AI for Americans First — Fabrice Pizzi, Universite Paris Sorbonne

Documents

Full study available in PDF — select language below. French edition complete; English and Portuguese editions in progress.

Ch.	Title	PDF
I	Theoretical Framework: Technological Protectionism and AI
II	Methodology: Scenario Matrix and CACI Index
III	Empirical Diagnosis 2020–2026
IV	Mechanisms of US Competitive Advantage
V	Prospective Scenarios 2026–2030 and Tipping Points
VI	Consequences for France and Europe
VI bis	Consequences for South America and Brazil
VI ter	Consequences for Asia
VI quater	Consequences for Africa
VII	Strategic Recommendations
Concl.	From AI Protectionism to the Reshaping of the Technological Order

Q&A

Questions & Answers — Volume 1

Detailed responses to key questions raised by the study.

Q&A

Questions & Answers — Volume 2

Extended discussion: regional scenarios, CACI methodology, strategic implications.

Q&A

Questions & Answers — Volume 3: CACI Ratio

In-depth exploration of the CACI index: methodology, interpretation, sensitivity analysis.

Appendix

Econometric Appendix — CACI

Full econometric analysis: panel data, PanelOLS, RandomEffects, robustness tests, CACI construction.

Working Paper

Working Paper — CACI AI Competitiveness

Detailed working paper on the CACI ratio and AI competitive advantage index.

Research Question

Artificial intelligence has emerged since 2023 as the primary driver of economic innovation and geopolitical competition. The AI value chain exhibits unprecedented concentration:

The United States controls 74% of global AI compute
Five US hyperscalers plan $660–690 billion in capex for 2026
Nvidia holds approximately 80% of the AI accelerator market

Research Question

To what extent does US AI protectionism create a measurable structural competitive advantage, and what are the differentiated consequences for France, Europe, and other regions of the world?

The originality of this research lies in the integration of four dimensions typically treated separately: energy, semiconductors, compute, and regulation. No prior study had formalized an integrated “Trump 2.0 — AI protectionism” scenario.

Key Findings

Export Controls

Inherited from Biden, transformed by Trump. The world segmented into three tiers: unlimited access for 20 close allies (Tier 1), quantitative caps (Tier 2), prohibition for adversaries (Tier 3).

Section 232 Tariffs

25% on advanced AI semiconductors (January 15, 2026). Direct cost differential between US firms (exempt) and non-US firms.

Capital Gravity

$660–690B annual capex. Japanese ($550B) and Emirati investments converge on US soil, self-reinforcing compute concentration.

CACI Index — Key Result

The CACI quantifies a US/EU ratio of 7:1 to 12:1, reflecting:

Compute gap: US 75 GW vs EU 35 GW
FLOPs cost: $0.5/TFlop US vs $1.2–1.8/TFlop EU
AI productivity: +30% US, +12% EU

Compute Infrastructure & FLOPs Gap: Epoch AI GPU Clusters Dataset (2025), analyzing over 150 top-tier AI data centers globally to estimate distributed aggregate compute power.
Energy Consumption & Grid Capacity: International Energy Agency (IEA) Electricity 2024 Report, evaluating nuclear vs. renewable grid stability for 1-GW-scale AI Factories.
Semiconductor Market & Foundries: Semiconductor Industry Association (SIA) and McKinsey reports on advanced logic fabs and HBM packaging bottlenecks.
Export Controls & Regulatory Shifts: Bureau of Industry and Security (BIS) rulings and Trump 2.0 proposed Section 232 tariffs (January 2026 scenarios).

Paradox 1

Chinese Acceleration

Restrictions accelerate the build-out of an autonomous Chinese AI ecosystem: Huawei Ascend 910c, DeepSeek-V3, $125B+ in 2025.

Paradox 2

Tier 2 Bifurcation

Brazil, India, ASEAN pushed toward Chinese partnerships (ByteDance: $38B in Brazil, $8.8B in Thailand).

Paradox 3

Allied Co-financing

Tier 1 allies (Japan, Korea) co-finance US supremacy ($550B Japanese investment) rather than building their own autonomy.

Comparative Regional Analysis

Differentiated impact of AI protectionism across five regions — structurally distinct dependency trajectories.

Region	Tier	Primary Dynamic	Strategic Asset	Key Risk
France / Europe	Tier 1	GPU+cloud dependency on US (72–80%); InvestAI €200B	Nuclear 70%, Mistral, ASML, AI Act	Geopolitical vendor lock-in
Brazil / S. America	Tier 2	US-China competition ground; TikTok megaprojects	83% renewable energy mix; fintech	Triple regional fracture
Japan / Korea / Taiwan	Tier 1	US co-investment ($550B); production transfer to US	HBM, TSMC 90% advanced chips	Asymmetric partnership; silicon shield erosion
India	Tier 2	Global South pivot; $200B+ ambition	1.4B population, tech talent	Structural gap (1.4 GW vs 53.7 US)
China	Tier 3	Forced autonomization; Huawei/DeepSeek ecosystem	1.4B domestic market; $125B+/yr	2–3 generation GPU lag

Source: author's construction from IEA, McKinsey, SIA, Epoch AI, BIS data (2020–2026).

Strategic Recommendations for France

Five strategic axes — critical action window 2026–2028.

Compute Infrastructure

13 EU AI Factories, Special Compute Zones, 5 AI Gigafactories InvestAI (€20B). Target: 30–40% of sensitive AI workloads on certified sovereign cloud by 2029.

Nuclear Energy

French advantage: 70% nuclear electricity. EDF 4 sites / 2 GW. Nuclear for AI 250 MW by end of 2026. 6 EPR 2 (Penly, Bugey, 9,900 MW).

Technology Alliances

Consolidate ASML-Mistral (€1.3B, ASML 11% stake). Second TSMC European investment. EU-Japan, EU-Korea bilateral agreements. GPU strategic reserves (6–12 months).

Offensive Regulation

AI Act as competitive lever. Priority to European models in public AI Factories. Brussels Effect. European CLOUD Act Shield.

Talent

AI scholarships and European talent visas (before end 2026). Frontier compute access: Fluidstack 500,000 GPUs, Mistral Compute, EuroHPC AI Factories.

Conclusion

The objective is not technological autarky but the capacity to choose. The question is no longer whether the reshaping of the global technological order will occur — it is underway — but whether we will be its architects or its subjects.

Study Structure

ITheoretical framework: technological protectionism and AI12 p. · 22 n.
IIMethodology: scenario matrix and CACI index8 p. · 10 n.
IIIEmpirical diagnosis 2020–202611 p. · 20 n.
IVMechanisms of US competitive advantage9 p. · 19 n.
VProspective scenarios 2026–2030 and tipping points11 p. · 16 n.
VIConsequences for France and Europe10 p. · 14 n.
VI bisConsequences for South America and Brazil11 p. · 19 n.
VI terConsequences for Asia12 p. · 16 n.
VI quaterConsequences for AfricaComing soon
VIIStrategic recommendations11 p. · 18 n.
Concl.From AI protectionism to the reshaping of the technological order8 p. · 3 n.

Total: 103 pages · 157 notes

Figures

Citation

@techreport{pizzi2026aifirst,
  author      = {Pizzi, Fabrice},
  title       = {AI for Americans First: US AI Protectionism,
                 Reshaping of the Global Technological Order
                 and Consequences for France and Europe (2026-2030)},
  institution = {Universite Paris Sorbonne},
  year        = {2026},
  month       = {February},
  type        = {Academic Research},
  note        = {103 pages, 157 notes, 22 figures}
}

License & Disclaimer This work, “America-First-IA”, is licensed under the Creative Commons Attribution – NonCommercial – ShareAlike 4.0 International License (CC BY-NC-SA 4.0).

You are free to share and adapt the material for non-commercial purposes, provided you give appropriate credit to Fabrice Pizzi (Universite Paris Sorbonne) and distribute your contributions under the same license.

This document is provided for educational and research purposes only.