School teaches sine and cosine without ever showing a real circle. It teaches networks without ever opening a patch panel. It teaches databases without ever configuring a physical server. This rupture between symbol and matter creates a developer who knows how to deploy a Lambda function without knowing what runs underneath. Free Tier reinforces this amputation: it gives access to cloud abstractions without ever requiring understanding of fundamentals. The State's budgetary abandonment eliminated lab rooms with real servers. The teacher does not choose between AWS and a sovereign alternative — they choose between AWS and nothing. AWS Educate does not exist out of philanthropy. It exists because Amazon identified this void as an acquisition opportunity.

Even if a CIO decides to recruit neutral skills, recruitment platforms (LinkedIn, Indeed, internal ATS) filter CVs on hyperscaler-certified keywords: "AWS Cloud Architect", "Azure Certified Developer", "GCP Professional". A "Senior Linux Administrator" or "Bare-Metal Engineer" profile is eliminated algorithmically before the interview. Lock-in is maintained by an invisible bureaucratic barrier — not by a deliberate human decision. Documented figure: −34% Linux administrator positions between 2019 and 2023 on French job platforms.

Terms §14.12: California jurisdiction · CLOUD Act applicable · non-cancellable commits · egress fees. Sovereignty Washing (cf. The Cloud-Washing S1) sells physical resolution as if it also resolved the legal dimension. "AWS European Sovereign Cloud" = servers in Europe (physical ✓) but Amazon.com Inc. remains subject to CLOUD Act (legal ✗). The four demands of the 8 September 2026 ultimatum target exactly this layer.

Lambda + DynamoDB + API Gateway + SQS + Cognito + CloudFront — each link does not exist outside AWS in this exact form. The chain is proprietary. Migrating = total rewrite. Cloud-Washing calls this "cloud-native." Architectural question: "Can this component run outside AWS without a rewrite?" If no — artificial complexity, not modernity.

73% of organisations have SaaS purchased outside IT. Salesforce = Sales business logic · Workday = HR history · M365 = the entire organisation's documents. Repatriating servers resolves the infra layer — not the applicative layer. Two sovereignties to reconquer simultaneously, the second involving business directorates without IT constraints.

TSMC fabricates ~90% of advanced chips worldwide in the Taiwan Strait. Intel ME / AMD PSP: non-auditable sub-processors since 2008/2013, subject to US law, active with server off. Taiwan blockade: hyperscalers will ration in order — internal needs → Enterprise Agreement → Pay-as-you-go. Without owned hardware: queue. With Grade A Refurbished already fabricated (cf. The Strategic Refurbished, 4p): complete autonomy. Chips exist, independent of future TSMC production. Potential blockade impact: $600-1,200 billion (PIIE 2023).

The CIO who evangelised cloud for 5 years on LinkedIn, received a promotion for the digital transformation, convinced their CEO — cannot retreat without publicly admitting they were wrong. The fear of losing face before peers is stronger than economic rationality. The exit narrative: "The 2019 decision was correct with 2019 information. 2026 information is different — CLOUD Act clarified, egress fees documented, TSMC geopolitical risk identified. Revising the position is good management." Not communication — the truth.

Three layers neutralising the three internal resistances: Sovereignty Washing reassures legal (servers Europe ✓ / California jurisdiction ✗) · Green Washing reassures sustainability (RECs ≠ green energy · Virginia 30-35% fossil EIA 2023) · Finance Washing reassures CFO (public money → hyperscalers → anti-regulation lobbying · CPF x3). The engineer who says "let us calculate the real TCO" stands alone against three arguments, three directorates, three boxes already ticked. Cloud-Washing is not an isolated lie — it is an architecture of conviction.

McKinsey, Accenture, Capgemini are "AWS Premier Partner", "Google Cloud Partner". Their business model depends on billing complex migration person-days. The simplicity of local Bare-Metal generates no consulting revenue. A Gartner report recommending AWS is funded, directly or indirectly, by the hyperscaler ecosystem. The CIO receives an "objective" report confirming what Cloud-Washing already told them. Two external validation systems — and both have structural conflicts of interest.

DINUM selects AWS, Azure and Google for French public administrations. Public hospitals migrate to Azure. Universities use Google Workspace. Ministries deploy on AWS. The EU uses US hyperscaler services while funding GAIA-X. Paradox: public money funds hyperscalers through procurement → same hyperscalers use this money for lobbying on GDPR, Data Act, AI Act → tax funds resistance to the regulations that would protect the taxpayer.

Cloud certification spending tripled between 2020 and 2023. Public professional training money — salary contributions — massively funds AWS, GCP and Azure certifications. These certifications reinforce dependency by creating a labour market that values proprietary skills above neutral skills. The public training system funds the cognitive capture documented in The Newspeak.

Local authority "tests" on AWS Free Tier — "free to start." Six months later: paying customer of a US hyperscaler for a public service funded by local tax. No procurement process — project below public contract thresholds. The taxpayer will pay indefinitely or invest in a costly migration.

The State eliminated physical lab budgets. AWS Educate filled this void — not out of philanthropy but because Amazon identified an acquisition opportunity. The closed loop: tax → eliminates TP budget → hyperscaler fills the void → student trained on AWS → CPF funds AWS certifications → CPF is funded by tax → loop. Every euro of cloud CPF reinforces the causal chain documented in this study.