A 15 Percent Premium To Rent Sovereignty. Here Is the Ownership Maths
The AWS European Sovereign Cloud went live at a consistent 15 percent premium. We do the honest sum against a one-time owned SIOS, and show where the premium and the residual actually go.
When the AWS European Sovereign Cloud went live on 15 January 2026, starting in Brandenburg and extending to Belgium, the Netherlands and Portugal, it arrived with a clean, honest number attached: roughly a 15 percent premium over the standard EU regions, run through an EU-controlled operating entity with EU-resident personnel and independent operational control.
We want to be the first to say it plainly. That is a serious piece of engineering and a genuine improvement for a lot of European institutions. A 15 percent premium to move sovereign control closer to home is, for many workloads, money well spent. This article is not an argument that the sovereign cloud is a mistake. It is the honest sum that most vendors will not put in front of you, because we sell the other side of it and we would rather you see both columns and decide for yourself.
What the 15 percent premium actually buys
The premium is not arbitrary. It pays for a separate operational perimeter, EU-resident staffing, independent control planes, and the legal restructuring that lets an EU entity make the operational calls. Those are real costs and the pricing reflects them. If your requirement is data residency, EU-controlled operations, and a strong contractual answer to "who can touch this," the sovereign cloud clears that bar for most regulated workloads.
Here is the part worth sitting with. A premium is a recurring line item. You pay it every month, on every eligible workload, for as long as you run there. Fifteen percent on a small footprint is a rounding error. Fifteen percent on a growing AI and data estate, compounded across a five-year or ten-year horizon, stops being a rounding error and becomes a structural cost of your operating model.
The residual that the premium does not remove
There is a second line that no cloud premium fully retires, and honesty requires we name it precisely rather than dramatically. Even with an EU-controlled operator, the underlying technology, code, and corporate lineage of a global hyperscaler carry a residual exposure to extraterritorial legal reach, including instruments like the US CLOUD Act. The sovereign cloud is explicitly architected to narrow that surface, and it narrows it well. We are not claiming it is defeated, and we are not claiming it makes anyone legally barred from using cloud. Almost every regime that matters, DORA, the FCA and PRA rules, the EBA guidelines, the NHS Data Security and Protection Toolkit, GDPR, permits cloud with the right controls. The genuine no-cloud bar is workload-level: classified material at SECRET and above, ITAR-controlled data, isolated OT and SCADA environments, and DPIA-negative cases. For everything else, this is a preference decision about control, cost, and data-exfiltration risk, not a prohibition.
So the honest framing is this. The premium buys you a much better answer to the residual. It does not buy you the absence of the residual. If your board's tolerance for that residual is "narrowed is fine," the sovereign cloud is a good fit. If your tolerance is "we want it inside our own walls with nobody else's lineage in the stack," that is a different requirement, and it is the one we built for.
The ownership maths, done straight
Let us set up the comparison the way a finance decision-maker would.
Renting sovereignty looks like this. Take your baseline EU cloud spend for the eligible estate. Add the 15 percent sovereignty premium. Multiply across your planning horizon. Add the residual exposure as a risk-weighted line, however your risk committee prices it. The total is a recurring, compounding cost that scales with your usage and never terminates while you run there.
Owning your intelligence layer looks different. Mickai is a Sovereign Intelligence Operating System, a SIOS, that regulated organisations own and run inside their own walls. It is built and live, air-gapped where you need it, with a cryptographically-signed audit record on every action. The commercial shape is a one-time owned deployment plus your own hardware and your own operations, rather than a per-workload premium that recurs forever. When you own the system, there is no 15 percent to pay on top, because there is no landlord. And because the stack runs inside your perimeter with no external corporate lineage in the control path, the CLOUD Act residual is not merely narrowed. There is no third party outside your walls who can be compelled, because there is no third party outside your walls who operates the system. That is a consequence of where the stack lives, not a legal opinion, and your counsel should still form their own view.
The crossover point is the whole game. On a small, static footprint, renting sovereignty is cheaper and simpler, and we will tell you so. As your AI estate grows and your horizon lengthens, the compounding premium and the risk-weighted residual cross over the one-time ownership cost. Where exactly that crossover sits depends on your usage curve, your discount rate, and how your risk committee prices the residual. We would rather hand you the model than assert the answer.
This is the same ownership logic behind our patent position: 104 UK patent applications, roughly 2,340 claims across 13 families, named inventor Mickarle Wagstaff-Irons, building toward examination and grant. We describe those filings by what they contain, sovereign control planes, signed audit, air-gapped operation, and we own the underlying architecture rather than renting it. Ownership at the IP layer and ownership at the infrastructure layer are the same idea applied twice.
Who should rent, and who should own
We will be the vendor who tells you to rent when renting is right. If your eligible footprint is modest, your growth is flat, your workloads sit comfortably inside "permitted with controls," and "narrowed residual" satisfies your board, the AWS European Sovereign Cloud is a strong, honest choice and the 15 percent is fair.
Own when the maths crosses over. If you run a large and growing AI estate, if your horizon is long, if you have workloads that genuinely cannot leave your perimeter, or if your risk committee wants the residual removed rather than reduced, then a one-time owned SIOS is the cheaper and cleaner answer over the life of the estate. The sovereign market this serves is not small: on register-backed counts, roughly 16,092 UK and EU institutions, 7,933 regulated core plus 8,159 large-private adjacency, inside an enterprise-AI-platform software TAM that Verdantix sizes from about 13 billion dollars in 2024 to 50.3 billion dollars by 2030, roughly 11.7 billion to 39.7 billion pounds. Most of those organisations are choosing on preference, not prohibition. Ownership is a preference we think a growing share of them will make once they run the sum.
The takeaway
The 15 percent premium is honest, and for the right workload it is worth paying. But it is a recurring, compounding line, and it narrows the extraterritorial residual without removing it. A SIOS you own outright converts that recurring premium into a one-time cost and collapses the residual by keeping the stack inside your walls with no external lineage in the control path. Run both columns across your real horizon. If renting wins, rent. If the crossover arrives, and for a growing estate it usually does, own.
Frequently asked questions
Is the AWS European Sovereign Cloud a bad option?
No. It is a genuine improvement for European institutions and the 15 percent premium is a fair reflection of what it costs to run a separate EU-controlled perimeter. For modest, stable footprints where "narrowed residual" satisfies the board, it is often the right call. The ownership case only wins once your estate grows and your horizon lengthens.
Does owning a SIOS mean I am legally barred from using cloud otherwise?
No, and we will not claim that. Almost every regime, DORA, FCA and PRA, EBA, NHS DSP Toolkit, GDPR, permits cloud with controls. The genuine no-cloud bar is workload-level, such as classified data, ITAR, isolated OT and SCADA, and DPIA-negative cases. Ownership is a preference decision about control, cost, and exfiltration risk, not a prohibition.
How does owning change the CLOUD Act residual rather than only narrowing it?
Because a SIOS runs inside your own walls, on your hardware, operated by your people, with no external corporate lineage in the control path. There is no third party outside your perimeter who can be compelled to act on the data, because there is no third party outside your perimeter who operates it. This is a factual consequence of where the system lives, not a legal opinion, and you should still take your own counsel's view.
Where is the crossover point between renting and owning?
It depends on your usage curve, planning horizon, discount rate, and how your risk committee prices the residual. On small, flat footprints, renting wins. As the AI estate grows and the horizon lengthens, the compounding 15 percent premium and the risk-weighted residual cross the one-time ownership cost. We would rather build the model with your numbers than assert a single figure.
If this maps to a decision you are pricing, our related writing on owning versus renting regulated AI infrastructure and on how signed audit changes the control conversation goes a level deeper into the same sum.
Micky Irons


