Yes, I know that predicting the downfall of SaaS and SaaS providers is all the rage right now, and no - this is not an AWS doomsday prophecy. AWS still holds about 30% of the cloud market (although Microsoft is really closing the gap with 20% market share) and remains, at least in my personal opinion, a market leader in terms of stability, features and innovation. AWS is not the best at everything, but is the best at most things, and that how it holds a third of the market (and half of Amazon's profits).
Do you remember when AWS was pushing serverless compute? Seems so long ago, a time when we were all still spinning VMs on our vCenters. They were offering S3, when companies were still piling on VNX storage racks. I know, I was there, at EMC², when it was happening.
In the Beginning There Was Cloud
That kind of forward thinking approach put AWS in the lead, and allowed it to maintain it's lead over the years. A few years ago- cloud is all anyone talks about. SaaS companies rise. Server racks are a thing of the past. Everybody is developing a cloud-native something. Consuming and managing resources made simple, accessible, cheap.
Unfortunately for AWS, this business model was disrupted recently. AI came barging into everyone's life - and is threatening AWS' dominance. I predicted the SaaSpocalipse almost a year ago, without even realizing. I wrote how we would shift from consuming software to consuming the underlying service this software provides - using agents.
AAS
Being the innovative company that it is, AWS also noticed the same paradigm shift I have. AWS is telling us what it's focusing on, as a company. All we have to do is listen (and be a little bit observant).
So, lets look at the facts.
AWS has been expanding its Bedrock AgentCore offering, packing it with features and trying to make it into a one-stop-shop for all of your agentic needs. Want to create an agent? easy. Need to safeguard it? they've got you there as well. Need some tools for your agent? no problem - a full MCP suite is available for you.
On top of providing a toolbox for agent builders and administrators, AWS has some agents of its own. Perhaps you've heard about the DevOps Agent and Security Agent. They are Agent-as-a-service type of offerings (or - service-as-a-service, if you will). Instead of building an agent, AWS provides a working one for you, tailored to perform a specific task. Enterprise customers are moving away from leasing static infrastructure or basic software tools, and are instead leasing automated digital labor. Who needs WIZ when you have a security auditing agent which is, according to a customer quoted by AWS "reducing the typical testing duration by more than 90%"?
Another honorable mention are the agent plugins, "an open source repository of curated plugins that bring packaged AWS expertise directly into AI coding assistants" said Laith Al-Saadoon, Principal AI Engineer at AWS.
The pattern is clear. AWS wants to be your one-stop-shop for any and all agentic needs, just like it has become a one-stop-shop for anything cloud related. Thus, I propose a new conceptual name for the provider - AAS: Amazon Agentic Services.
Agentic Tides
The AI tsunami did not spare AWS. Even though they are still actively developing "legacy" cloud native services , the shift in focus is evident. Even new features, like Lambda Durable Functions, is clearly a direct response to the rise of specialized durable execution engines such as Temporal, Azure Durable Functions, Cadence, and Cloudflare Workflows. These competing platforms have explicitly targeted the orchestration of agentic workflows.
AWS wants to be a Swiss army knife when in comes to building anything agent \ LLM related. Any model, any complimentary service, on demand.
We can assume that AWS will continue expanding its AI-related portfolio, with even more ready to go agentic services and more tools for builders, striving to become an AI platform, by spending 200 billion dollars on AI infrastructure. To accurately conceptualize this scale, Amazon will be spending approximately $548 million every single day, or $21 million every hour procuring, powering, and deploying artificial intelligence infrastructure for AWS data centers globally.
Unlike Wall Street, AWS leadership views this monumental investment not as a speculative risk, but as an absolute, existential necessity.
Top comments (0)