AI advantage becomes a 'strategic decision' as nations build local AI grids
The race is now on to build "National AI Grids" to protect sovereignty.
The shift toward sovereign AI factories is fundamentally changing how leadership views technological investment, moving the conversation from general efficiency toward strategic national infrastructure.
Abhishek Srivastava, principal at Arthur D. Little, suggested that AI is evolving into a fundamental layer of national infrastructure, similar to how telecom was in the previous century.
Speaking at the Asian Telecom Summit 2026 held in Singapore, he noted: “Very early, countries and companies recognised that telecom is national infrastructure. It is fundamental to national sovereignty… and we expect a similar fundamental nature of AI to play out. Increasingly, AI is quite important, not just for efficiency, for better, faster, etc., but also for national security.”
Srivastava said this shift means every country will need to develop its own national AI system. “We anticipate as this evolves, each country will build its own national AI grid, and companies and enterprises will build national AI factories on top of that.”
According to Srivastava, roughly a third of enterprises are already exploring their own AI capabilities. This involves not just deploying infrastructure like GPUs, but also fine-tuning and adapting foundational models to create a competitive advantage, rather than relying solely on off-the-shelf solutions.
Low-strategic-relevance tools such as co-pilot or ChatGPT are expected to remain foundational, but investment focus is shifting toward creating products and services using enterprise data that differentiate companies in the market.
Srivastava cited the digital transformation of DBS as an example, noting how the bank built an AI factory on top of foundational models to gain a competitive edge: “We have seen how the digital transformation of DBS helped it to become the best bank in the world, and we anticipate a similar cycle on the AI front to differentiate the winners versus losers across industries.”
This approach requires more than capital. For DBS, talent investment was fundamental, with AI experts centrally pooled and deployed across all functions to maximise impact.
To avoid commoditisation, organisations are adopting multi-tenanted platforms, allowing them to choose multiple GPU providers and build vertical, narrow AI solutions tailored to specific industry needs.
Srivastava stressed that whilst a handful of companies currently provide most of the world’s AI computing, that will not remain the case: “In future, each country, or ideally, groups of countries, will take ownership of not just their data, but also the AI infrastructure that goes with it.”
The AI factory approach is becoming mainstream, and telcos — traditionally guardians of trust — are expected to play a critical role as AI infrastructure orchestrators. In the current geopolitical environment, Srivastava said, “AI advantage will be a strategic decision… the one who leads in building this out with the right setup would win.”
