Amazon averted a worst-case sales scenario during the final three months of 2022, results released after the bell on Thursday suggested. Revenue was $149.2 billion in Q4, up 9% and comfortably ahead of the $145.8 billion consensus expected. Excluding a $5 billion FX headwind, sales rose 12% YoY. The top-line beat was a relief. Last quarter, Amazon's guidance suggested revenue growth might slow even more sharply, prompting consternation among already nervous investors. Andy Jassy cited the
7 thoughts on “AAA Letdown”
I believe we are about to hit peak public cloud usage. As these lines of business fail to meet growth expectations, costs will be levered higher to try to make them more profitable. As the costs of the services come up for a business, they will be forced to calculate whether or not it is actually more cost effective to leverage a cloud service provider over building out their own on premises cloud. And eventually the answer for many will be it’s more cost effective in the long run to just buy and maintain your own equipment and migrate the now very portable apps back home.
I really couldn’t disagree more with the argument that anyone would look to stay onprem these days. In addition to infrastructure costs it’s important to look at the engineering skill set you want to hire, and I don’t think most companies should want to invest in the skill set required to design and maintain their own infrastructure. I also think the people you’d hire to do that are not as strong as the people AWS hires, and I think that the overall infrastructure design would be worse than what AWS designed.
Cloud providers like AWS have also figured out how to make difficult distributed technologies self-service. There’s no longer a need for Elasticsearch experts or Kafka experts or Kubernetes experts on the control plane side. There’s also no longer a need for DBAs or Storage Engineers at most companies. This sort of paragraph could extend on and on with examples of how onprem companies require specialist Engineers that are not as needed in Cloud based companies who are consuming the well designed services rather than designing and maintaining them. It also empowers the entire Engineering department to use technologies they want at the scale they want without needing to go through specialist teams and limitations of their designs.
So in the end I think the decision to go to the cloud versus staying onprem is not just about saving cost on infrastructure.
Your point is well made. I agree that in addition to the investment in the infrastructure there will also require an investment in engineering skillset. But with the major tech companies sending their talent to the job market, that skillset is easier to find. While I don’t disagree that you will need a broad range of skills, I do disagree that you need to a hire a bunch of different resources to fill those skill gaps. Over the past decade the emphasis on tech skill growth has been to make yourself more of a T and less of an I. This means that engineers now know how to write front ends and back ends with database components all leveraging CI\CD in a secure environment. Gone are the days where you require a person or team to represent each piece of the Tech stack. The modern day full stack engineer already knows all of the pieces or is able to piece together what is missing through available online resources. Taken with the also now more readily available and refined open source counterparts to the tooling you mentioned, it is possible to reconstruct an AWS, GCP, or Azure style setup with very little licensing cost associated with it. Given an economic downturn, more tech layoffs, and a focus on cost cutting for the sake of survival, I still believe my theory holds true.
However, I would like to caveat my synopsis with the disclaimer that we won’t see a wholesale departure from the usage of these cloud service providers. I think it will become more of a piecemeal hybrid configuration where you leverage public cloud where it makes sense and move away from them where it is more cost effective. I am not predicting the demise of Cloud, I’m just saying I think evaluating Cloud usage will get more scrutiny. But I do still believe that Cloud growth will continue to decline and that will mean pricing will have to account for that loss of growth.
CD – is it just possible that IBM is on target with their emphasis on hybrid clouds?
Might that shake the conviction of the IBM perma-bears, along with their strength in AI, quantum chip design and their new 2nm chip JV? (Reprinted from The Onion)
I think IBM is positioned well, especially because they now own Red Hat and that is essentially the Microsoft or Apple of the Linux world. Included with Red Hat is the Ansible Automation Platform which makes automating things easy enough for pretty much anyone to do it. They also own Open Shift which has both strengths and weaknesses as a hybrid cloud provider. I wouldn’t discount their future viability at all.
I think where we differ is I see what you’re describing as possible only via the Cloud. But I agree with what you’re describing with the large difference that I don’t see it as possible onprem.
I feel that I could go on and on about this sort of stuff, and I do on tech forums and I’m very active on StackOverflow etc. I know this is a finance forum and so I won’t derail everything. I normally get particularly interested in the articles and comments related to tech stuff I guess.
I guess I view anything as possible given the desire to execute. Granted, some things will remain just plain easier leveraging already created cloud tools, I just think that in some cases these tools will prove to be more expensive than they are worth when the cost dynamics start shifting. Add to it a reduction in headcount and some really impactful operational outages and I think the landscape will look very different in a year or two.