WhatsApp, Twitter, Facebook all started life without a business model, and they all made it to greatness by building a huge subscriber base via network effects over time. So why shouldn't you do the same: Don't worry about making money, just grow the subscriber base until it is huge, and then figure out how to monetize the install base?
The inconvenient truth is that these companies are all outliers in the land of startups. Most startups work their way to greatness over a period of 5 to 10 years through hard work, perseverence, and constant iterations based on market feedback. If you start out with no business model, you are basically ceding the first couple of years of potential business model iterations to the competition. Why lose that incredibly valuable opportunity to learn from the customer when you have a small and loyal install base that will work with you as you try new things?
The other inconvenient truth is that all these three companies started a long time ago in startup years: 2009, 2006 and 2004 respectively. The world was a very different place back then. What it took to build a business is very different because consumers are at a whole different level where it comes to savviness about new products and services, and because investors are raising the bar every year on what constitutes a fundable startup. Increasingly startups (even on the west coast) are starting to look at their options where it comes to balancing growth with profitability.
This 2019 Hackernoon story about the counterintuitive choices Bill Boebel, Founder and CEO of Pingboard, had made about this balance, and offers another view than the common startup wisdom of chasing growth above all else.
Our recommendation to startups is always the same: Come up with a hypothesis for a business model from Day 1, and then get out of the building as quickly as possible and start to test and iterate it in the field. That is usually the most cost-effective way to lead you to a scalable and repeatable business model while spending the least possible elapsed time.