Title: Policy Brief: Europe, the United States and the AI Bifurcation
Tags: geopolitics, compute-sovereignty, ai-regulation
---
A polemical brief written in December 2025 in reply to statements by Donald Trump and Elon Musk, and to the new United States National Security Strategy, about the "civilizational decline" of Europe and the European Union's regulatory burden. It applies the projections of the author's paper *Thermodynamics of Cognitive Power: When {1=1} Breaks the Fortress* to that dispute.

The thesis is that between 2027 and 2030 the AI ecosystem splits in two: centralized cloud services for training and frontier models, and **sovereign compute**, in which organizations and individuals run capable open models on their own hardware. On this reading Europe is well placed. The brief cites announced European investments in AI infrastructure, the demand for data sovereignty in health and finance, and regulation (GDPR, the AI Act) as a force that creates a market for locally run systems. It estimates that market at 35 to 55 billion euros by 2030, beside 50 to 80 billion for enterprise cloud.

The risk is placed on the American side: an oligopoly of a few cloud providers that fails to adapt is compared with Nokia after the arrival of the smartphone. China is described as sovereign by necessity, because of export controls, and as the likely beneficiary of a transatlantic quarrel. The second quarter of 2027 is named as the point of no return, tied to the price of consumer hardware and to the narrowing distance between open and proprietary models. An interview with Jared Kaplan of Anthropic about the decisions of 2027-2030 is read as support for distributed inference as a safeguard.

The brief closes with a scorecard and an open letter, and names 2027 as the year that will show who was right.