Sovereign AI: How governments are seeking technological independence in artificial intelligence

This article has been translated using artificial intelligence

Artificial intelligence (AI) is evolving beyond a purely digital technology to become a key component of the physical world. From autonomous robots to intelligent assistance systems, Embodied AI is redefining how we interact with machines. At the latest Future Trends Forum, the Fundación Innovación Bankinter brought together 40 international experts to examine the impact of this revolution on industry, mobility, and society.

At this event, David Shrier, Professor of AI and Innovation at Imperial College London and CEO of Visionary Future, offered a crucial perspective: beyond human-machine interaction, AI has become a strategic and geopolitical issue. His talk focused on the concept of sovereign AI—a model in which governments aim to develop their own artificial intelligence systems rather than relying solely on Big Tech. In a world where AI increasingly influences the economy and national security, the question is: who should control it?

If you’d like to watch David Shrier’s presentation, here you can do so:

David Shrier: “Sovereign AI Highlights” #EmbodiedAIForum

AI sovereignty: why governments are reacting

Artificial intelligence is transforming the global economy. According to David Shrier, up to 60% of the workforce could be affected by automation, yet at the same time, AI could add $13 trillion to global GDP by 2032, based on a study conducted with Evercore ISI.

Beyond the disruption to employment, AI can also boost productivity. A recent study by Harvard Business School and Boston Consulting Group (BCG) indicates that workers collaborating with generative AI tools can significantly improve their performance. According to this analysis, professionals who used AI completed tasks 25% faster and with 40% better quality than those who didn’t use it.

This impact hasn’t gone unnoticed by governments, which see AI as more than just a technological tool—it’s a strategic asset. Until now, AI development has been in the hands of a few corporations like OpenAI, Google DeepMind, and Anthropic. However, more and more countries are pushing strategies to build their own AI, ensuring that their values, languages, and economic priorities are reflected in these systems.

What is sovereign AI?

Shrier defines sovereign AI as a model in which governments, in collaboration with companies and multilateral organizations, develop their own AI systems rather than relying solely on foreign private actors. But he draws a key distinction: sovereign AI doesn’t mean government-controlled AI. It’s not about governments directly managing AI models, but about setting the conditions for development aligned with national interests.

This paradigm shift is driven by several factors:

• Geopolitics and technological autonomy: The trade war between the U.S. and China has shown that technology is a strategic weapon. Some governments fear that future restrictions could limit their access to AI from American companies.
• Data sovereignty and cultural values: OpenAI and other companies have trained their models primarily on English-language data with an Anglo-Saxon cultural bias. Several countries argue for developing AI models that reflect their own languages, values, and social norms.
• Resilience against political changes: A potential return of Donald Trump to the U.S. presidency (note: this talk was given in early December 2024) could lead to new technology restrictions, blocking access to advanced AI models for other countries.
• Security and privacy: AI is being used in critical sectors like healthcare, finance, and defense. Not controlling these systems poses a national security risk.

The challenges of sovereign AI

Shrier warns that AI sovereignty is neither an easy nor a cheap goal. Developing an advanced AI model can cost up to $1 billion, and keeping it operational requires robust digital infrastructure. Countries must also consider:

• Cybersecurity: AI models can be vulnerable to attacks that manipulate their outputs or steal sensitive data.
• Regulation and governance: Policies must strike a balance between innovation, control, and ethics.
• Digital infrastructure: Only a few countries have the technical and economic capacity to develop their own models. For others, multilateral alliances may be the answer.

One example is JAIS, a consortium driven by several Arab countries to jointly develop AI. Initiatives like this help share costs, talent, and technological capabilities.

How should countries respond?

The rise of sovereign AI has put governments in a complex position: how to develop strategies to avoid exclusive dependence on Big Tech without falling into inefficiency or wasting resources. David Shrier suggests countries should address AI sovereignty across several key dimensions:

1. Create regulatory frameworks and national AI policies
   Shrier mentions that many countries are crafting or revising national AI strategies to ensure AI aligns with their economic, cultural, and security interests. These policies should cover:

• Governance and regulation: Clear standards for AI use, focusing on transparency, security, and ethics.
• Pro-innovation: Avoiding overly restrictive regulations that could stifle the development of local startups and tech companies.
• Data protection: Ensuring AI is trained on information that represents the country, without depending solely on foreign data.

Many governments have already started down this path, from the AI regulations in Europe to the push for national models in regions like the Gulf and Asia.

2. Invest in digital infrastructure and data sovereignty
   One of the biggest hurdles for sovereign AI is its immense cost. As mentioned earlier, training an advanced AI model can cost $1 billion, and maintaining it requires a robust infrastructure. Shrier stresses that governments need to:

• Develop national data centers and enhance storage and processing capacity.
• Ensure data sovereignty by making sure AI models reflect national realities and values.
• Explore public-private partnerships to reduce costs and foster collaboration with industry.

3. Cybersecurity and resilience against threats
   Shrier highlights that governments are increasingly aware of AI-related risks—disinformation, cyberattacks, and model manipulation. Protecting AI systems is key, and this means:

• Implementing strategies to detect and mitigate attacks on AI models.
• Regulating access to critical technologies to prevent external interference.
• Protecting the integrity of data used in AI training.

4. Develop talent and AI training
   For sovereign AI to be viable, countries need AI experts—still a challenge in many regions. Shrier notes that some governments are investing in:

• AI education and training programs in universities and research centers.
• Incentives to attract and retain tech talent.
• Support for startups and AI hubs to stimulate local innovation.

5. Multilateral alliances to share resources and costs
   Shrier emphasizes that developing advanced AI models is out of reach for many countries due to the costs involved. However, some governments are turning to international consortia like JAIS. These collaborations allow:

• Sharing data and compute infrastructure.
• Developing AI models trained on data from multiple regions.
• Reducing dependence on Big Tech without taking on all the costs alone.

Countries face a dilemma: build their own AI from scratch or rely on foreign models? According to David Shrier, the key is to strike a balance between independence and collaboration, ensuring AI reflects national values and needs without isolating from the global ecosystem.

The future of sovereign AI is still taking shape, but one thing is certain: it will play a critical role in the economy and geopolitics of the coming decade.

Conclusion: The future of AI will be geopolitical

Artificial intelligence is no longer just a technological issue. As Shrier emphasized at the Future Trends Forum, who controls it will determine which countries lead the economy of the future and which are left behind. The question is no longer whether governments should develop their own AI, but how they can do so sustainably without falling into authoritarian models.

At the Future Trends Forum, we will continue to explore the impact of AI on society, the economy, and global geopolitics. Stay tuned for our next analyses!