Concluding the theme of concept-related posts this week, let’s talk about scale. Top of mind is an excellent article by Ben Thompson titled “The Amazon Tax.” In it, Thompson outlines his thoughts on the way the “tax” business model for Amazon Web Services (AWS) is both being copied into other parts of Amazon’s business as well as being a reflection of the core drive for growth imbued within Amazon’s very DNA as a company. But it was in conjunction with much of the “cost of the cloud” content of the CIO Forum that I attended last week that really anchored the topic of scale in my mind. Why is scale so singularly important in the cloud computing realm? And is there any reasonable way to prevent an oligopoly or monopoly from forming when taking the argument to its logical conclusion?
Let’s start with the first question: “why is scale so singularly important in the cloud computing realm?” In essence, the more computing resources that are pooled together, the less of an impact any one customer can have on the available resources for others. It’s analogous to the value of diversification in the stock market – an index of the S&P 500 will prevent the drop in stock of any one company from drastically dropping the value of your investment. The key to getting to those diversification benefits with lots of computing resources is having enough paying customers to maintain an ever-growing infrastructure. The more customers you bring in, the more computing resources in which you can invest, and this in turn allows you to bring in more customers. It’s a virtuous cycle for ‘cloud inertia‘.
That cloud inertia brings us to the second question: “is there any reasonable way to prevent an oligopoly or monopoly from forming when taking the argument to its logical conclusion?” The short answer is: probably not. Market share is a key element to the success of any cloud business. The efficiencies of scaling massive data centers make the cost-per-customer drop significantly. So in addition to the drive for acquiring more customers to grow one’s business, the profit margin per customer exceeds the incremental cost of overhead (generally-speaking). In large organizations, getting too large can be a hazard because of the management overhead. But when the ‘resources’ you’re managing are servers and switches and not people necessarily, scale doesn’t have the same magnitude of drain on management resources (there are some, of course).
The double incentives of company growth (common for any profit-seeking business) and practically unlimited economies of scale provides a powerful incentive for the growth of cloud computing. And in relation to the ever-increasing and rapid growth of companies like Amazon, that edge of higher efficiencies of scale will eventually push out smaller competitors who simply can’t compete with the prices from companies like Amazon. (It doesn’t help that Amazon has managed to convince shareholders that long term sustained growth is an ample substitute for profitability – profits from the company get directed back into growing the company even further).
The final straw is that “cloud computing” is the ultimate commodification of computing resources – no one cares where you are located, what hardware you’re using, or how your main office lobby looks – providing reliable, cost-effective computing resources is the only name of the game. When taken together, it’s hard to imagine a scenario that doesn’t end up with only a few large cloud computing vendors. Mom and Pop cloud computing just isn’t going to be a thing. And while we’re relatively early in the arms race of cloud computing vendors, don’t be surprised when your choices are someday limited to two or three options … or maybe just one.