AI sovereignty : The new frontier of digital power

AS the country prepares for the India AI Impact Summit 2026, it is important to note that AI sovereignty was a central issue at the Paris AI Action Summit held in February 2025. Importantly, the concept of AI sovereignty primarily determines the direction of today's massive AI-related investments worldwide.

According to French philosopher Bernard Stiegler, one of the first and most influential theorists of the digital age, to comprehend the world, it is necessary to go beyond the conventional concept of technology. He maintained that "psychopolitics" — where media and technology have the ability to shape our minds and attention — defines the 21st century.

Stiegler contends that technology is a "pharmakon" — a poison and a cure at the same time — and that we must choose whether to let technology "use" us or to use it for "critical intensification" to reclaim our rational sovereignty from modern agencies that would take it away from us if they could.

However, the early 21st century saw a reorientation of the concept of "tech sovereignty", mostly as a result of geopolitical conflicts and sanctions. In the 2000s and early 2010s, internet sovereignty was first discussed in China, perhaps. Then, to avoid dependence on foreign powers in this AI era, various countries began exerting control over digital infrastructure, data and AI. The 2018 trade war between the first Trump administration and China further stoked the demand for sovereign AI.

AI sovereignty is the ability of a nation or company to independently control its AI technologies, including its data, infrastructure and models, to ensure national security, regulatory compliance and strategic autonomy. It entails avoiding reliance on foreign control points, maintaining sensitive data within a jurisdiction and developing and implementing AI systems within domestic frameworks.

To put it simply, sovereign AI seeks to guarantee that AI is produced domestically, including the data that the AI uses for training, explores during research and produces as output in answer to a query. Thus, "data sovereignty" and "AI sovereignty" are linked, yet distinct. But sovereign AI ensures that AI conforms to their cultural, ethical and legal standards.

Nations today know the risks and benefits of GenAI and their potential impact on national security and economic growth. For instance, by 2035, AI could boost India's economy by $1.7 trillion.

According to a World Economic Forum article published in April 2024 titled ‘Sovereign AI: What It Is, And 6 Strategic Pillars For Achieving It’, the push for sovereign AI doesn't necessarily imply digital isolation but a push for strategic resilience, which can be carried out in tandem with international cooperation.

The strategic pillars — which include digital infrastructure, workforce development, research, development and innovation, regulatory and ethical framework, stimulating the AI industry and international cooperation — will help countries enhance their sovereign AI capabilities.

Therefore, the path to sovereign AI is intricate and necessitates careful, long-term strategic planning at the national level. Among examples of sovereign AI in different countries are a government-funded AI model in Singapore which is able to communicate in 11 languages; ILMUchat in Malaysia, built by a local construction company; and Apertus in Switzerland.

South Korea intended to pour $735 billion into making "sovereign AI" built on Korean language and data. In India, in March 2024, the Union Cabinet approved the IndiaAI Mission, where over Rs 10,300 crore was allocated over five years with 38,000 GPUs deployed with the aim of "Making AI in India and Making AI Work for India."

Jensen Huang, CEO of Nvidia, declared in early 2024 that "every country needs sovereign AI." As the global race for AI sovereignty continues, it should be highlighted that sovereign AI services are mostly provided by some US and Chinese tech giants and countries are frequently compelled to long-term reliance on them for infrastructure and chips.

These big tech companies are influencing the discourse surrounding sovereign technology by essentially providing sovereignty as a service. Nvidia referred to the parts that make up this worldwide infrastructure as "AI factories". And it is installing chips and hardware infrastructure in countries around the world.

Microsoft has contracts with the UAE and other countries. Amazon Web Services provides an "AWS European Sovereign Cloud." Private businesses can also access sovereign clouds from Microsoft Azure and Google Cloud, which include data localisation and encryption. And different open-source and multilingual AIs have also been hailed as promoting digital and AI sovereignty at the model construction and dataset annotation levels.

However, rhetoric soon emerged characterising such ‘sovereignty as a service’ as a contemporary form of colonialism. For example, Laleh Khalili's June 2025 article in the London Review of Books titled ‘Collective Property, Private Control’ described this.

Historically, build-operate-transfer (BOT) schemes were used by European powers in the 19th century as a means of colonial expansion, where private, metropolitan companies supplied the funds, expertise and resources needed to build vital infrastructure, such as roads, railroads, ports, canals, telegraph lines, etc, either in their colonies or in places where their government was attempting to increase its power and influence. How much this resembles the present scenario of AI sovereignty — and where they differ — are components of a completely new political economy of international politics.

In any case, governments everywhere are now compelled to spend billions, in one way or another, on their own "sovereign" AI technologies. There are also proposed alternatives. For instance, to develop a competitive rival to the US and Chinese behemoths, researchers from the University of Cambridge's Bennett School for Public Policy suggested a public AI firm that would be divided among a group of middle-income nations.

In allusion to Europe's successful attempt to create a rival to US multinational corporation Boeing in the 1960s, they refer to it as "Airbus for AI." In fact, open-source AI can be a "cornerstone of digital sovereignty", serving as the basis for "autonomy, innovation, and trust" in countries all over the world, according to Hugging Face, a French-American firm based in New York.

AI sovereignty may prove to be an Achilles' heel for many middle-income countries. They must make wise decisions by weighing their ability to make financial investments, the anticipated economic return, data privacy and national security. In this AI era, it is an evolving dynamic where concepts of sovereignty are manifesting in engineering, technology and nation-state politics alike.