I can demo an alternative to Google Search in a weekend. I can throw comparatively small amounts of money to improve it (e.g. add more sites) such that it is extremely useful to find sites that I designate to have crawled (and it's not difficult to crawl the top N sites). My algorithms will be worse, my crawler less efficient, I probably will ignore javascript, my database will be larger than it needs to be, but none of that matters to an end user who wants to type in some text and get a list of results. My product would be worse in every way something can be worse, yet still satisfy the needs for a sizeable majority of users. It can scale out in known ways (way more expensive than google does it, slower than google does it, etc) by largely paying salary and hosting costs.

I absolutely cannot decide I want to install some poles to run electricity through a town without permits, insurance, inspections, and a ton of capital. I can't decide I want to build my own power plant and sell energy to my new grid without jumping through a lot of hoops, some of those hoops only a mountain range of cash can solve (compared to the merely large pile you'd need to run a credible regional alternative to google search for some segment of the population).