A National AI Model Is Born
On July 1, 2026, Portugal unveiled 'Amália', its first national large language model (LLM). Named after the iconic fado singer Amália Rodrigues, the model is more than a piece of software; it's a declaration of intent. Unlike consumer-facing chatbots
like ChatGPT, Amália is designed as a foundational platform. This means its source code, models, and datasets are freely available for companies, universities, and public institutions to adapt and build upon. The project, backed by €5.5 million from EU recovery funds, was developed by a consortium of Portuguese universities and research institutions. The initial applications are telling: a virtual guide for museums, an AI assistant for teachers, decision-support tools for the Portuguese Navy, and a digital helper for public services.
The Open-Source Advantage
Choosing to make Amália open-source is a deliberate strategic decision. For a nation of just over 10 million people, competing head-on with the colossal R&D budgets of Silicon Valley or Beijing is unrealistic. Open-source offers a more pragmatic path. It allows Portugal to build national capability without the immense cost of training a frontier model from scratch. By making the system's inner workings transparent, the government can audit, inspect, and trust the technology it plans to wire into critical public services. This approach fosters a local ecosystem of innovation, allowing startups and researchers to create tailored applications for the Portuguese language and market, a niche often underserved by English-centric global models.
From Digital Dependence to Sovereignty
This is where technology connects directly with national autonomy. Policymakers across Europe are increasingly wary of becoming passive 'consumers' in an AI era dominated by a few foreign firms. Relying exclusively on proprietary models from the US or China creates strategic vulnerabilities. What if access is restricted due to geopolitical tensions? What if a foreign company's commercial priorities conflict with a nation's public interest? By developing its own sovereign AI capabilities, Portugal aims to mitigate these risks. Prime Minister Luís Montenegro stated the case plainly: “Europe's strategic autonomy is today, perhaps more than ever, tied to AI.” This move is about ensuring that the digital infrastructure of the future is not entirely controlled by external actors, giving the nation greater control over its data, its economy, and its public services.
A European Counter-Narrative
Portugal's initiative fits into a broader European push for 'digital sovereignty'. Countries like France and Germany have similarly backed homegrown AI companies like Mistral AI and Aleph Alpha. This trend represents a slight counter-narrative to the EU's primary focus on regulation, as embodied by the comprehensive EU AI Act. While fully complying with the Act, Portugal is also pursuing a proactive strategy of building and owning its technology. The Amália model itself builds upon a European foundation model called EuroLLM-9B, demonstrating a collaborative, pan-European approach to reducing dependency. This strategy is not about digital isolationism, but about creating credible regional alternatives and ensuring a diversity of players in the global AI landscape.
The Road Ahead Is Not Easy
Despite the optimism, Portugal's path is challenging. The government has an ambitious goal for 75% of its companies to adopt AI by 2030, but the country faces underlying issues like low wages and an ageing population. There is also the question of resources. While the Amália project leverages powerful supercomputers, maintaining a competitive edge in the fast-moving field of AI will require sustained investment far beyond the initial €5.5 million. Furthermore, some critics argue that relying on rented cloud computing power, often from the very US giants one seeks to gain independence from, creates an "illusion of sovereignty rather than the substance of it." The success of Portugal's push will be measured not by this single launch, but by whether it can foster a resilient and self-directed digital future over the long term.
