Let's suppose you have a really good idea for a business.
Not just "we're going to be an easy-to-use centralized middleman for a previously unorganized and possibly unrecognized market" which has been done enough times that half of your success will be in getting customers to recognize you and the other half will be in not screwing up -- where was I? -- but a really good idea. The sort of thing that could make your grandkids into spoiled rich parasites who vote Republican.
It's true, you might not need a lot of money, or people, or resources -- at first. All the really good ventures end up needing all of those things in order to cope with the demand. And the big thing is people. You can trade money for an awful lot of other things, but somewhere in your great idea is a core competency that cannot, by definition, be outsourced. If you can outsource it then all you are is an organizer, and someone else will out-organize you. Odds are not in your favor.
Different businesses will be able to outsource different things. Accounting is popular. Every business needs to keep their books straight, manage purchasing and payroll and taxes. Hiring an accountancy firm part-time is usually a good move, because you don't have enough business to justify a full time accountant yet but you do have the need. If you're successful, you get to reduce your relative cost by hiring an in-house accounting team to replace a service that needs to cover its own overhead and profits. Unless your really good idea is about accounting, in which case you should not outsource it.
Cloud computing is a big deal these days because managing production- quality computing systems is expensive. Most businesses do not need to own redundant data centers, and cannot justify the expense. Therefore, they pay relatively more for a slice of a service run by Google, Amazon, or some other reputable company. Note that Google and Amazon are in the business of leasing out computer services because they already had business requirements to own the data centers and the networks and the rest of the infrastructure, so expanding that and charging for it is a good idea -- for them. At a surprisingly small scale, AWS and GCE become much more expensive than doing it well in-house. It could be that your really good idea requires a level of responsibility or security that a cloud service cannot readily sell you, in which case you should not outsource it at all.
If you make a physical product, you might outsource the production, but you must not outsource the design. What do they call someone who makes lots of money on an outsourced design? A great marketing and sales team. What do they call someone who make reasonable profits on that? A good factory. There are lots of good factories and good marketing and sales teams around the world. If you can outsource your design, they don't need you.
Figure out what makes your product or service much better than anybody else's version, and do not outsource that. Every so often, evaluate whether you are doing well enough that a particular function will be less expensive or more profitable if you bring it inside.