Sovereign AI and the race for data control and economic opportunity

The race for data control and economic opportunity

Boardroom Bytes is the new HotTopics column where our editorial team explores the news making waves at the intersection of enterprise technology, business strategy and C-suite leadership. 

Today, we’re diving into the latest fanfare around artificial intelligence: the rise of sovereign AI.

Sovereign AI and the race for data control: Overview

• TDLR: What is Sovereign AI?
• Why it matters
• What it means for the C-suite: The HotTopics take

TLDR: What is Sovereign AI?

Sovereign AI refers to artificial intelligence systems that are developed, deployed and regulated within a specific country or region to ensure data sovereignty, security and compliance with local laws. Or, in other words, sovereign AI represents a singular nation or region’s ability to produce their own AI systems that utilise local infrastructure, data, workforces and business networks.

Sovereign AI is an attractive proposition for nation states because it arguably reduces the risk and the reliance on US-based technology behemoths while giving control to local jurisdictions on data processing and decision-making. 

Furthermore, with a focus on local talent and infrastructure, sovereign AI could help nations drive innovation and chase economic growth in an AI market expected to contribute over $15 trillion to the global economy by 2030.

Why it matters 

To understand the value of sovereign AI, you have to understand the intersection of Silicon Valley dominance, tech deglobalisation, a patchwork of AI (and data protection) regulation and governments as enthused about the transformative effects of AI as enterprise executives.

Just take outgoing U.S President Joe Biden, who late last year issued a national security memorandum ordering an assessment of the country’s  ‘relative competitive advantage’ in AI. 

Big Tech has seen this same shift, to the point where Nvidia CEO Jensen Huang has called this a ‘second wave of AI investments’, where ‘every region and every country needs to build their sovereign AI’ in order to serve their specific cultural needs and leverage their unique business strengths. The CEOs of other key hardware players, such as Dell and IBM, have followed suit with similar comments.

Small wonder then that countries such as Canada, Denmark, France — already one of the big AI players with LLM start-up Mistral – India, Italy, Japan, Singapore, Sweden, Thailand. and the USA have made substantial investments into new sovereign AI strategies. The Middle East has become a hive of AI activity, while just last week, the UK established a 50-point plan for AI ‘dominance’, including the launch of a new supercomputer and localised ‘AI Growth Zones’.

This is not to say Big Tech is completely altruistic in their support of localised AI. 

Besides the obvious appeal of multi-million (or billion) dollar government contracts, the insatiable demand for infrastructure and data mean that there’s an almost infinite need for high-performance compute power and data processing, preferably localised to the region of use. 

By localising AI systems, technology providers also get closer to local regulatory approval and to resolving issues such as AI bias and discrimination. Building and training LLMs on local datasets is a good example of this. By using local people with local dialects, there’s arguably less likelihood of unconscious or conscious bias creeping into AI decision-making.

What it means for the C-Suite: The HotTopics take

In a cyclical technology industry, technology executives could argue we’ve been here before; we’ve had similar conversations in the past on data sovereignty with cloud providers, the overreliance on Big Tech hyperscalers and deglobalisation in the supply chain (see the continued investments in global semiconductor plants).

Nonetheless, sovereign AI could challenge C-suite leaders to rethink strategies around data sovereignty, compliance and vendor reliance.  It also means balancing innovation with the growing number of local regulations, and scrutinising AI investments against a backdrop of geopolitical and economic pressures — not straightforward at a time when GenAI's ROI has increasingly been called into question.

At the same time, while sovereign AI appears to be as much about economic gains and opportunity as technological deglobalisation and de-risking, this is not to say there are no detractors. 

Cost, resourcing, isolated tech development and scalability are potential barriers to sovereign AI taking hold, as are the possibilities of government oppression and varying maturity levels emerging between nations. AI ethics continues to be a challenge too; in one of the more unusual tales, Denmark’s new AI ‘Gefion’ supercomputer has reportedly been bankrolled by pharmaceutical giant Novo Nordisk, which is behind Ozempic, the celebrity drug of choice for weight loss. 

An alternative approach for sovereign AI, as promoted by the World Economic Forum, could be a Global AI Compact, where compute power is shared as a central resource, like electricity. 

As sovereign AI evolves, will nations collaborate to create a balanced global AI ecosystem, or will we see greater fragmentation? Let us know what you think in the comments below or on our LinkedIn page.