Once upon a time, Amazon was a dot-com-era technology company best known for selling books. Then, in 2003 and 2004, Amazon wanted to streamline its internal process between the programmers and the hardware engineers. It was a move that many other companies were taking, but an Amazon engineer had a brilliant idea: Why not use the same project to design an application that could rent chunks of Amazon’s computing facilities to customers?
On August 24, 2006, the public beta of Amazon’s “Infrastructure as a Service” (IaaS). And so, the ability to rent computing capacity managed by someone else was born. It was a gamble that has, so far, paid off. Amazon includes IaaS revenue in a larger unit called Amazon Web Service, which includes other cloud products. That, in turn, is under a part of the financial reports that includes non-cloud products called “other.” Besides Amazon Web Service, Amazon’s other revenue includes non-retail activities, such as marketing and promotional activities, co-branded credit card agreements, and other seller sites. Yet most analysts studying the industry believe that the mass majority of the “other” is cloud computing, and the growth is stunning:
IaaS at Amazon went from a thought project in 2004, to a startup in 2006 and it is almost certainly heading toward a billion-dollar business, if it is not there already.
Renting computers, called servers, that were managed by someone else, somewhere else, was not a new concept. What was new was the pricing that allowed customers to buy servers by the hour with a click of a button. That rental concept allowed businesses with uncertain future demand to buy computing capacity rapidly, 24 hours a day, as needed. Other companies followed Amazon’s lead. Rackspace, Terremark, CSC, Savvis, among others have similar options now, and technology research-firm Gartner estimates that businesses will spend $6.2 billion, or 45.4% growth, in 2012 on IaaS.
Here’s why server space matters. In 2002, Friendster was the first site to introduce social networking to millions. The demand grew exponentially as news outlets jumped on the new phenomenon, yet the code behind the site was asking too much of the servers. (In comparison, Facebook uses a programming paradigm called AJAX that taxes servers less.) Its popularity stressed Friendster’s computing capacity and eventually, for a period of time, the site became unusable because it was too slow. It is unknown whether Friendster could have innovated rapidly enough to remain dominate over the more nimble MySpace and Facebook, but it never had a chance. Other companies were forming while it was down. Years later in 2007, Zynga combined gaming with social networking into online applications that became known as social gaming. It first received venture capital in 2008 and by April 2009 it was the largest app developer for Facebook with 40 million monthly active users. In 2011, Zynga brought in 12% of Facebook’s revenue. Unlike Friendster, Zynga was able to handle the explosive demand of its service. It did so while not owning most of its own servers or hardware infrastructure. It was almost completely in the cloud. In Amazon’s cloud.
IaaS cloud computing is not always cheaper than owning your own hardware, but it provides liquidity for compute capacity. No longer does a company, researcher, or individual need to invest large amounts of capital to purchase hardware that will be used and amortized over three or four years. Now they can buy computing capacity by the hour. That allows websites to increase the ability to handle rapidly growing demand like Zynga did. Other sites have been able to avoid going down from 375 million page views a month for only $15 a month. In fact, the instructions to avoid a Friendster-like site slowdown are so easy to digest that someone right out of high school or college can use the technique. IaaS means that pioneering companies can start small yet still grow rapidly while other websites do not need to go down just because they posted something popular.
For startups, Infrastructure as a Service’s cloud computing increases the amount of money that can be returned to investors (salvage value), reduces the time to set up equipment, and cuts the amount of capital that needs to be raised. Servers and other computing components that run websites and are used for research no longer need to be bought outright, saving money and time. Indeed, most of the tech startups I know use cloud computing and it is exciting to see the rate of innovation increase as classical deterrents are removed. And cloud computing is not only changing business. Academic and corporate researchers can now rent the 102nd fastest supercomputer in the world, according to the Top 500 Supercomputer Sites through Amazon.
Ten years ago, if asked what company would revolutionize computing, a book merchant with a tech edge probably did not come to mind. However, that is exactly what Amazon has accomplished. Infrastructure as a Service is now widely available in many forms. While other cloud vendors have developed their own systems from scratch or are working with open (free) software, Amazon’s continues to grow rapidly and change the way businesses form and run—while reaping larger and larger revenue streams for vendors in the market.