Some are very niche paid businesses like SaaS or privacy-focused mail that provide very specific and uniquely tailored services to a small set of clients. These are probably the easiest to understand, as at their core is a simple, logical transaction. However, they don't defeat companies like Google at scale due to the fact that not many people want to pay for bits that they already pay their Internet provider to deliver, and the fact that accommodating everyone's needs in such a business is overly costly.
All the gargantuan scalable platforms are not really selling a technology - technology is the vehicle connecting customers to their service. Their real money maker lies outside the Internet, whether it is shipping and merchandise (Amazon), or selling an audience (Google/Facebook). I find it to be a similar scenario to McDonalds's famous franchise business model of owning the land the burger is made on, while not selling the burger itself. If you want real customer service, you need to find the local restaurant in your area and pay more (Plausible/Fathom, Hey.com, Fastmail).
A common trend in response to this kind of environment has been to offer some services for free and drive users to pay more for specific use cases if they need to (especially power users and enterprises). However, doing this kind of monetization wrong can easily lobotomize your business and push users away, as seen with the ongoing news media crisis and paywalls. And it also turns out that companies used to one kind of business model find it difficult to pivot to others, for example with Google's continual failure to make paid services.
In all cases, when looking at the Internet you need to examine your customers (or your needs) and focus on accessing them, even if that means tearing down any conventions you already have. This can apply to anything from the news media looking to fully digitize, or independent bloggers looking for analytics that serves them.