US tech giants warn Europe that ditching US tech giants could cost €115B

A CCIA Europe-commissioned study estimates that the EU’s proposed Cloud and AI Development Act could impose €115 billion in costs over 10 years, versus the European Commission’s upper estimate of €18 billion. The estimate includes €33.3 billion in direct public-sector costs and a modeled €80 billion for private companies, assuming sovereignty requirements extend to 25% of cloud workloads in essential industries; the report says software redevelopment represents 69% of costs. The findings come from an industry group representing major US cloud companies that could benefit if Europe maintains purchases of their services, and the actual cost remains uncertain.

Analysis

The key economic distinction is between sovereignty rules that require technical controls and rules that effectively discriminate by provider ownership or staffing. The former could let AWS and Google retain workloads through localized offerings, potentially adding compliance and infrastructure spend rather than causing wholesale displacement. The latter raises the risk of durable share transfer to European providers such as OVHcloud and Deutsche Telekom, with migration costs acting as a barrier to switching in both directions. Meta has much less direct cloud exposure; its relevance is primarily through the broader AI-infrastructure and regulatory precedent.

The €115bn estimate is a lobbying-group-commissioned scenario, not an established liability. Its private-sector component depends on a specific expansion of Article 31, so treating the headline as a quantified revenue threat to hyperscalers overstates what is known. Near term, expect policy headlines to matter more than reported earnings. Over 1–3 months, legislative scope and procurement language are the catalysts; over 6–18 months, actual tender awards and customer migration decisions determine whether sovereignty produces meaningful share shifts.

Contrarian point: tighter rules need not be negative for US hyperscalers if they can satisfy requirements through sovereign products. The more adverse outcome is not “Europe buys less US cloud” in the abstract, but rules that make corporate origin itself disqualifying. No trade on the study alone; the earnings exposure is not established and is diluted within AMZN, GOOG, and META.

Key Decisions for Investors

• Do not trade the €115bn headline as a direct earnings estimate. Track the final CADA text, especially whether Article 31 reaches essential private-sector workloads and whether eligibility hinges on technical safeguards or provider ownership/personnel.
• Watch European public-cloud tender wins, sovereign-cloud contract disclosures, and cloud growth commentary from Amazon and Alphabet. A shift in those indicators—not the study’s modeled cost—is the evidence needed to underwrite a share-loss thesis.
• Conditional positioning: if the legislation permits US providers to qualify through technical controls, consider AMZN or GOOG on policy clarity rather than headline volatility; keep sizing modest because cloud exposure is only part of each company. Falsify the bullish read if final rules exclude foreign-controlled providers or major tenders explicitly favor European operators.
• If ownership-based restrictions expand beyond public procurement, reassess European cloud providers as potential relative beneficiaries and US hyperscalers as exposed incumbents. Do not initiate that trade until scope and tender evidence confirm the scenario; migration cost estimates alone do not establish who captures the displaced workloads.