Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
Cloud

Amazon's AWS Shows Signs of Weakness as Competitors Charge Ahead (bloomberg.com) 25

Amazon Web Services basically invented the cloud computing business and once held nearly half the market. That dominance is slipping. AWS captured 38% of corporate spending on cloud infrastructure services last year, down from almost 50% in 2018, according to Gartner. Microsoft now grows its backlog of corporate sales faster than Amazon. The company that brushed aside incumbents and transformed an internal startup into Amazon's profit engine now faces internal bureaucracy that has slowed it down.

Bloomberg interviewed 23 current and former AWS employees who described management layers that proliferated after a pandemic hiring binge. One sales engineer who was six managers from Jeff Bezos before the pandemic found himself fifteen rungs from CEO Andy Jassy earlier this year. AWS hesitated to invest in Anthropic when the AI startup was spending most of its cash on Amazon servers.

Executives doubted the Anthropic AI could be monetized and were culturally reluctant to pay for external technology they believed could be built in-house. Google invested in early 2023. Amazon followed that September with $4 billion in commitments. On Thursday, Google said it will supply up to 1 million AI chips to Anthropic.

Amazon's AWS Shows Signs of Weakness as Competitors Charge Ahead

Comments Filter:
  • by haruchai ( 17472 ) on Friday October 24, 2025 @04:12PM (#65748412)

    and mine isn't particularly favorable, AWS is a FANTASTIC achievement.
    for it to have been built by a glorified department store is downright ASTOUNDING when NONE of the tech industry titans had a comparable offering still boggles my mind.
    their stocks should have dropped to $10 the day AWS launched.

    • The idea was good, but being the first mover is always brings it's problems. Competition is now eating AWS' lunch with regard to usability. Maybe it starts to shine when you are running huge deployments, but if the first impression sucks and a client goes elsewhere because of it, they will not be coming back, and you lost your chance.
      • by haruchai ( 17472 )

        " Competition is now eating AWS' lunch with regard to usability"
        which?
        I haven't looked at any except Azure in years and it looks like I'll soon be living & breathing Azure as my org is looking to dump VMware for Azure Local

        • Hello, former AWS user, current Azure user - and not by choice.

          Azure fucking sucks. It's missing basic things like being able to rename a resource.
          Seriously, you cannot rename a VM in Azure.

          It's also a buggy mess.
          • by lucifuge31337 ( 529072 ) <daryl&introspect,net> on Saturday October 25, 2025 @10:23AM (#65749742) Homepage
            I'm not saying a agree with it, but for your very first complain you need to understand: you're doing it wrong. That's not how public cloud is intended to be used. VMs are herd not pets. There should never be a reason to rename a VM. You spin up another and move on.

            "But I need it to....." Nope. You're doing it wrong. It's not what that system was designed for and if you want to use it you need to design your workloads to operate within the theory of how it's intended to function. It's really that simple. Things will continue to be hard as long as you continue to attempt to use the wrong tool for the job at hand.
            • you are the one thinking it wrong. not everyone needs to deploy their vm under an elastic load balancer that automatically replaces their instances on an autoscaling group if they go down, and on top of that, do immediate failover to a different DR region.

              heck, not even AWS does that. there use cases for both. everyone that went down past monday didn't have all of what i mentioned above. why? it's not easy, quick and cheap to have all of that. there's room on the cloud for all use cases.

              • Please go learn some more before trying this again. Or just keep digging in and looking foolish, whichever works for you.
      • Agree 100%. As a current user of both Azure and AWS, I can attest that both are clunky and often difficult to use. But at least Azure generally has actual *screens* for editing settings, not forcing you to edit json config files that you have to read a manual to find out what keys do what.

  • What does Anthropic do that Bedrock can't?

    AWS's big sell is that it is a 'one-stop-shop' for all your infrastructure and hosting needs. My bet is when the dust settles the GenAI applications that are actually valuable will get migrated to where the rest of the resources are, especially as any capability gap shrinks.

  • by leonbev ( 111395 ) on Friday October 24, 2025 @04:24PM (#65748442) Journal

    I've been a long time AWS user, and I noticed that their quality is slipping. Their tech support is becoming increasingly unhelpful, and escalating to your TAM is useless because they can't do anything to help solve the problem.

    I also noticed that unexplained small outages are becoming more frequent, along with the larger outages like we saw in us-east-1 last week. Worst yet, the cost for compute is actually starting to creep upward in terms of cents per hour for server runtime, after years where it constantly dropped.

    If someone asked me where to set up a new greenfield IT deployment now, I'd probably tell them to go to Azure instead.

    • AWS is a terrible place to work (by design) and has a brain drain problem. The system is now so complicated that the people remaining no lnoger have sufficient intitutional knowledge to reason about it in an effective manner, so issues are being sorted out by rediscovery rather than knowledge. This creates consistent new issues as code is deployed that nobody thought could cause issues in a system they don't truly understand.
  • Most value in today's market comes from controlling the supply. I suspect AI will be a commodity purchased based on cost with no one really able to control the supply. That will likely mean that a lot of the investment being made will turn out to be of much less value than anticipated. Surgeons are expensive because the supply is controlled through licensing. Once they are replaced with AI and a robot, the cost of surgery is going to be determined by how easy it is to build a robot using AI, not how much su
    • You are assuming that AI will be a commodity which will be in demand. If that's not the case, your argument completely breaks.
      • AI already is a commodity in demand for good or ill. My point was that doesn't guarantee a return on the current investments.
        • The dotcom boom and bust of the 2000s shows that what one era thinks will become a commodity can be very far off the mark.
  • metastasized tech debt
  • by laughingskeptic ( 1004414 ) on Friday October 24, 2025 @04:48PM (#65748508)
    AWS revenues have been growing at + ~$6B per year for a few years and that has not changed. Just because MS is currently growing faster because they have a bunch of locked-in customers whose easy-path is Azure, doesn't mean AWS is "weak". Since Azure growth is coming from a largely fixed pool of customers, it will taper off at some point. There are plenty of things to not like about AWS' corporate culture, but the business seems to be firing on all cylinders when it comes to revenue growth.
  • They can't be all that weak if half the web goes down when one of their servers does ...
  • Is the article suggesting that AWS isn't as good as it once was (a valid argument), or is it claiming that not backing AI is causing it to slip (a much less valid argument)?

"People should have access to the data which you have about them. There should be a process for them to challenge any inaccuracies." -- Arthur Miller

Working...