Public cloud platforms have often been marketed as the cheaper option. In many cases the cloud is cheaper than the alternatives, but I think it was a mistake for the cloud industry to focus on price as the leading benefit. In an apples-to-apples comparison, the cost benefits of the cloud aren't always as clear-cut as many have assumed.
When companies are carrying out cost analyses of cloud platforms, the mistaken assumption that cost is the primary benefit of cloud deployments can lead them to dismiss the cloud option because the cost difference isn't as substantial as they'd hoped. But such comparisons are inapt: the cloud is not the same as in-house physical deployments.
Cloud platforms offer companies enterprise-grade servers and networks that they'd otherwise be unable to deploy without enormous upfront and ongoing costs. They enable companies to make those deployments in minutes and with substantially less expertise than they would otherwise need.
If you compare the direct costs of running a large network of cloud servers for three years with the direct costs of running an in-house network of servers of the same size, there is likely to be a price differential but it may not be enormous.
But, such a comparison misses the point of public cloud platforms.
Firstly, the flexible nature of the cloud means that you're highly unlikely to be running the full complement of servers continuously. With a physical network, you have to buy the servers, deploy them, administer them, and maintain them. Those are fixed costs entailed by physical deployments and they apply regardless of the utilization of the servers: idle servers accrue the same fixed costs as working servers. Cloud deployments are different. Users pay for what they use. Of course, a poorly managed cloud can be underutilized, but fixed costs are lower because the public cloud gives users on-demand access to compute and storage resources.
Secondly, the cloud offers capabilities that would otherwise be extremely expensive and labor intensive. The ability to quickly scale up or down with minimal effort is not to be dismissed. It allows companies to do things that would have been simply impossible with traditional physical deployments and it makes business models that would have once been uneconomical possible.
Netflix is the canonical example of a business that probably couldn't have existed and certainly couldn't have begun its journey without the cloud. Because of the radical shifts in viewing between primetime and fallow periods, to support the busy periods Netflix would have had to maintain thousands of servers that were underutilized most of the time. The cloud has allowed the company to scale daily as required. But it's not just big companies that benefit from rapid scaling and deployment: simply having the ability to deploy one server for testing or development without having to use IT procurement processes can change the pace of iteration and innovation within a business.
The cloud is a different beast altogether, and simple price comparisons with traditional infrastructure deployments fail to adequately capture those differences.