Often cited as the top driver for cloud adoption is agility. A common issue facing companies for decades has been long lead times in deploying hardware to their data centers, causing all kinds of issues, including the expansion of project timelines. Removing this constraint by having virtually limitless capacity and providing the ability to use that capacity within minutes is critical for an organization with high business velocity requirements. As the world moves toward an even faster flow of ideas due to the internet and social media, companies will be chasing those ideas and trying to monetize them for their business models. Since an idea or fad can often come and go in a few months, having to wait that long just for the physical resources means the opportunity could easily be missed. In the case of cloud computing, this is not an issue and allows the organization to develop a market strategy based on their core competencies, while allowing the cloud vendor to focus on commodity IT resources.

Failing fast is another area where cloud agility and experimentation is important. The disruption of ideas and companies is happening at record speeds these days and, for a company to be able to withstand that disruption, they need to be constantly iterating on their business model to match their customers' demands. To quote Jeff Bezos on the concept of an experiment – "It's not an experiment if you know it's going to work". This means that continuously trying new ideas, designing new products or services, and doing it quickly, is important. It also means that some of these ideas, probably a lot of them, will fail to fulfill the objectives of the business, and that has to be OK, since, every once in a while, they will succeed.

Cloud computing allows for this agility by enabling resources to be created and destroyed exactly when needed, or when no longer needed. It means that a minimally deployable system of resources, if properly designed, will grow exponentially along with demand and never put the company in a position where hardware is idle, or otherwise underutilized.