UK govt office admits ability to negotiate billions curbed by vendor lock-in

> But in practice - is this important? The hyperscaler clouds are similarly priced

Similar prices as advertised maybe, but big companies and governments do enough volume that they can negotiate discounts. If they're locked in they lose leverage though

The hyperscale clouds double in cost every 3 years compared to the prices dictated by Moore's law for hardware sold or rented in competitive markets.

Certainly, but you need stakeholders who are willing to make the investment, do the hard work over years, decades even, and not use it as resume building who are going to bounce to something else and leave a pile of garbage behind.

Are governments willing to keep technologists happy to run sovereign clouds with comp and working arrangements? Or are they going to treat them poorly and pay them poorly?

This is why mainframes still run while other workloads churns back and forth between on prem and cloud: hype cycle peaks, folks moved to cloud, found out it was super expensive and they didn't need the agility it provides, and now they have sticker shock and are trying to figure out how to come back on prem. But come back on prem to what? VMware gonna rake you over the coals, and there isn't a great on prem story besides say, OpenShift after Mesos.

The problem at that point is that you have the government (if that's what you meant by organization) competing with the private market for the talent required to do so. It's hard for the government to win that battle, and bad for the market if they do. And they wouldn't get the benefit of all the higher-level services these cloud providers have developed. It's not just about the infrastructure.

And, ignoring the money for a moment - it's hard to see how it benefits society to have so many different organizations (including businesses) essentially duplicating the effort to build the same thing. Much more efficient to let a few larger companies do it who can spread out the development and maintenance effort more. There's room for lots of smaller companies to compete in the gaps that aren't well served by the large cloud providers at that moment.

Sure. Long-running argument where the cross-over point is, with some voices on each extreme of the cross-over point, some saying "never" while others are saying "always".

Me, I think that if you want innovation, it's best to have competition, which itself is best done in a free market*.

Once you've done all the innovating, even if nobody ends up as a monopoly, then it makes sense for a government to buy up those companies and use them for the public good. Examples of low (but not necessarily zero) innovation include the water and sewage systems, and the railways (the rails themselves, not necessarily the rolling stock).

Computing is a high-innovation area, so it makes more sense for a government to buy stuff from the private sector than to do this stuff itself.

* One exception is warfare, which is high innovation, but there are game theoretic/monopoly on violence issues with allowing private military companies (as distinct from defence contractors) becoming dominant.

Anything's possible. But the vendor can increase price or refuse a discount instantly. While that government client would require a years-long large and costly project to migrate - and their track record on completing such projects is terrible. They might be able to run new projects on different vendors but it's not likely they will ever completely migrate existing applications.