What we argue,
and why.
Long-form arguments from the people who build and run Oasis Stack, for operators, governments and enterprises deciding what kind of cloud to own. Each one takes a single decision and argues it from the ground up.
The long version.
Each article takes one decision — who owns the cloud, what it costs, where the boundary goes — and argues it from the ground up.
Anyone who wants to sell cloud is offered two roads: resell someone else’s, or spend years building your own. This is the third.
For a decade, cloud providers were measured by size. That was a race only a handful could run. The next generation is measured by what it lets others build.
The cloud was offered as speed. The price was the estate. Now the board wants to know who controls it, how you prove it, and how you would ever leave.
Independent providers were offered two paths: become a hyperscaler, or resell one. Federation is the third, and it changes what being small means.
Buying accelerators does not make a cloud AI-ready. What makes it ready is everything that turns them into a service people can order.
Almost every AI offer comes with a condition: send your most valuable data somewhere else first. That condition is the real decision.
Reducing dependency is usually told as a replacement story. For the provider that runs a national cloud, it is a harder job: being chosen.
Building your own cloud looks like an infrastructure project. It is a software business you did not mean to start, and the clock runs for years.
A cloud can sit entirely inside your borders and still answer to someone else’s roadmap. Sovereignty has a technology layer, and it is the one most often skipped.
The choice is usually presented as cost against control. It should be a decision about where to draw the boundary — workload by workload.
You already manage your customers’ clouds. The question is why the cloud revenue still goes to someone else.
Roads, ports and power grids built the last economy. The next one runs on infrastructure too — the question is who owns it.
Cloud was supposed to make infrastructure spend simple. For a growing number of organisations it became the hardest line in the budget to predict.
The region is not choosing between modern cloud and control any more. It is asking why it was ever told to.
Space, power and ports built the business. Customers now buy something else, and whoever sells it owns the relationship.
Six things we think this industry has wrong.
The articles argue from these. Each is a line published elsewhere on this site, linked to the page that makes it in full.
A local region on a foreign platform answers one layer of the question. Identity, operations, key custody, the roadmap and the right to continue all stay abroad. We argue sovereignty is seven decisions, made deliberately, not one box ticked in a procurement form.
Per-core licensing grows with your hardware. Consumption billing grows with your customers. Revenue share grows with your success. All three are the same arrangement: the better your cloud does, the more of it belongs to someone else.
A certificate attaches to an organisation, a scope and a set of operating practices — not to software. Vendors who imply otherwise are selling a comfort that will not survive an assessor. What software can do is make controls enforceable and evidence automatic.
Every cloud built outside the hyperscalers hits the same ceiling: alone, you can only offer what you can build. We argue that ceiling was a consequence of clouds being islands, not a law of the industry.
That gap is the real reason operators accept the trade. Running a cloud properly looks like it needs a headcount only a global provider can carry — which is an automation problem, not a destiny.
Move to someone else’s cloud, and modernise on the way. The estate you spent years building becomes a line in a decommissioning plan, and the dependency you were reducing doubles.
Every one of them is the same argument.
Read together, the six positions are one claim wearing different clothes: the cloud industry settled on a trade — you can have a cloud that works, or a cloud that’s yours — and then sold that trade back to everyone as though it were a law of physics. It is not. It is a consequence of clouds being built alone.
Where the argument startsDisagree with one of them?
These are positions, not marketing. If one of them is wrong about your market, that is the conversation worth having.