Of course when someone says startups are getting cheaper, one has to ask "cheaper than what?" Cheaper than during the dot com boom, sure, because people don't get fresh ice sculptures delivered to their office every morning anymore. But that's nothing new, we knew that for a while, so comparing today's startups to those of the dot com boom isn't very useful. A much more useful statement would be that startups today are cheaper than they were two years ago, but in my experience this just isn't true. The biggest expense of modern startups is paying the employees, and the salaries of great people (engineers and otherwise) are rising, so I'd argue that startups are getting more expensive relative to their cost a few years ago.
Most (but not all, of course) things that have huge value take a few years to build and a lot of careful work to market, which means that underfunding most startups is a terrible idea. This article does an excellent job pointing it out.
You could have a lean start-up that burns $10m a month.
"Getting by" is pointless if you haven't found your product/market fit. You need to invest in finding it until it's found. If that takes 100 people, then that's how many people you should hire (if you can afford it).
Horowitz agrees with sgblank and ries!