Amazon Web Services is likely to have 1.3 million servers that are more than three times more efficient than enterprise systems, data centers that use space better and generate better returns of about 20 percent.
Those high level takeaways come from an Oppenheimer research note raising Amazon's stock price target to $930.
Much of the Oppenheimer tag-team analyst note revolves around AWS and how it'll fuel earnings into the future. And yes, Oppenheimer analysts were upbeat on Amazon's e-commerce expansion plans too. But let's get real: AWS is the profit margin machine on team Amazon.
In a nutshell, Oppenheimer concludes that AWS' capital spending won't need to be as high as expected. Why? "AWS' competitive advantages in procuring, designing and architecting datacenters and compute/storage resources are driving even higher profitability and lower capital intensity than previously expected," said Oppenheimer.
Also: Public cloud computing vendors: A look at strengths, weaknesses, big picture | AWS Lambda garners interest, production workloads as serverless world evolves | Salesforce and AWS pair up: Here's what it means for cloud computing | Why Amazon is the king of innovation: AWS, a cloud above the rest | AWS at 10: A look at how Amazon revamped the enterprise cloud computing pecking order
Add it up and an additional 100,000 square feet of incremental data center capacity will run AWS about $376 million and generate a return of more than 20 percent.
This efficiency for AWS---and likely other cloud providers---is going to create a flywheel of scale. Oppenheimer's team of analysts noted:
We believe that Internet and the major cloud providers represent over 45% of total global compute capacity, which is likely to be closer to 65% in five years, at which point we believe will represent a steady state in terms of global capacity.
In fact, Oppenheimer is betting that AWS and Microsoft Azure will combine for 12.6 percent of the server installed base by 2020. If you toss in IBM and Google you see where this is headed: Server vendors will largely be selling to just a few big buyers.
If this efficient cloud model plays out AWS should be on track to produce $5 billion in free cash flow by 2018, up from Oppenheimer's previous estimate of $3.2 billion.
Oppenheimer's AWS revenue estimates are a bit stunning through 2023. Oppenheimer is projecting AWS to have 2016 revenue of $12 billion with $16.81 billion in 2017. In 2018, AWS revenue will be $22.2 billion jumping to $33.29 billion two years later in 2020. By time 2023 AWS revenue is in the books it'll hit the $57.5 billion mark.
The AWS flywheel effect isn't an entirely original idea. Evercore analyst Ken Sena handicapped Google's cloud investments of late, but ultimately concluded that the AWS lead was too much. Sena said:
Amazon's level of investment and geographic scale is conservatively nearly 5x that of our current Google estimate,looking at just the last two years, and many times that of other providers. Moreover,as it currently stands, we see Amazon's lead in market share, capabilities, and geographic scale as developing a flywheel effect whereby enterprises and/or developers can adopt core AWS functions like Compute and ultimately transition into newer services such as Redshift (a data warehousing service), QuickSight (businessanalytics), or Lambda/Elastic Beanstalk, making AWS a fuller-service PaaS offering as opposed to simply infrastructure. Analogously, as Amazon has attracted a host of managed service partners on top of AWS, the ecosystem around their products has inevitably grown stronger.