Nvidia pitches national AI factories for governments

Nvidia has outlined a framework for countries to build domestic AI infrastructure and local AI models, arguing that governments increasingly see AI as a strategic national priority.

National AI efforts now centre on building local computing capacity, training models on domestic datasets and developing local skills. Nvidia argues this approach lets countries keep data and model development under local rules while tailoring systems to local languages, public services and industries.

The company set out five elements of a national AI strategy: treating AI as an economic and security priority, building an AI-ready workforce, training models with local data, supporting a domestic AI ecosystem and creating locally owned AI factories. It uses the term "AI factories" to describe data centres designed for AI training and inference.

Countries are pursuing different ownership models for this infrastructure. Some are working with state-owned telecommunications groups or utilities to procure and run AI cloud systems, while others are backing local cloud providers that can offer shared computing resources to public and private users.

The argument reflects a wider shift in government technology policy. Many states have long invested in domestic infrastructure for transport, communications and healthcare. Nvidia is positioning AI infrastructure as the next layer of that national investment.

Strategic push

The rise of generative and agentic AI has increased pressure on governments to establish local model development and deployment. Nvidia says the technology is reshaping markets and changing work across sectors including gaming and healthcare, while also being used in software coding, drug discovery, fraud prevention and robotics.

Nvidia also links national AI investment to resilience in climate, energy and cybersecurity. It says accelerated computing is becoming more important in efforts to improve energy efficiency, respond to cyber threats and address wider sustainability challenges.

Local datasets are a central part of the approach. Nvidia says domestically trained foundation models and large language models can better reflect regional dialects, cultural context and specialist domains. It adds that speech AI systems can also support the preservation and revival of indigenous languages.

The company also ties this to governance as well as performance. Local hosting and operation of AI models means systems can be subject only to local laws and policies, rather than external rules or overseas infrastructure constraints.

Examples abroad

To illustrate the approach, Nvidia points to projects in Europe, Asia and Latin America that use its technology in public services. The examples focus on automation in government, multilingual language systems and legal administration.

In France, AI agents from ThinkDeep are being used by the Ministry of Economy and Finance to automate public-service workflows by processing large volumes of documents and data sources, according to Nvidia. The company says this cut document search times from two days to two minutes, saved 2 million euros for 10,000 employees and reduced energy use through more efficient in-country infrastructure control.

In India, Nvidia highlighted the Sarvam platform, which it says was built entirely on domestic infrastructure and uses Nvidia GPUs to deliver multilingual AI models and voice agents across the country's 22 official languages. According to the company, the aim is to support government and enterprise services while keeping data, computing resources and governance under national control.

In Brazil, Nvidia cited work by Widelabs for the Public Ministry of Rio Grande do Sul. The company says the AI systems are helping staff streamline internal investigations and make justice records easier to search and use for more than 8 million citizens across nearly 500 municipalities.

Industry context

The emphasis on sovereign or national AI has grown as governments weigh the economic and political risks of relying on foreign cloud providers and imported models. For Nvidia, the trend also points to a growing market for advanced chips, AI systems and the software stack needed to run them.

Nvidia says its AI Nations initiative has worked with countries since 2019 to help develop AI ecosystems and workforce training. The goal, it says, is to create conditions in which engineers, developers, scientists, entrepreneurs, creators and public-sector officials can pursue AI work in their home markets rather than relying entirely on imported infrastructure and expertise.

Jensen Huang, Founder and Chief Executive Officer of Nvidia, has previously framed AI infrastructure in broad economic terms. "The AI factory will become the bedrock of modern economies across the world," Huang said.