Here's a for instance: If I were to create a new social network today, I might host it on Heroku, send email thru Sendgrid, store my data in Neo4J and make it searchable with ThriftDB, etc etc. There is only one consumer-facing brand, but hundreds of small companies contribute to creating that brand. But each of them also have hundreds or hundreds of thousands of other customers, which creates the vertical economies of scale.
This is not new, corporations have had sub-contractors and suppliers throughout history. But what is new is that the small one-man shop will have a dozen suppliers from day one. It's a brand new ecosystem still in it's infancy, it didn't even exist to this extent when Facebook and Youtube was started. What existed then was the infrastructure services like server colocation and CDNs. So they relied on open source software, but they still had to deal with a lot of messiness you don't have to with SaaS.
This is the embryo of that eco system. Give it a decade to mature and my bet is there will be a one man operation totally competing with Google on their core business.
Oh wait, that already happened.
You don't see the same effects with little pizza shops, etc, because their reach is usually gated by locality, but the "flatness" of the internet causes many more "winner take most/all" scenarios.
A Swedish email company may very well co-exist with Sendgrid, catering to other Swedish companies. The Swedish startup Kundo competes successfully with GetSatisfaction, in Sweden. A company specializing in sending out the same email to many people instead of one-to-one, may co-exist with Sendgrid, just as Mailchimp is.
Create a matrix with the dimensions [every thinkable service a company needs] x [the number of possible functional specializations for each service] x [the number of geographical markets] x [the number of possible target groups in each geographical market] x [the number of companies that can co-exist catering to each target group] and you get a search space that can fit as many startups as there are sands in Sahara. Even with a very low fill-factor of that matrix there is room for millions of companies to co-exist over the entire planet.
The debate is not to what extent there are huge numbers of market opportunities, because it's a fact that there are. But whether a larger corporation is more fit to compete within that matrix than many smaller ones. That balance is rapidly shifting. The flatness of the Internet does cause a lot of "winner takes all" scenarios, but it also makes it possible for the winner to be a small company.
A company like Google still has to do PR to compete with Sendgrid if they create their own email sending service. Yes, they have their brand recognition to leverage, but it's possibile for Sendgrid to say "Google has mediocre customer services and may be a great search company, but we do only what we do best and that is one-to-one email sending with excellent customer service".
And people are increasingly discovering that it is on a personal level more gratifying to be the founder of a small company competing for a very specialized market opportunity than to be an employee in a large company competing for a very broad market opportunity.
Huge companies exist because historically communication inside an organization was cheap while communication between organizations expensive. Guess what, with the Internet that doesn't apply anymore. Many small companies can innovate faster than one big, and contribute to faster growth. It's the big companies that are overdue with the reality check.
They are dinosaurs dying in a climate more fit for leaner creatures.
(Oh, if there is such a need for talent, why doesn't the bigger companies scoop up some talent by doing acquisitions? Valuations shouldn't be to steep in early phases. And it wouldn't take long for the entrepreneurs to get lobotomized, sorry I meant institutionalized, and be happy cogs in the big machine)
</rant> ...sorry.