In its early years, much of Amazon Web Services' (AWS) growth came from the avid embrace of its bread-and-butter cloud computing and storage services by startups and cloud service providers such as Netflix (NFLX) , Instagram, Dropbox, Pinterest and Spotify. Quite often, these companies wanted to be spared the trouble of running their own data centers, and found AWS to be a convenient alternative.
Traditional enterprises that already had their own data centers were also spending money on AWS. But much of it was the result of individual teams and departments taking the initiative to use Amazon (AMZN) for certain workloads or projects, rather than large-scale deals that CIOs were signing off on.
Today, as AWS barrels towards a $15 billion annual revenue run rate, the story looks a lot different. Global 2000-type companies are making "strategic" decisions to move a large chunk of their workloads to AWS, and in some cases shutter many of their data centers along the way. GE and Capital One were among the companies to announce such moves last year.
This transformation has a lot to do with how AWS has evolved from a provider of basic cloud infrastructure (IaaS) services into a platform supporting thousands of apps and value-added managed services that collectively make it easy for enterprises to migrate critical workloads to the cloud and generally replicate much of what they've been doing within their own data centers.
This year's AWS re:invent conference, which runs from Monday to Friday in Las Vegas and is expected to attract thousands of partners, developers and customers, should provide the next chapter in the company's efforts to get big business to stop worrying and love the cloud. Another goal is to keep clients from giving Microsoft (MSFT) and Alphabet's (GOOGL) fast-growing cloud operations a look.
Fortune reported last week that Amazon plans to unveil a new managed database service at AWS re:Invent, as well as more advanced AI/machine learning services for developers. They would follow the August launch of encryption key, streaming data analytics and application load-balancing services targeting enterprises, as well as last year's launch (at the 2015 re:Invent conference) of database-migration services and a storage appliance (named Snowball) that lets companies quickly move large amounts of data to an AWS data center.
The database service will be based on the enterprise-friendly PostgreSQL database, and will complement Amazon's existing Aurora service, which is based on the MySQL database (popular with many startups and open-source developers). Aurora-based workloads produce over a billion dollars in annual revenue for Amazon, making it a thorn in the side of database giants such as Oracle (ORCL) and Microsoft.
The machine learning services could help protect Amazon's flank, given that Google and Microsoft have been steadily winning converts for their machine learning solutions. Google's TensorFlow software library, which underpins its machine learning efforts, has won a lot of developer support, and the search giant has also leveraged its AI expertise to launch application-specific cloud services for fields such as text-analysis, image-analysis and language-translation.
The Wall Street Journal, meanwhile, reports that shipping firm Matson is the latest enterprise to shut down its data centers and move their workloads to AWS. The paper also provides some interesting comments from Matson CIO Peter Weis: He says Matson has only been using AWS's basic computing and storage services, in part to make it easy to switch cloud providers should the company wish, but is intrigued by Amazon's reported plans to offer a PostgreSQL database service. "I don't think we'll go anywhere, and we'll probably use more of their tools," he adds.
The remarks shine a light on how Amazon's switching costs -- and perhaps also its pricing power -- are growing as more customers embrace the value-added services that either aren't supported by rival clouds or would take a lot of work to replicate on them.
This stems not only from Amazon's home-grown cloud services, but the many services provided by third parties through the AWS Marketplace. Some of these third-party offerings, such as a recently-launched analytics software solution from Tableau Software (DATA) , compete with value-added Amazon services. But both Amazon and the third parties generally seem comfortable with this arrangement.
There are some parallels here with how Amazon has increased Prime's switching costs by getting customers hooked on media services that would disappear if they cancelled their Prime memberships. For AWS, creating higher switching costs may be crucial to maintaining the platform's dominant position as Microsoft and Google price aggressively and otherwise pull out all the stops to grow their public cloud businesses.
Microsoft's Azure, widely seen as the world's #2 public cloud platform, saw its revenue rise 116% annually in the September quarter (no revenue figure was disclosed). And The Information reported today Google's cloud revenue is expected to more than double this year to over $1 billion. That still makes it much smaller than AWS, which might do close to $13 billion in revenue this year, but also suggests the Google Cloud Platform (GCP), like Azure, is growing faster than AWS. Amazon's AWS revenue rose 55% in the third quarter, and 59% over the first nine months of 2016.
Fifty-five percent growth is still pretty remarkable for a business of AWS's size. But with competition intensifying, continuing to grow at such torrid rates, or close to it, will require Amazon to keep landing enterprise mega-deals and creating the kind of "stickiness" for its platform that lets it avoid seeing heavy churn or be badly hurt by pricing pressure.
That certainly isn't easy. But given its recent moves, and the ones it's expected to announce shortly, Amazon's execution still appears to be rock-solid.