Before we can discuss the cost of building cloud native applications, we need to turn back the clock to examine infrastructure before the cloud, namely on-premises systems. When an enterprise deems it necessary to establish or expand its IT infrastructure, a number of actions have to be taken regarding:

• Physical space: Finding space in the existing data center or buying/renting new space for the data center buildout.
• Electric power: This can be easily overlooked. Deploying a large scale array of servers requires large power conduits that may not exist in a new location or be available in the current space. In addition, redundancy and backup power generators are often required to continue critical operations in case of a power outage or disaster.
• Physical Security: Paramount in enterprise deployments, establishing physical security to the hardware supporting a data center requires key/badged access points, security personnel, cameras, and security equipment.
• Network connectivity: Depending on the site chosen, broadband connectivity may not exist or be of insufficient bandwidth to support steady state data center functionality. In most cases, data centers require a physically redundant network connection to support network failover scenarios. This may require digging, burying, and trenching another line on the part of the internet service provider (ISP), which may take months (in the worst case scenario, permits are required from the local and regional governing authorities to bury a single network strand).
• Cooling: Computing equipment generates lots of heat, and they need to operate within certain temperature ranges to maintain optimal performance. The newest data centers have passive cooling systems (if built in cold climates), but the vast majority require large air conduits that are built-in to move the waste heat from the computers and pull in cool air.

• Physical hardware: After all of the preceding points have been addressed, the actual compute, storage, and network equipment that gives the data center its business value needs to be ordered, shipped, received, racked, stacked, connected, tested, and brought online. This takes teems of data center operations resources to accomplish and dozens of man hours per rack.
• Staff: While all of the preceding points are being dealt with, the company in question needs to hire, train, and compensate a large number of staff to design, order, install, test, and operate all the items noted previously.

It's no surprise that data center buildouts are planned out months and years in advance. Large capital investments are needed to bring together the necessary teams and resources so that they can complete the build and become production ready.