General comments on AWS/Azure/Cloud moats
The link is to my entry for the September Ideas Comp: Amazon. As AWS is such a large portion of the market value of Amazon I thought I'd take some time and talk about cloud moats.
First, if you talk to any AWS engineer they will tell you they'd rather leave their company than be asked to work in a different cloud environment. I'm sure there are exceptions to this, but I've heard the same thing from 3 different people I talked to while doing my research for my post on The Cloud, one of whom was a senior AWS engineer at an f500 company who was involved in a mass migration effort of the company's internal apps from on-prem to AWS. I would guess that the same is true to a lesser extent for Azure, and to a yet lesser extent for GCP.
One comment he made to me stood out and speaks to the MOAT of the big 3 generally. He said that they expected to save money from the migration, but the migration was really about making life simpler, more scalable, and more elastic. Even still, they ended up saving more money by an order of magnitude than they had expected. As long as the value proposition of the cloud is so strong the profits will be there.
Second, one of the core MOATs of the cloud is the cost required to build out the infrastructure. Not only does it require billions of dollars to build a data center, you must have data centers around the globe to service companies who sell their products internationally (which is basically everyone). Beyond the normal infrastructure costs, the MOATs are being strengthened (especially with Amazon) via the introduction of their own silicon (e.g. Graviton) - this is another capex intensive process that smaller businesses won't be able to compete with. Amazon alone will spend $30B in capex for AWS this year.
Third, despite what it may feel like reading financial news, there isn't any evidence so far (that I'm aware of) that companies are moving workloads en-masse from one cloud to another once they have been established. When an AWS customer ads Azure or GCP it is usually to do something else, but once a workload is established it stays in place.
Fourth, there are enormous economies of scale outside of the infrastructure costs. Optimizing the hardware stack (GPUs, CPUs, Asics, networking equipment), services (AI/ML/routing/Other), etc and being capable of handling in the most energy efficient way any workload, is incredibly complex. The R&D costs of optimizing everything can be amortized across the customer base - meaning that the cost of optimization drops per task as scale increases.
Fifth, these companies are better positioned than anyone else to add on new services. It's just like Microsoft with Teams - you don't need to have the best product to win the market, you just need a decent offering, and the convenience of having things in the same place will do the rest.
Lastly, as to the cloud upstarts - I think it's still too early to understand the impact they will have on the big 3. Most of these companies are just now hitting $1B in revenue. AWS should cross $100B next year.
No comments yetBe the first to add your insight!