CIOInsider India Magazine

Last month, during the India AI Impact Summit in New Delhi, the Indian government revealed plans to enhance national computing capabilities and focus more on creating AI models within the country. This change in perspective suggests that Indian policymakers now view AI not just as a secondary technology, but as a vital strategic asset.

This transformation is part of a larger worldwide trend known as the "sovereign AI agenda," where countries are reconsidering the control of crucial components like chips, cloud infrastructure, data, models, and applications that have a significant impact on areas such as public administration, economic growth, and democratic governance.

Avoiding Dependence on Foreign Technology
One of the most concerning risks lies within the domain of national security. Picture a future where an AI system has control over a nation's power grid, healthcare system, financial markets, and military defense. Now, envision that this AI system is at the mercy of a foreign entity that could potentially cut off access during a crisis. This situation, known as an "AI embargo," could have devastating effects similar to a modern-day blockade. This vulnerability cannot be overlooked by any government.

Also Read: Nebius Signs AI Infrastructure Deals with Meta

Developing a Sovereign AI is crucial for national defense as it ensures the security and continuity of vital infrastructure, free from external influence. It also guarantees that sensitive government and citizen data remains within the country's borders, safeguarded by local laws and shielded from foreign surveillance. The most concerning risk lies within the realm of national security.

Nations that are unable to develop, educate, and utilize models within their own borders run the risk of becoming reliant on digital technologies from other countries. This can lead to a situation where they are only consumers of external innovation and lack the ability to influence or govern it due to a lack of legal jurisdiction.

Canada Taking a Lead with its Three Pillared AI Strategy
Canada is making significant efforts to enhance its control over AI by implementing a $2 billion Sovereign AI Compute Strategy. This initiative is designed to offer researchers, startups, and innovative companies with access to robust computing resources.

Canada is improving its internal AI capabilities by investing in infrastructure and prioritizing secure data management, increased innovation access, and self-sufficiency. The Canadian Sovereign AI Compute Strategy is based on three core principles. The AI Compute Challenge aims to encourage private sector funding for AI data centers with a potential investment of up to $700 million.

The AI Sovereign Compute Infrastructure Program (SCIP) seeks to establish a cutting-edge public supercomputing system that is completely Canadian-owned and operated, with a potential budget of up to $705 million. The AI Compute Access Fund looks to provide subsidies for compute resources to small and medium-sized enterprises as well as research institutions to overcome financial obstacles, with a potential allocation of up to $300 million.

IndiaAI Mission: Building a Self-Reliant AI Ecosystem
In 2024, India launched the IndiaAI Mission, a national initiative aimed at enhancing its control over AI technology. With a budget of Rs.10,371.92 crore, the mission aims to create a self-sufficient AI ecosystem by integrating computational infrastructure, locally developed models, open datasets, and ethical regulations.

The plan will consist of deployment of over 10,000 GPUs for both public and private purposes. An IndiaAI Innovation Center will be set up in order to construct specialized foundation models in various domains. Going forward an IndiaAI Datasets Platform will be built for accessing secure and varied non-personal data. Additionally, the future will see the establishment of an AI Safety Institute and introduction of preliminary governance guidelines by MeitY.

Japan Advances AI with ABCI 3.0 Supercomputer
Japan is enhancing its control over AI through the development of ABCI 3.0, a project spearheaded by the National Institute of Advanced Industrial Science and Technology (AIST) in partnership with Hewlett-Packard Enterprise (HPE) and NVIDIA. ABCI 3.0 is engineered to provide an impressive 6 AI exaflops of processing power, utilizing Quantum-2 InfiniBand networking and numerous NVIDIA H200 GPUs, thus establishing itself as a globally renowned and highly accessible AI supercomputing system. The joint effort exemplifies Japan's dedication to improving its AI proficiency and fostering its own technological autonomy.

France Champions Sovereign AI with Trusted Cloud
France has shown its commitment to sovereign AI through its investment in secure cloud infrastructure, aiming to maintain national authority over data and processing. The nation's initiative, introduced in 2021, introduces the concept of the "trusted cloud" label overseen by ANSSI's SecNumCloud certification. By utilizing this framework, businesses and government entities can run AI operations on a platform that is shielded from external intrusion, a crucial aspect of sovereign AI.

Also Read: Center Launches Nationwide Survey of Manuscripts Under Gyan Bharatam

To facilitate large-scale implementation in industry, France is endorsing hybrid cloud models such as Bleu, a collaboration between Orange and Capgemini that utilizes Google technology while adhering to rigorous legal protections. This initiative is bolstered by an initial funding of 107 million euro through the government's Programme d’investissement d’avenir.

UAE: Building Secure, Transparent AI Infrastructure
The UAE's vision for the future includes becoming a dominant force in artificial intelligence by implementing strong regulations surrounding data, advancing national infrastructure, and promoting ethical practices. The country acknowledges the importance of data in the modern world and is dedicated to creating a secure environment for data storage, ensuring the protection of sensitive information, and facilitating the growth of AI technology through effective model training. At the core of its plan is the development of standardized, transparent, and AI-compatible data sets that are gathered and controlled within the country. This involves constructing systems such as Estonia's X-Road and Australia's SURE to provide secure, ethical, and authorized entry to government data, safeguarding data independence in industries like healthcare, banking, and power.

The Challenge of True AI Sovereignty
Even though AI is being portrayed as independent, achieving real sovereignty in the field still proves to be difficult. Currently, there are two main frameworks guiding countries in their endeavors to create autonomous AI environments:

The Stargate Plan: Within OpenAI's initiative for Countries, nations have the ability to swiftly implement AI services like ChatGPT tailored with local data governance, region-specific APIs, and safety measures. This method utilizes data centers within the country and localization controls while still utilizing US-origin models, chips, and cloud infrastructure. However, critical elements like model architecture, foundational training data, and semiconductors are not under national control.

Also Read: Karnataka Forms AI Panel Led by Kris Gopalakrishnan

In-house sovereignty: Various countries are striving to create their own AI frameworks through open-source models, public datasets, and local infrastructure. For instance, India is developing multilingual LLMs like Bhashini, establishing GPU clusters through the IndiaAI Mission, and investing in data platforms like AI Kosh. France has also created models like Mistral and viable alternatives to mainstream search engines such as Qwant. Yet, these initiatives are closely tied to global supply chains. France is collaborating on AI data centers with the UAE, while India still depends on imported chips and foreign cloud infrastructure. Even the most independent efforts are seldom without strategic and technical complications.

A Fragmented AI World
The world is not trending towards complete AI domination, but instead heading towards a divided AI environment with countries adopting different strategies centered around two main influences:

The US influences various regions such as Europe, the Gulf, and parts of the developing world. These areas heavily depend on technology developed by US companies like GPT and Claude, chips from NVIDIA and AMD, as well as services provided by giants in cloud computing like AWS, Azure, and Google Cloud.

The realm of China: This encompasses nations that are connected to China's technological network via initiatives like the Belt and Road project or services from Huawei in AI and cloud computing.

Developing Countries Navigate AI Uncertainty
In the midst of uncertainty, developing countries are looking for ways to protect themselves strategically. They are determined to avoid getting caught in the political turmoil caused by trade disputes, limitations on exports, or bans on technology.

Currently, no country has complete control over all aspects of AI technology. The global situation is evolving towards a system of layered sovereignty, where nations strategically choose which parts of the AI realm they want to dominate based on their strengths and limitations.