A year ago, Amazon Web Services (AWS) CEO Andy Jassy declared that his company builds out its offerings based on what customers are asking for, rather than what rivals are up to.
That customer-centric mindset feels very much alive today, as AWS unveils dozens of new services at its annual re:Invent conference that collectively address numerous customer pain points and strengthen Amazon.com's (AMZN - Get Report) hand in a slew of cloud services fields witnessing very strong developer interest and usage growth.
But while AWS' R&D decision-making might still be heavily informed by customer feedback, the public cloud giant's marketing message now seems to be more influenced by its ongoing battles with the likes of Microsoft (MSFT - Get Report) and Alphabet/Google (GOOGL - Get Report) .
Near the beginning of his re:Invent keynote on Wednesday, Jassy went out of his way to favorably compare AWS to Microsoft's Azure cloud platform, its largest rival. He shared research firm Gartner's estimate that AWS had 51.8% of the public cloud infrastructure services (IaaS) market last year, or nearly four times Azure's 13.3% share. He also noted -- just before unveiling a managed file system service for apps running on Microsoft's Windows Server OS -- that IDC estimates AWS is responsible for 57.7% of Windows public cloud instances to Azure's 30.9%.
In addition, with a lot of media coverage of Azure noting that its revenue growth rates have been higher than AWS' -- in calendar Q3, Azure's revenue grew 76%, and AWS' 46% -- Jassy suggested Azure revenue rose by $1 billion annually last quarter, much less than the $2.1 billion increase reported by AWS. The remark suggests Microsoft, which breaks out Azure's growth rate but not its revenue, had Azure revenue of about $2.3 billion, or slightly over one-third of AWS' $6.7 billion.
Indirectly, Jassy also sought to favorably compare AWS to the Google Cloud Platform (GCP), whose AI/machine learning services for cloud developers have been praised by Gartner and others. He declared AWS' machine learning services have twice as many reference customers as any rival, and that 85% of cloud workloads involving the popular TensorFlow machine learning framework (it originated at Google, but is open-source) run on AWS. Gartner estimates GCP had a modest 3.3% 2017 IaaS share, trailing Alibaba's (BABA - Get Report) 4.6% as well as AWS and Azure.
Meanwhile, as he went over the breadth of his firm's offerings, Jassy made a point of how no rival can match AWS in terms of the sheer number of solutions it provides in various fields. Among other things, he noted AWS now supports a whopping 175 cloud computing instances, and that its Lambda platform for serverless computing -- it still requires servers, but offloads the creation and management of server virtual machines to AWS -- is integrated with 47 services, 30 more than that of any rival serverless platform.
Certainly, the feature set lead claimed by AWS became even larger this week, as new offerings -- enabled by a large R&D budget that AWS' tremendous revenue base and profitability help justify -- have been announced at a dizzying pace. The biggest announcements include:
- AWS Outposts, a solution that enables hybrid cloud deployments by letting companies run AWS services on servers installed within their own data centers. AWS still manages the servers, which can have VMware's (VMW - Get Report) server, networking and storage virtualization software layered on top if a customer wishes. Outposts has a thing or two in common with Microsoft's Azure Stack appliances, which are sold by Microsoft's server OEM partners and used to locally run Azure services.
- New computing instances that rely on Graviton, an internally-developed ARM server CPU. Amazon claims Graviton-powered instances can deliver cost savings of up to 45% relative to traditional, Intel-powered instances for certain types of scale-out workloads.
- A slew of new services to aid machine learning work. They run the gamut from an AWS-optimized version of TensorFlow that's promised to more efficiently use GPUs (typically Nvidia's (NVDA - Get Report) ) when training AI models, to a solution for simplifying data-labeling when training models, to a solution for quickly adding GPUs to computing instances to assist with inference (the running of trained AI models against real-world content).
- New AI-powered services for analyzing medical text, making business predictions, deciphering the contents of paper documents and personalizing and recommending content.
- Inferentia, a custom chip for powering AI inference workloads that will be available next year. The product has some similarities with Google's Tensor Processing Unit (TPU), which supports both inference and training, and will compete to a degree against Nvidia's inference GPUs. Tech analyst Patrick Moorhead notes that unlike Google's TPUs, which only support TensorFlow, Inferentia also supports rival frameworks such as MXNet and PyTorch.
- Several new database services, including one (called QLDB) for creating an immutable, blockchain-like ledger with a central trusted authority, as well as a managed service for creating blockchain networks.
The public cloud market is still by no means a winner-takes-all game. Azure still has some important strengths, such as its hybrid cloud abilities, integration with locally-installed Microsoft software and ability to be bundled with traditional Microsoft enterprise deals. Google can still sell businesses on its powerful AI and analytics offerings and the developer-friendly nature of many of its services, and though it has seen limited traction elsewhere, Alibaba remains a very popular choice in China. Many large enterprises rely on multiple cloud providers, and will continue to do so.
However, between its massive and constantly-expanding feature set, as well as the huge ecosystem of software and services partners it has built, AWS is still very much in a league of its own, and quite often the best choice for a company that wants to do a large-scale migration to a public cloud rather than simply launch or move a few workloads.
And with Microsoft, Google and Alibaba's cloud efforts getting a little more attention these days, Amazon now seems more eager to remind businesses about these things.
This column originally appeared on Real Money, our premium site for active traders. Click here to get more great columns like this.