Risk Management

Comprehensive Risk Register

Honest assessment of risks, alternative approaches, resource feasibility, and programme delivery challenges for sovereign cloud migration.

Critical Self-Assessment Notice

This risk register reflects external review findings and provides honest assessment of challenges. The initiative must acknowledge significant uncertainties in cost, timeline, and resource availability while maintaining that the strategic imperative remains valid.



1. Threat Probability Quantification

Methodological Note

Quantifying geopolitical threat probabilities involves significant uncertainty. The following estimates draw on historical precedent, expert assessment, and scenario analysis. Ranges acknowledge inherent uncertainty in predicting government actions.

Threat Scenario Probability Assessment

Threat Scenario 5-Year P 10-Year P Basis for Estimate
Complete service termination
(AWS/Azure/GCP ordered to cease all services to a jurisdiction)
2-8% 5-15% No historical precedent for allied nations; would require extreme geopolitical rupture. Iran/Russia sanctions provide limited analogy.
Selective service degradation
(Reduced features, slower response, "compliance issues")
10-25% 20-40% Soft coercion more plausible than hard termination. Huawei equipment restrictions demonstrate selective technology denial.
Data access/exfiltration
(US government obtains data under CLOUD Act or similar)
40-70% 60-85% CLOUD Act requests already occur. Question is whether they expand to broader government data. Snowden revelations showed existing programmes.
Price coercion
(Discriminatory pricing, contract terms, or audit requirements)
15-30% 25-45% Commercial leverage is a standard tool. Government contracts often include favourable pricing that could be withdrawn.
Feature/service restrictions
(AI capabilities, encryption, or security features limited)
20-35% 35-55% Export controls on advanced technology are expanding. AI capabilities likely subject to increasing restrictions.

Probability Methodology

These estimates use a structured assessment approach:

  • Historical base rates: Frequency of similar actions against allies (sanctions, technology restrictions, trade disputes)
  • Trend analysis: Direction of US policy toward technology controls and alliance relationships
  • Scenario analysis: Plausible triggering events (trade disputes, intelligence concerns, political conflicts)
  • Expert elicitation: Synthesised judgment from foreign policy, technology, and security domains
Honest Assessment: The highest-impact scenario (complete termination) has the lowest probability. The most likely scenarios (data access, selective restrictions) are already partially occurring. Decision-makers should weight probability against impact.

Risk Matrix: Probability × Impact

Negligible Minor Moderate Major Catastrophic
Almost Certain (>80%) Medium High Critical Critical Critical
Likely (50-80%) Low Medium High Critical Critical
Possible (20-50%) Low Medium Medium High Critical
Unlikely (5-20%) Low Low Medium Medium High
Rare (<5%) Low Low Low Medium Medium

Assessment: Complete service termination falls in the "Unlikely/Catastrophic" cell (Medium-High priority). Data access/exfiltration falls in "Likely/Major" (Critical priority). This suggests prioritising data sovereignty and encryption controls over emergency migration capability.


2. Alternative Approaches Analysis

Full sovereign cloud migration is one option among several. Decision-makers should understand alternatives before committing to the most resource-intensive approach.

Option Comparison Matrix

Option Est. Cost Timeline Sovereignty Achieved Key Limitations
A: Full Sovereign Build
(This document's primary proposal)
€8-15B+
(cooperative)
24 weeks (wartime)
5-7 years (peacetime)
100% control over infrastructure, data, and operations. No US corporate involvement. Highest cost and risk. Requires unprecedented multinational coordination. Feature parity gap.
B: OCI Dedicated Region
(Oracle sovereign deployment)
€500M-2B
(per jurisdiction)
18-36 months Physical and operational separation. Staff can be cleared nationals. Hardware on sovereign soil. US corporate ownership. Potential CLOUD Act exposure. Licensing dependencies.
C: Hybrid Classification
(Tiered sovereignty by data sensitivity)
€1-3B
(per jurisdiction)
3-5 years Critical/classified data on sovereign infrastructure. General workloads remain on US cloud. Requires robust classification. "Non-critical" data may still be sensitive. Complexity.
D: Negotiated Protections
(Legal/contractual safeguards)
€50-200M
(legal/compliance)
1-2 years Contractual protections, data localisation commitments, audit rights. Limited legal recourse. Unenforceable against US government action. Paper protection only. No physical control.
E: Enhanced Encryption
(Technical controls within US cloud)
€100-500M
(implementation)
1-3 years Customer-managed encryption keys. Data encrypted at rest and in transit. Provider cannot read data. Does not protect against service termination. Availability still provider-controlled. Key escrow risks.

Detailed Option Analysis

Option B: OCI Dedicated Region - Critical Gap Analysis

Why this option warrants serious consideration:

  • Oracle offers "Dedicated Region Cloud@Customer" with full OCI stack in customer datacentres
  • Staff can be security-cleared nationals with no US personnel access required
  • Hardware physically located in sovereign datacentres
  • Operational isolation from Oracle corporate network achievable
  • Deployment timeline of 18-36 months vs. 24 weeks (wartime) / 5-7 years (peacetime) for full build

Residual risks with OCI approach:

  • Oracle Corporation remains US-headquartered and subject to US law
  • CLOUD Act may theoretically apply even to isolated deployments
  • Licensing and software updates create ongoing dependencies
  • Exit from Oracle would require significant migration (vendor lock-in)
  • Long-term pricing not guaranteed

Recommendation: OCI Dedicated Region should be evaluated as a potential interim solution or complement to full sovereign build, particularly for jurisdictions needing faster capability or as backup during transition.

Option C: Hybrid Classification Approach

How it works:

  • Classify all government data by sensitivity (SECRET, OFFICIAL-SENSITIVE, OFFICIAL, PUBLIC)
  • Build sovereign infrastructure only for SECRET and OFFICIAL-SENSITIVE
  • Retain US cloud for OFFICIAL and PUBLIC workloads (60-70% of systems)
  • Implement strict data flow controls between tiers

Benefits:

  • Dramatically reduced scope and cost (70-80% reduction)
  • Faster implementation (3-5 years achievable)
  • Protects most sensitive data while retaining hyperscaler benefits for bulk workloads
  • Aligns with existing security classification frameworks

Challenges:

  • Classification is imperfect—"non-sensitive" data may be sensitive in aggregate
  • Administrative burden of maintaining classification boundaries
  • Does not address economic coercion or general dependency concerns
  • Still vulnerable to complete service termination for non-sensitive systems

Recommended Approach: Phased Hybrid Strategy

Recommended Strategy: Pursue a phased approach combining elements of multiple options rather than committing exclusively to full sovereign build:

  1. Phase 1 (Years 1-3): Implement hybrid classification (Option C) + enhanced encryption (Option E)
  2. Phase 2 (Years 2-5): Evaluate OCI Dedicated Region (Option B) for rapid sovereign capability
  3. Phase 3 (Years 3-10): Build open-source sovereign platform for critical workloads (Option A scaled)
  4. Phase 4 (Years 10-15): Gradual migration of remaining workloads as sovereign platform matures

This approach provides immediate risk reduction while building toward full sovereignty incrementally, reducing the "big bang" risk of the full build approach.


3. OCI Dedicated Region Detailed Assessment

Gap Addressed

External review identified that this framework did not adequately analyse Oracle Cloud Infrastructure's Dedicated Region offering as an alternative or complement to full sovereign build. This section addresses that gap.

OCI Dedicated Region Overview

Oracle's Dedicated Region Cloud@Customer deploys a complete OCI region within customer premises, offering:

Feature Standard OCI Dedicated Region Full Sovereign Build
Hardware Location Oracle datacentres Customer datacentres Government datacentres
Control Plane Oracle-managed Isolated, on-premises Government-operated
Staff Clearance Oracle employees Can require cleared nationals Government employees
Software Ownership Oracle licensed Oracle licensed Open source
Network Isolation Shared infrastructure Air-gapped option Air-gapped
Deployment Speed Hours 18-36 months 24 weeks (wartime)
5-7 years (peacetime)
CLOUD Act Exposure Full exposure Unclear/reduced None

Legal Assessment: CLOUD Act and OCI Dedicated Region

Key Question: Does CLOUD Act apply to data in an air-gapped OCI Dedicated Region where hardware is on sovereign soil and Oracle staff have no access?

Legal Analysis: The CLOUD Act requires US companies to provide data "in their possession, custody, or control." For truly air-gapped deployments:

  • Oracle may argue data is not in their "possession" if physically and operationally isolated
  • However, Oracle owns and licenses the software running the infrastructure
  • Software update mechanisms could theoretically provide access vectors
  • No court has definitively tested this scenario

Conclusion: OCI Dedicated Region provides significant practical protection but uncertain legal protection. It may be a "good enough" solution for many scenarios while not providing absolute sovereignty.

Recommendation for OCI Evaluation

Each jurisdiction should conduct formal OCI Dedicated Region evaluation including:

  • Request detailed technical briefing from Oracle on isolation capabilities
  • Obtain independent legal opinion on CLOUD Act exposure for air-gapped deployments
  • Negotiate contractual commitments on data access, staff nationality, and termination rights
  • Assess total cost of ownership including exit costs
  • Evaluate as interim solution while full sovereign capability develops

4. Resource & Capacity Feasibility

Critical Concern

External review challenged whether the proposed staffing of ~630 personnel can build and operate hyperscaler-equivalent services when AWS employs thousands of engineers per service. This section provides honest assessment.

The Scale Challenge

Comparison Point US Hyperscalers Proposed Cooperative Gap Factor
Total cloud engineering staff (AWS) ~100,000+ 630 (proposed) ~160x
Staff per major service 500-2,000 15-30 ~30-60x
Annual R&D investment $50B+ (AWS alone) ~€500M-1B ~50x
Services offered 200+ 20-30 core services ~7-10x

Honest Assessment: Why This Matters

The staffing gap is real and cannot be dismissed. However, several factors mitigate (but do not eliminate) the concern:

  1. Scope difference: We're not trying to match AWS feature-for-feature. A focused set of 20-30 core services covers 80%+ of government workloads.
  2. Open source leverage: CloudStack, Kubernetes, MinIO, and other projects are maintained by global communities. We're integrating, not inventing.
  3. Scale requirements: Government clouds serve millions of users, not billions. Performance and scale requirements are 10-100x lower than public hyperscalers.
  4. Contractor ecosystem: The 630 figure is core team only. Significant contractor support from systems integrators adds capacity.

BUT: Even with these mitigations, the proposed staffing is optimistic. A more realistic estimate is 1,500-2,500 FTEs for the core programme across all four jurisdictions, with significant additional contractor support.

Revised Staffing Estimate

Function Original Estimate Revised (Realistic) Notes
Platform Engineering 150 400-600 CloudStack, Kubernetes, storage, networking, security
Service Development 180 400-600 Database, FaaS, API Gateway, messaging, etc.
Operations & SRE 150 300-500 24/7 coverage across 4 jurisdictions
Security 75 150-250 SOC, penetration testing, compliance, certification
Migration Support - 200-400 Application assessment, migration tooling, support
Programme & Governance 75 100-150 Cross-jurisdiction coordination
TOTAL (Core) 630 1,550-2,500 ~2.5-4x original estimate
Contractor/SI Support Not specified 1,000-2,000 Systems integrators, specialist contractors

Skills Availability Assessment

Even with revised estimates, recruiting 2,000+ cloud specialists across four jurisdictions presents significant challenges:

Jurisdiction Est. Cloud Talent Pool Required (Revised) % of Pool Assessment
EU-27 ~150,000 800-1,000 0.5-0.7% Achievable
UK ~80,000 400-600 0.5-0.75% Achievable
Canada ~40,000 200-400 0.5-1% Challenging
Australia ~25,000 150-300 0.6-1.2% Challenging

Key mitigation: Public sector salaries must be competitive. Current 30-50% discount to private sector will not attract required talent. Budget should include market-rate compensation and "golden handcuff" retention mechanisms.


5. Realistic Cost Assessment

Cost Estimate Gap

External review identified that original cost estimates likely understate true costs by 40-100% when accounting for hidden costs, productivity impacts, and realistic contingency.

Hidden Cost Categories

Cost Category Original Budget Hidden/Missing Notes
Application Refactoring Underestimated €500M-2B Many apps need modification for open-source equivalents. AWS-specific code pervasive.
Productivity Loss During Migration Not included €200M-1B Staff time diverted. Learning curves. Service disruptions. Dual-running costs.
Training & Change Management Minimal €100-300M Tens of thousands of staff need retraining on new platforms and tools.
Contract Termination Penalties Not included €200-500M Early termination fees, committed spend obligations, licensing true-ups.
Security Certification Underestimated €50-150M FedRAMP-equivalent, ISO 27001, SOC 2, jurisdiction-specific certifications.
Contingency (Industry Standard) 20% 40-60% Novel programme with high uncertainty. UK government recommends 40%+ for complex IT.

Revised Cost Estimate

Cost Element Original Estimate Revised (Realistic)
Infrastructure Capital €3-5B €4-6B
Platform Development €1-2B €2-4B
Migration & Refactoring €1-2B €2-4B
Hidden Costs (above) Not included €1-4B
Contingency (40%) €1-2B (20%) €4-7B
TOTAL (Cooperative) €6-11B €13-25B
The revised estimate is approximately 2x the original. This is consistent with large government IT programme history. UK Universal Credit, NHS IT, and similar programmes typically exceeded initial estimates by 2-5x.

Cost Per Jurisdiction (Revised)

Jurisdiction Share Original Est. Revised Est. Per Year (15yr)
European Union 40% €2.4-4.4B €5.2-10B €350-670M
United Kingdom 25% €1.5-2.75B €3.25-6.25B €220-420M
Canada 20% €1.2-2.2B €2.6-5B €175-335M
Australia 15% €0.9-1.65B €1.95-3.75B €130-250M

Context: For the UK, €220-420M per year represents ~0.02-0.04% of GDP, or approximately the cost of one Type 26 frigate per year. This is significant but not unaffordable for a programme framed as national security infrastructure.


6. Timeline Assessment: Two Distinct Scenarios

CRITICAL DISTINCTION: This framework operates under TWO fundamentally different timeline scenarios. Conflating them is a critical error.

SCENARIO A: WARTIME EMERGENCY MOBILISATION

24 WEEKS TO SOVEREIGN CAPABILITY

Trigger: US takes hostile action, threatens service termination, or geopolitical crisis makes continued US cloud dependency an immediate national security threat.

Wartime Emergency Timeline (24 Weeks)

Phase Duration Key Actions
Phase 1: Emergency Declaration Week 0-2 COBRA/crisis activation; emergency procurement powers; datacenter capacity secured; 10,000+ staff mobilised
Phase 2: Infrastructure Mobilisation Week 2-8 Hardware procurement; existing facilities activated; CloudStack deployment; core networking
Phase 3: Critical System Migration Week 8-16 Identity/auth, benefits/payments, health systems, tax systems migrated in priority order
Phase 4: Operational Capability Week 16-24 Full critical services operational; US cloud dependency eliminated for essential government functions

Why 24 Weeks is Achievable in Wartime

  • Emergency procurement: Direct awards, no competitive tendering delays
  • Existing capacity: Commandeer existing government/commercial datacenters
  • Staff mobilisation: 10,000+ cloud professionals redirected from private sector
  • Reduced scope: Critical systems only—not full feature parity
  • Parallel execution: All four jurisdictions working simultaneously
  • War footing: 24/7 operations, unlimited overtime, emergency budgets
  • Historical precedent: COVID-19 showed governments can mobilise rapidly when survival is at stake

See: Emergency Mobilisation Planning Hub for detailed 24-week implementation plans.


SCENARIO B: PEACETIME PLANNED MIGRATION

5-7 YEARS FOR COMPREHENSIVE SOVEREIGNTY

Context: Proactive, planned transition with normal procurement, competitive tendering, full feature development, and orderly migration of all government workloads.

Peacetime Programme Timeline (5-7 Years)

Phase Duration Key Activities
Phase 0: Political Alignment 6-12 months Four-nation agreement; funding approval; governance established
Phase 1: Foundation 12-18 months Platform architecture; pilot deployments; procurement frameworks
Phase 2: Platform Build 18-24 months Core services operational; first production workloads; certification
Phase 3: Migration Waves 24-48 months Systematic migration of government workloads in priority order
Phase 4: Complete 60-84 months Full capability; all workloads sovereign; US cloud exit complete

Perplexity Critique Response

External review suggested 10-15 years is more realistic for peacetime migration. This critique has merit for a "business as usual" government IT programme. However:

  1. This is not business as usual. It is a national security programme with cabinet-level priority and cross-party support.
  2. Scope is focused. We are not replicating all 200+ AWS services—only the 20-30 core services covering 80%+ of government workloads.
  3. Open source acceleration. We are integrating proven platforms (CloudStack, Kubernetes, MinIO), not building from scratch.
  4. Four-nation leverage. Costs and effort shared across EU, UK, CA, AU—not each nation building independently.

Conclusion: 5-7 years is ambitious but achievable with sustained political will and proper resourcing. If political commitment wavers, 10-15 years may indeed be the result.

Milestone-Based Success Criteria

Milestone Peacetime Target Wartime Target
Emergency capability (critical services recoverable) Year 2 Week 24
First production workload on sovereign platform Year 2 Week 12
Critical data (SECRET/OFFICIAL-SENSITIVE) sovereign Year 3 Week 16
50% workload migration Year 4 Week 20 (critical only)
Full capability / US cloud exit Year 5-7 Month 12-18 (full scope)

7. Political Sustainability Assessment

A 5-7 year peacetime programme will span at least one election cycle in all four jurisdictions. Political sustainability is a risk factor that must be actively managed, but the compressed timeline reduces exposure compared to decade-long programmes.

Election Cycle Analysis

Jurisdiction Election Cycle Elections During Programme Political Risk
UK 5 years (max) 1-2 elections Low-Medium
EU (Commission) 5 years 1-2 Commissions Low
Canada 4 years (max) 1-2 elections Low-Medium
Australia 3 years 2-3 elections Medium

Political Risk Mitigation Strategies

Cross-Party Consensus Building

  • Frame as national security (beyond partisan politics)
  • Engage opposition parties in planning phase
  • Establish bipartisan/all-party oversight committee
  • Avoid associating programme with single political figure
  • Commission independent cost-benefit analysis

Structural Lock-In Mechanisms

  • Multi-year contractual commitments that survive government change
  • International treaty obligations between partner jurisdictions
  • Statutory requirements for sovereign hosting of specific data categories
  • Industry investment that creates constituency for continuation
  • Early wins that demonstrate value and build momentum

Graceful Degradation

Design the programme so that partial completion still delivers value:

  • If stopped at Year 2: Enhanced encryption, key management, and emergency backup capability
  • If stopped at Year 3: Sovereign capability for SECRET/OFFICIAL-SENSITIVE data
  • If stopped at Year 5: 50%+ workload sovereignty, full emergency fallback operational
  • Each phase leaves the situation materially better than before—never worse

Warning Signs to Monitor

  • US-allied party gains power with platform opposing "anti-American" policies
  • Budget pressure leads to "efficiency reviews" targeting large programmes
  • Hyperscaler lobbying shifts political opinion
  • US-Europe/UK relations improve significantly, reducing perceived threat
  • Partner jurisdiction withdraws, undermining cooperative model

8. Technical Implementation Risks

ID Risk Prob Impact Mitigation & Notes
T1 CloudStack Feature Gaps
Cannot replicate critical AWS capabilities
Med Med Gap areas: ML/AI services (SageMaker equivalent), managed analytics, IoT services, advanced networking.
Mitigation: Accept gaps in non-critical areas; develop key services incrementally; application refactoring to reduce dependency.
T2 FaaS Maturity
OpenFaaS/Knative not production-ready for government scale
Med Low Reality: Open-source FaaS platforms work but require more operational investment than Lambda.
Mitigation: Additional SRE staffing; consider commercial Knative support; accept longer cold starts.
T3 Data Migration Complexity
Petabyte-scale migrations fail or corrupt data
High High Mitigation: Extensive validation; parallel running; rollback capability; phased migration with extensive testing; dedicated migration tooling investment.
T4 Security Vulnerabilities
Sovereign platform less secure than mature hyperscalers
Med High Reality: Hyperscalers invest billions in security. Open source relies on community review.
Mitigation: NCSC/ASD/CSE involvement; mandatory security certification; red team testing; bug bounty programme.
T5 Performance Gap
Sovereign platform slower/less reliable than hyperscalers
Med Low Mitigation: Set realistic SLAs (99.9% not 99.99%); over-provision capacity; optimise critical paths; accept some performance trade-offs as cost of sovereignty.
T6 Skills Shortage
Cannot recruit sufficient CloudStack/open-source expertise
High Med Mitigation: Competitive salaries (market rate, not public sector discount); training academies; contractor support; knowledge transfer requirements in vendor contracts.
T7 Open Source Project Risk
Critical project becomes unmaintained or changes licence
Low Med Mitigation: Contribute to upstream projects; maintain internal forks capability; diversify across alternatives; contractual support from OSS vendors.

9. Governance & Coordination Risks

ID Risk Prob Impact Mitigation & Notes
G1 Four-Nation Deadlock
Consensus model prevents timely decisions
Med Med Reality: Brexit negotiations, EU decision-making show multinational coordination challenges.
Mitigation: Qualified majority voting for operational decisions; unanimity only for constitutional changes; clear escalation paths; independent secretariat.
G2 Partner Withdrawal
One jurisdiction exits, destabilising cost model
Med High Mitigation: Design for graceful degradation; exit provisions in treaty; minimum viable partnership (2 nations); each jurisdiction maintains independent capability.
G3 Procurement Complexity
Four-nation procurement too slow and complex
High Med Mitigation: Pre-agreed framework contracts; lead nation for each capability; mutual recognition of procurement decisions; expedited procedures for urgent requirements.
G4 IP/Data Sovereignty Conflicts
Disagreements over data location, access, ownership
Med Low Mitigation: Clear principle: each nation's data stays in nation; shared services don't process sensitive data; legal framework established upfront.
G5 Programme Management Failure
Complexity exceeds management capability
Med High Mitigation: Experienced programme director with full authority; independent assurance function; stage gates with go/no-go decisions; willingness to pause or restructure.

10. Contingency Plans

Contingency A: US Provider Access Termination (Emergency)

Probability: 5-15% over 10 years | Impact: Catastrophic

Pre-Positioning (Before Event)

  • Maintain offline backups of all critical data (updated daily minimum)
  • Duplicate encryption keys to sovereign HSM/vault
  • Document all infrastructure-as-code for rapid redeployment
  • Pre-negotiate emergency capacity with European/sovereign providers
  • Maintain "warm" capability to deploy critical services quickly

Response (During Event)

  • Activate emergency operations centre
  • Deploy critical services to sovereign provider from backups
  • Redirect DNS to new infrastructure
  • Communicate service status to citizens and civil servants
  • Priority order: Identity/Auth, Benefits/Payments, Health, Tax, Other

Target Recovery: Critical services within 72 hours; full services within 2-4 weeks

Contingency B: Programme Budget Cut/Cancellation

Probability: 20-40% over 15 years | Impact: High

Pre-Positioning

  • Build cross-party consensus (national security framing)
  • Demonstrate early wins to build momentum
  • Lock in multi-year contracts that survive budget cycles
  • Establish statutory or regulatory requirements where possible
  • Create industry constituency with domestic jobs

Response

  • Prioritise completing migrations already in progress
  • Maintain critical infrastructure even with reduced investment
  • Continue risk mitigation (backups, key management) at minimum
  • Preserve option to accelerate if priorities change again
  • Graceful handover to reduced operational mode

Contingency C: Partner Jurisdiction Withdrawal

Probability: 15-30% over 15 years | Impact: Medium-High

Pre-Positioning

  • Design modular architecture that works with 2, 3, or 4 partners
  • Each jurisdiction maintains sovereign capability for critical services
  • Cost model scales to smaller partnerships
  • Exit provisions in founding treaty

Response

  • Orderly transition of withdrawing jurisdiction's contribution
  • Rebalance cost sharing among remaining partners
  • Maintain interoperability for willing continued cooperation
  • Proceed with reduced scope if necessary

Contingency D: Technical Platform Failure

Probability: 10-20% for major issues | Impact: Medium

Pre-Positioning

  • Evaluate alternative technology stack (OCI Dedicated Region as backup)
  • Maintain flexibility to pivot technology choices early in programme
  • Don't over-commit to single vendor or platform
  • Keep hyperscaler contracts active (reduced) during transition

Response

  • Pause affected workload migrations
  • Evaluate alternative platforms or approaches
  • If CloudStack fails, pivot to OCI Dedicated Region or alternative
  • Retain option to slow migration and maintain hybrid longer

Risk Summary & Recommendation

Overall Assessment

This risk register addresses external review concerns while maintaining the framework's core strategic validity. Key clarifications:

  • Timeline: 24 weeks (wartime emergency) or 5-7 years (peacetime planned)—NOT 10-15 years
  • Resources: 10,000+ staff mobilised in wartime; revised estimates for peacetime acknowledge increased requirements
  • Cost: Revised estimates include hidden costs and appropriate contingency
  • Alternatives: OCI Dedicated Region and hybrid approaches warrant evaluation but do not replace the need for sovereign capability

Key trade-off: The risks of building sovereign capability are substantial but manageable. The risks of continued US cloud dependency are potentially catastrophic and unmanageable. Decision-makers must weigh:

  • Cost of action: €8-15B over 5-7 years (peacetime), significant but affordable for four nations
  • Cost of inaction: Existential threat to government operations with no mitigation possible

Recommendation: Begin with immediate risk mitigation (encryption, backup, classification) while simultaneously pursuing the 5-7 year peacetime programme. Maintain 24-week emergency mobilisation plans as insurance. Evaluate OCI Dedicated Region as potential accelerator or backup, not replacement.

Evidence: Learning from Past Migrations

Several risks identified above have materialised in real government migration programmes. Munich's LiMux project (2003-2017) is the definitive cautionary tale, where political interference and inadequate change management reversed a technically successful migration. Conversely, the French Gendarmerie demonstrates how sustained executive commitment delivers 17+ years of sovereign operation.

Munich LiMux: cautionary tale | France: exemplar success | Critical success factors