While some companies will run their own data centers, most companies will use the cloud. It's the age-old trade-off of CAPEX vs. OPEX, but the convenience and flexibility come at a premium. I usually call this the cloud tax.
The cloud tax usually eats into margins – especially true the closer your service is to an underlying cloud primitive (the case for repatriation). But there are other interesting downstream effects of the cloud tax on how we buy, run, and architect software. A few implications:
- Passing through the cloud tax to customers: The emergence of the cloud-prem deployment model can be seen as passing on the cloud tax to customers. Instead of running infrastructure in a vendor's cloud account, they run it in the customer's cloud account. This not only has the benefit of a simpler security model for the customer but letting customers deal with the cloud tax instead of vendors.
- Companies that find "tax breaks": A crop of cloud-cost management startups aim to help customers reduce their cloud bills. Now that interest rates are rising, companies are looking for more ways to reduce costs. Cloud infrastructure is usually a good place to start – overprovisioned machines, unused resources, and other ways that companies can reduce their overall spend.
- Competitive pricing for cloud providers: In AWS Is Not a Dumb Pipe, I argued that AWS could compete in higher-level services partly because they had a significant cost advantage. When you are making a margin on the underlying infrastructure (the cloud tax), you can competitively price the higher-level services you build on top of them.