Amazon Web Services VP Adam Selipsky sits down to talk about Amazon.com's DNA and its low-margin approach to cloud services vs. competitors.
"It will be extremely hard for most companies to drive to the cost structure Amazon has and will continue to improve upon. … We will absolutely continue to drive down our business costs and customer prices," he said.
Still, getting to the heart of AWS' balance sheet is impossible. Amazon delivers low-cost compute cycles because of the way it builds and operates cloud data centers, but Selipsky was not at liberty to divulge any details of how it designs the servers with which it populates them or how it keeps operations staffs small. How, for example, does Amazon solve the problem of chilling the equipment? Does it use the big air-conditioning units known as chillers or some lower-cost method?
"We don't talk about that. We have a lot of our IP tied up in our designs," he said.
That, of course, is in contrast to Facebook, which has published its server designs as part of the Open Compute Project and opened its data centers to inspection tours. Is Amazon watching its power usage effectiveness -- the amount of power actually used in computing out of the total delivered to the data center -- as Facebook is?
Of course, said Selipsky. "We analyze everything. We're highly data driven." But PUE – power usage effectiveness - is another stat he was not at liberty to disclose.
As a result, part of the belief in Amazon's "high volume, low margin" ethos has to be taken on faith. Parent company Amazon.com doesn't directly report revenues for AWS. In the first quarter of 2013, it reported $798 million in revenue in the "other" category, widely believed to mainly represent AWS but the accounting is not precise on whether the parent company's retail cash flow in part supports AWS operations.
Nor was Selipsky prepared to provide such an accounting. Rather, without mentioning names, he wanted to draw a line in the sand and dare would-be competitors to cross it. Amazon continues to build out data centers, adding per day the servers it once took to support the $5.3 billion retail business in 2003. Some of this addition could take the form of upgrading old servers to more powerful models but Selipsky said Amazon is also filling more data center space with equipment. "We continue to expand geographically. We're not done yet," he said. It's already got centers in Northern Virginia as U.S. East, and centers in Oregon and Silicon Valley as U.S. West and GovCloud. Overseas, it's in Dublin, Sydney and Singapore, as well as in two locations in Japan and one in Brazil.
Macquarie Capital says Amazon Web Services had likely revenues of $2.5 billion in 2012 and could be valued as a $19 billion company by 2015. As competitors reach their first billion or struggle to maintain stock value as growth slows, AWS is pressing its case.
High volume, low margin is part of every aspect of Amazon's business. "It's the way we designed our headquarters and offices," said Selipsky. Challenged to illustrate, he says he and other executives work from desks that were originally made from doors as a low-cost office-equipment option. "Those desks are a visual reminder of the early culture of the company," he said.
"It's difficult, retrofitting your company from a DNA perspective," he says to competitors. "If you're looking to make 70%-80% margins," you don't think about costs in such a manner.
The scale AWS has achieved and its pace of innovation will continue to be based on that culture. Catch us, Selipsky seemed to say, if you can.
Multicloud Infrastructure & Application ManagementEnterprise cloud adoption has evolved to the point where hybrid public/private cloud designs and use of multiple providers is common. Who among us has mastered provisioning resources in different clouds; allocating the right resources to each application; assigning applications to the "best" cloud provider based on performance or reliability requirements.
InformationWeek Tech Digest, Nov. 10, 2014Just 30% of respondents to our new survey say their companies are very or extremely effective at identifying critical data and analyzing it to make decisions, down from 42% in 2013. What gives?