How did OpenAI crawl the web so effectively without getting blocked?

It didn't. Others did it, OpenAI used it. https://commoncrawl.org/

I'm pretty sure scraping is easier than building a groundbreaking LLM used by 100M people

It's not hard to scrape the internet

Very carefully