Gilgit-Baltistan, where peak traffic dropped nearly 60%. Sindh saw a decline of about 35%, Khyber Pakhtunkhwa 30%, and both Islamabad and Azad Jammu and Kashmir approximately 20%.

BLOG-1857 Embedded Image - 6Jkxqr

The disruption was equally visible at city level. Peshawar in Khyber Pakhtunkhwa was hit hardest, with peak traffic down nearly 55%. Faisalabad, Karachi, and Multan each saw peak traffic fall by approximately 40%.

BLOG-1857 Embedded Image - 6ner3h

Censorship and the 1.1.1.1 effect

Alongside the mobile shutdowns, authorities ordered ISPs to block major social media platforms including Facebook, Instagram, YouTube, and Twitter. Testing by the Open Observatory for Network Interference (OONI) suggests the blocking combines TLS-level interference with DNS-based blocking. When DNS-based blocking is in play, Cloudflare’s 1.1.1.1 resolver typically sees a usage spike as users hunt for a path around the filters.

That pattern repeated here. Overall 1.1.1.1 request volume from Pakistan rose roughly 40% over the period. Requests for specific blocked platforms also surged. Traffic to facebook.com started climbing around 14:00 UTC on May 9 and more than doubled at peak. Instagram, also owned by Meta, followed a similar timeline and grew nearly 50%. Twitter requests began spiking earlier — around 08:00 UTC on May 9 — and grew as much as 150% that afternoon. YouTube requests increased by about 40% on May 9, and remained higher on May 10 than earlier in the week, though below the initial spike.

BLOG-1857 Embedded Image - FtG5LX

The cost of disconnection

The shutdowns carry real economic weight. Tens of thousands of “gig workers” and freelancers who depend on mobile connectivity have been affected, and many point-of-sale terminals rely on the same mobile networks. Transactions through Pakistan’s main digital payment systems fell by approximately 50% after the blackout began. Telecom operators have estimated the damage so far at Rs. 820 million — around $2.8 million USD.

Real-time tracking of such disruptions is available via Cloudflare Radar, with updates on @CloudflareRadar on Twitter or @cloudflareradar on Mastodon.