I believe renting dedicated servers is often overlooked. You pay someone else to install hardware, ensure network connections and be on-site for hardware swaps etc but still have the maximum degree of flexibility.

Even larger companies can work well with that model, traffic also tends to be cheap enough that you can spread across different vendors to avoid lock-in. And in that case, your sysadmins can sit wherever they want, no need to be physically close to the servers.

Also, as there's much less knowledge to be a dedicated server provider, competition is strong and prices are comparably low.

Many providers will now also bring up dedicated servers for you so fast, or offer you API based provisioning of single-tenant VMs or similar that it's really rare that the difference relative to cloud providers becomes much of an issue.

I used to spin up dedicated servers and then put an overlay network + a simple set of tools to spin up containers on them years before Kubernetes etc. was a thing, and we'd have a "global" (we had VMs in Asia, dedicated servers in Germany and colocated own servers in the UK) unified deployment mechanism that let me spin up containers wherever with a one-liner. Having a few extra dedicated servers with spare capacity standing by still made the whole system far cheaper than e.g. AWS, even if you attributed my entire salary towards it (I spent nowhere near all my time keeping that running).

It's easy enough to find consultants that can set up systems like this that abstracts away the dedicated hosting providers so you can mix and match and move with ease - especially today with options like Kubernetes.

If I was to go back to doing consulting I'd probably look at finding a way of packaging this kind of offering up behind lots of marketing speak and offer some sort of "abstract" hybrid private cloud layer on top of a choice of dedicated hosting providers to make that kind of hosting palatable to execs who refuse to believe the cost saving potential because they've never dived into the actual numbers (oh, the amount of time I've spent building out spreadsheets with precise cost models that'd get promptly ignored because someones had heard from a friend that company X swore vendor Y was cheap and believed it blindly)

I don't have first hand knowledge but my understanding is large companies have procurement departments and they get over half off sticker price for Azure. My guess is this is why the sticker price needs to be overinflated because people in procurement need to show that they are doing their jobs.

Also it is a major pain point getting anything done with IT operations.

Like the Oracle database server that half the department relies on stops responding on a Friday morning and it takes all day to determine the hard disk is full and fix it. I had never before worked at a company where this happened multiple times.

Or operations saying they were unable to restore a windows server hosting a database server and now everyone has to scramble to update their connection strings because operations somehow cannot use the same domain name for the new machine.

It's true there are huge rebates to be had if you're big enough, which is one thing smaller companies should bear in mind when they look at big company X using cloud provider Y as justification for thinking Y must be cheap.

If you're Netflix, cloud is probably not that much more expensive than owning your own servers. Maybe even cheaper. But you're not getting Netflix prices.

But even if you're small fry you should however start regularly talking to your provider and go through a regular cost-cutting exercise and talk to them about how you're looking at provider Z and have been asked to cost out managed servers and on prem options.... You won't need to get very big before that starts paying off.

If your competition is doing this and you're not, and hosting costs starts becoming a big part of your cost base, you won't be able to compete.

Long term I think we're going to see disruption here to the point of startups failing because of competitors copying their idea but being better at driving down hosting costs by not being afraid of going to dedicated hosting or hybrid solutions (hybrids are my favourite - if your stack can be deployed semi-transparently both on dedicated servers and cloud you can go much closer to the wire on your dedicated servers by being prepared to spin up cloud instances to take care of spikes; ironically having the ability to spin up cloud instances makes relying on cloud services even less cost-effective)

I'd also expect to see more "hybrid" cloud offerings with companies offering you operations-as-a-service by giving you a virtual cloud type interface where they don't actually own a cloud service themselves but helps you abstract away cheaper hosting providers. You can already find plenty of people who'll e.g. run Kubernetes setups for you, so taking the step to do more cost-optimization on the backend is natural (and I'm sure there are people who'd do this for you today - if I was still doing contracting I certainly would be offering that - and maybe someone is already wrapping it up as a service offering; I haven't kept up on that market)

> I think the case is for big companies that have a hard time attracting IT talent.

Companies that have made a name for themselves by outsourcing to the cheapest IT contractor that will promise them the moon and fill the seats with barely warm bodies? I was one of those bodies so I know exactly why they can't attract talent - they don't bother, and don't reward it. They treat IT as a cost center and are surprised when they get disrupted. The only good options in those companies are to work on the business side or worst case as a project/product/program manager interfacing with the warm fungible contractor bodies.

Many Enterprises are only alive because of inertia and goodwill from earlier decades.