Skip to main content

The Remake Framework

AI Sovereignty Trilemma

Trust, speed, and control cannot be maximised together.

Version 1.0 Introduced Model

What it is

The proposition that organisations and jurisdictions can optimise their AI posture for trust, speed or control, but not all three simultaneously, forcing deliberate strategic positioning rather than attempting to serve every market at once.

The trilemma is a positioning model, not a maturity one: there is no ladder to climb, only a posture to choose. Its claim is that the three poles cannot be simultaneously maximised — every AI sovereignty posture is a trade of one pole against the other two, and the postures that fail are the ones chosen by default rather than by decision.

The three poles

Optimise for any two; the third is the cost. The model forces the choice to be deliberate rather than discovered.

  1. TrustAlignment with the jurisdictions and stakeholders you serve: compliance, explainability, and the confidence of regulators and customers.
  2. SpeedAccess to the frontier at the pace the market moves: the newest capability, adopted as it ships, without sovereign friction.
  3. ControlSovereignty over the stack: where models run, whose terms govern them, and what survives a provider’s or jurisdiction’s change of heart.

How to read it

Locate your organisation’s current posture honestly: which two poles has your architecture actually optimised, and which is exposed? Then read the exposure against consequence — what happens to you when the unoptimised pole is tested. The model deliberately ignores vendor promises to deliver all three; the reading is of your position as built, not as marketed.

The Insight

The hidden cost of the convenient default

The visible cost of sovereignty deters Boards — the infrastructure, the duplication, the slower frontier access. What the trilemma makes visible is the cost that was never shown: the convenient, frontier-speed default is itself a position on the triangle, with control as its exposed pole. The organisations that felt that exposure did not choose it; they discovered it.

A worked reading

On 12 June, a frontier model was withdrawn from service and organisations built on it discovered their position on the triangle in a single morning: speed and trust optimised, control absent — dependencies with no sovereign fallback. A trilemma reading beforehand would not have prevented the withdrawal; it would have made the exposure a Board decision with a mitigation, instead of a surprise with a bill.

The Framework

In the framework

Within Remake, the AI Sovereignty Trilemma is declared a Model — read at Remake: Advise, where posture becomes guiding policy: which two poles this organisation optimises, what the exposed pole costs, and what mitigates it. A deliberate position, on the record.

Cited in

Writing that builds on this asset.

History

Provenance and revision history. Every asset in the library carries this record, so what you are applying, and which version of it, is never in doubt.

Asset
AI Sovereignty Trilemma
Kind
Model — a way of seeing or structuring something
Status
Current
Version
1.0
Origin
Developed by Mario Thomas; introduced 7 September 2025 in The AI Sovereignty Trilemma
Registry
Recorded in the Remake Library and defined in the glossary

Versions

  1. 1.0 Introduced in The AI Sovereignty Trilemma.

I use cookies to understand how my website is used. This data is collected and processed directly by me, not shared with any third parties, and helps me improve my website. See my privacy and cookie policies for more details.