TL;DR
Thorsten Meyer AI has challenged its own five-week campaign for AI sovereignty, arguing that most organizations should use the strongest available model behind a multi-provider router. The publication says dedicated sovereign infrastructure remains justified for workloads restricted by law, classification rules or binding regulation.
Thorsten Meyer AI has challenged its recent advocacy of owning AI infrastructure, concluding that most organizations should prioritize the best-performing model and use a multi-provider router for resilience. The July 16 analysis retains a narrower case for sovereign systems where law, classified work or regulated data prevents use of foreign-controlled services.
The publication said its previous eight analyses had repeatedly favored model ownership, local computing capacity and scrutiny of provider control. Its new article argues that this pattern risked turning reporting into a predetermined thesis. After reconsidering the same evidence, it found that capability gaps, qualification costs and delayed product delivery weaken the sovereignty case for buyers without binding restrictions.
The analysis cited tests in which Inkling reportedly scored 77.6% on SWE-bench, compared with 95.0% for Fable 5, and 63.8% against 89.5% on Terminal-Bench. It described the figures as drawn from Artificial Analysis and vendor tables, while acknowledging that some results are self-reported and awaiting replication.
Thorsten Meyer AI also cited estimates including a $75,000 to $100,000 annual staffing cost, an idle-capacity penalty approaching tenfold and major European infrastructure commitments. These numbers came from the publication’s earlier reporting and named outside sources, but the supplied material does not include the underlying reports or methods needed to verify each comparison independently.
Against sovereignty: the strongest case for just using the best model
This publication has spent five weeks arguing one thing — and every piece converged. That should bother you. It bothers me. When eight analyses reach the same verdict, you’re not running an analysis. You’re running a thesis, and the evidence has started arriving pre-sorted.
So here’s the case against — argued properly, with the same evidence, turned around. Not a strawman erected to be knocked down. The version a smart CTO would put to me across a table, and which I have not yet answered in public. The claim: for almost everyone, sovereignty is an expensive hedge against a risk they’ve mispriced — and the rational move is to use the best model and get on with it.
Defence · classified · national health data · DORA-bound finance. The foreign-legal-order risk isn’t theoretical and isn’t insurable by other means — it’s a legal gate. No benchmark opens it. Your alternative isn’t a worse model; it’s no deployment at all.
Statistically, you are. You have a reasonable, politically legible, entirely unbudgeted feeling — and an industry built to monetize it. The capability compounds, the tax is real, the opportunity cost is brutal, and 18 days is survivable.
I’ve spent five weeks arguing you should own your stack. The strongest case against says: for most of you, that’s an expensive way to be worse, sold by people whose real product is a feeling. And that case is mostly right. What survives is smaller and sharper — everything above the router line (the qualification programme, the owned cluster, the custom pre-training run, the €11B data centre) you should buy only if a law requires it, never because a narrative does. A router is the sovereignty most people actually need. 90% of the resilience for ~2% of the cost — and it would have made 12 June a non-event. So run the honest test: are you bound, or are you performing?
Performance Gains Change the Buying Test
The argument matters because AI infrastructure decisions can commit organizations to years of spending, specialized staff and slower deployment. If leading hosted models complete more coding or agent tasks successfully, choosing a weaker system for symbolic autonomy may impose daily operational costs that exceed the expected benefit.
The publication draws a firm line between mandatory sovereignty and preference. Defence, classified operations, some national health-data systems and finance subject to binding regulatory limits may have no lawful alternative. Other buyers, it argues, can obtain much of the required resilience through provider switching and business-continuity planning.
AI model performance testing tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Five Weeks of Sovereignty Advocacy
The new position follows five weeks in which Thorsten Meyer AI examined European model providers, ownership structures, computing capacity and the possibility that a foreign government or supplier could withdraw access. Those articles repeatedly supported owning the model rather than relying on an API.
The reassessment focuses on a disruption the publication dates from June 12 to July 1. It says a Commerce directive removed access to Fable 5 and Mythos 5 for 18 days before service returned, with alternative providers available. On that account, the episode was a survivable vendor interruption, although the supplied source does not provide the directive or independent documentation of the incident.
multi-provider AI routing software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Benchmark and Cost Claims Need Testing
It is not yet clear whether the cited benchmark gaps will hold under independent testing or across real production workloads. Benchmark scores may not capture privacy, latency, language coverage, customization or sector-specific accuracy, and the publication acknowledges that some figures await replication.
The source also does not establish that a router delivers the claimed 90% of resilience for about 2% of the cost across different organizations. Migration complexity, incompatible provider features and data-transfer restrictions could reduce that benefit. The exact boundaries of regulated or legally restricted workloads also depend on jurisdiction, contracts and deployment design.
enterprise AI infrastructure solutions
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Procurement Teams Face a Binding Test
The analysis calls on procurement and technology leaders to determine whether a specific legal rule requires sovereign deployment before funding dedicated clusters, qualification programs or custom model training. Organizations without such a requirement would instead compare model capability, switching options and outage tolerance. Independent benchmark replication and fuller documentation of the cited costs will be needed to test the recommendation.
AI model benchmarking devices
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Is Thorsten Meyer AI abandoning AI sovereignty?
No. It is narrowing the case to legally or operationally restricted workloads while opposing broad adoption based mainly on perceived geopolitical risk.
Who does the analysis say should buy sovereign systems?
It identifies defence, classified operations, regulated finance and some national health-data workloads where foreign control may create a binding legal barrier.
What does the publication recommend for most companies?
It recommends selecting the best-performing suitable model, placing a multi-provider router in front of services and maintaining tested fallback arrangements.
Are the benchmark and cost figures confirmed?
The publication attributes them to earlier reporting, vendor tables and named research sources. Some results are self-reported, and the supplied material does not permit independent verification of every figure.
Source: Thorsten Meyer AI
