Summary of an article published on The Web Scraping Club on 2026-06-23 (free to read). Read the full article

Proxy IPs have to come from somewhere, and most of us buying them don’t know where. ISP proxies and residential exit nodes are the backbone of any scraping job on protected targets, and AI companies and every other industry that relies on web-scraped data depend on them. That makes the sourcing practices of proxy providers something we should understand, not just consume. Browser infrastructure is the other half of the problem: running browsers at scale is expensive and slow to spin up.

This is Episode #0 of a new series where I share the best things I’ve read about web scraping. This time I pick two pieces from Spur Intelligence, on how NetNut sources ISP IPs and how Smart TV apps become residential proxy exit nodes, plus an engineering post from Browser Use on rebuilding their cloud browsers on Firecracker microVMs. I add my own view from years of working with the big proxy names, and on how we should behave as guests on both the target and the proxy network.

What’s inside

  • How the DiviNetworks partnership puts real ISP address space on the market while actual subscribers stay on the same addresses.
  • Which detection signals Spur found for co-opted ISP networks, and the Howard University case as a real example.
  • Why proxy SDKs inside Smart TV and mobile apps work as a monetization model, and what the consent screens show users.
  • Why Browser Use runs Firecracker on plain EC2 instead of bare metal, and their case for fully headless over headful.

In the full article: The links to the Spur and Browser Use articles, the embedded webinar on ISP proxies, the specific figures and technical details I picked from each piece, and pointers to Scraping News and the Web Scraping Club Wiki.

Read the full article on The Web Scraping Club