Sovereign Artificial Intelligence
Tom Everill | 10 April 2024
In November 2023, Nvidia CEO Jensen Huang, in an interview with iliad Group Deputy CEO Aude Durand, suggested that the world is witnessing a second wave of AI adoption driven by two major catalysts. Firstly, what he describes as recognition by governments that every country must build ‘sovereign AI’ capacity, and secondly, the adoption of AI in new industries. Huang argues that sovereign AI is the only way for countries to serve their specific cultural, language, and regulatory needs and leverage their unique business strengths in the era of AI.
A blog post by Nvidia from 28th February points to the long-standing tradition of nations investing in domestic infrastructure to harness power and economic gains and argues that this has never been more vital for any technology than AI. Nvidia describes sovereign AI as a way of nations owning ‘the production of their own intelligence’. At the World Governments Summit in Dubai this February, Huang told the UAE’s AI Minister, Omar Al Olama, that this type of state-owned AI capacity allows nations to codify their culture, societal intelligence and history.
According to Nvidia, the foundation of sovereign AI capability will be ‘AI factories.’ These factories will serve as the essential infrastructure into which data is input and intelligence is output. These next-generation data centres will host advanced, full-stack accelerated computing platforms intended to facilitate intensive tasks. Currently, two main models are being used to build up domestic AI capacity. Some nations are procuring and operating sovereign AI cloud infrastructure in collaboration with state-owned telecommunication providers, whereas others sponsor local private cloud providers to deliver shared computing platforms for use in both the public and private sectors.
Many countries are already investing heavily in sovereign AI capabilities, most notably those partaking in Nvidia’s AI Nations Initiative, launched in 2019. The initiative focuses on ecosystem enablement, workforce development, and creating favourable conditions for individuals and firms to pursue their AI ambitions. The following is a list of some key developments in the realm of sovereign AI:
France:
French-based firm Scaleway is building Europe’s most powerful cloud-native AI supercomputer.
Switzerland:
Swisscom Group (majority owned by the Swiss Government) announced recently that its Italian subsidiary Fastweb will build Italy’s first and most powerful supercomputer powered by Nvidia DGX servers.
Fastweb intends to launch an end-to-end system for Italian companies and public administration organisations in different fields.
India:
Tata Group is currently developing large-scale AI infrastructure.
Reliance Industries plans to develop a large-language model (LLM) that is trained on India’s diverse range of languages.
Nvidia is working with top universities in the country, to help expand local AI research and development communities.
Japan:
The Japanese Government is working with Nvidia to upskill its workforce, support LLM development in the Japanese language, and build AI tools to help respond to natural disasters
Nvidia has also partnered with telecommunications provider Softbank Corp. to build generative AI platforms for 5G and 6G applications, as well as a network of distributed AI factories.
Singapore:
The Singaporean Government is partnering with Nvidia to upgrade its National Super Computer Center.
Leading communications services provider Singtel is building ‘energy-efficient’ AI factories across Southeast Asia.
United Kingdom:
Cambridge University, Intel and Dell collaborated to build, Dawn, the UK’s fastest AI-capable supercomputer.
Huang and Nvidia are not the only high-profile proponents of sovereign AI. IBM Corp. CEO Arvind Krishna made a similar statement as Huang’s, calling on his home country of India (and others) to develop state-level AI capabilities, including large language models. There are, however, opponents to the concept. In November 2023, US Ambassador at Large for Cyber Space and Digital Policy, Nate Fick, testified before the Senate, stating that, ‘there is no room for digital sovereignty.’ Instead, he suggests, the United States must focus on ‘digital solidarity’ to coordinate regulatory interoperability and R&D with close allies. The logic is business-focused, arguing that companies want large, harmonised markets involving integrated efforts between countries, the world’s best universities and highly skilled labour.
It must be remembered that Nvidia, as the leading proponent of sovereign AI today, stands to gain enormously if states adopt its semiconductors for use in their national computing infrastructure. Earlier this year, driven by mass AI adoption, Nvidia’s market cap surged to over USD 2.3 trillion (GBP 1.82 trillion), surpassing Alphabet and Amazon to become America’s third most valuable company behind Apple and Microsoft. However, according to GlobalData’s report titled 2024 Enterprise Predictions: AI, Nvidia is set to lose some of its market dominance due to rising competition in the AI Chip sector. Schemes such as its AI Nations Initiative serve as an attempt to consolidate the chipmaker's large market share moving forward.