Weaving a Sovereign Web: Project Tapestry's Bid to Remake Global AI

PARIS, France – June 19, 2026 – As the leaders of the G7 nations convened in Évian this week, the air was thick with a new kind of geopolitical anxiety: artificial intelligence. The quiet consensus is that the world’s most powerful technology is being consolidated in the hands of a few Silicon Valley giants and their state-level competitors, creating a new form of digital colonialism. Most of the world consumes AI, but it does not shape it.

Against this backdrop, a radical new blueprint for the future of AI is taking shape. Here in Paris, the AI Alliance has announced a major expansion of Project Tapestry, a global consortium designed to build frontier-level AI as open, collaborative infrastructure. The latest move sees India’s sovereign AI company, BharatGen, commit to anchoring the nation's participation, a significant step for a project that aims to directly challenge the centralized model of AI development.

Project Tapestry is not just another open-source initiative. It represents a fundamental rethinking of how we build and govern the systems that will define the 21st century. It proposes a third way: a model where nations can pool resources to build world-class AI without surrendering control of their data, their digital infrastructure, or their destiny.

Sovereignty by Architecture, Not Just Trust

The central dilemma of our time is how to foster global collaboration without sacrificing national autonomy. Project Tapestry’s answer is not found in policy treaties, but in code. Its foundational principle is “sovereignty by design,” a concept that moves beyond flimsy promises of corporate goodwill.

“AI is becoming essential infrastructure,” said Dr. Yann LeCun, the project’s Chief Science Advisor. “No single company or country should determine who can build on it, adapt it, or benefit from it.”

The project’s architecture is a clever solution to this challenge. It employs a federated model where a central “base model” is developed by the consortium. This base is then distributed to participating nodes—nations, research labs, or companies. These nodes then continue to train and refine the model using their own private, local data, which never leaves their control. Instead of sharing raw data, they share back the “model updates”—the mathematical improvements and learnings. These are then integrated into the base model, creating a virtuous cycle that strengthens the collective intelligence while preserving individual sovereignty.

As one architect of the project explained, the guiding principle is that “sovereignty cannot depend on trust alone. It must be protected by architecture.” This is a crucial distinction. Participants retain full ownership of their unique derivative models, allowing them to create AI systems tailored to their specific legal, industrial, and cultural contexts.

A Global Coalition Takes Shape

What began last month with a technical workshop of 30 AI leaders from the US, France, India, Vietnam, and the UAE, is rapidly evolving into a global movement. The commitment from India, a nation with a stated mission for “self-reliant AI,” is a powerful endorsement. BharatGen, supported by IIT Bombay, will now co-lead workstreams in distributed model training, lending significant weight and expertise to the consortium.

“India has long believed that science advances fastest when it is open and shared,” noted Professor Ajay Sood, Principal Scientific Adviser to the Government of India. The project’s model, he added, “reflects the collaborative, self-reliant innovation India is advancing.”

This sentiment is echoed in Europe, where French President Emmanuel Macron has made AI capability a cornerstone of his G7 presidency. The heavy involvement of French organizations like Pleias, PRAIRE, and AMI Labs underscores a continental ambition to chart a course independent of American and Chinese tech dominance. The AI Alliance, a global non-profit, serves as the neutral ground for this collaboration, providing the governance and technical backbone to hold the ambitious project together.

“The knowledge and intelligence of humanity that is used to build AI does not live in any one organization,” stated Dr. Anthony Annunziata, Chairman of the AI Alliance. “The most capable and trusted AI can’t either.”

From New Delhi to Hanoi: Crafting Culturally-Native AI

Perhaps the most compelling promise of Project Tapestry lies in its potential to create AI that is not just multilingual, but truly multicultural. For participating nations, this is an opportunity to leapfrog the status quo and build technology that reflects their unique identity.

Vietnam has emerged as a particularly clear use case. As explained by Dr. Christopher Nguyen, the project's Chief Architect, “Aligning an AI model to a culture requires much more than language translation.” In collaboration with partners like the National Innovation Center and FPT Corporation, the project is already demonstrating how to build models that respect and understand local customs, preferences, and social practices. This is the difference between an AI that can speak Vietnamese and an AI that can think in a way that is relevant to Vietnam's society and industry.

For India, the stakes are similar but on a vastly different scale. With its immense linguistic diversity and burgeoning digital economy, the ability to build sovereign AI models is a strategic imperative. “As India builds frontier AI rooted in its own languages and knowledge, IIT Bombay and BharatGen...are proud to support and participate,” said Shireesh Kedare, Director of IIT Bombay and Chairman of BharatGen.

The Consortium's Gambit

How can a distributed, non-profit consortium hope to compete with the centralized, trillion-dollar machinery of big tech? The answer lies in a different kind of competitive advantage. The next breakthroughs in AI will not come from scraping more of the open web; they will come from high-quality, domain-specific data currently locked away in national archives, scientific databases, and industrial systems.

This is data that nations and corporations cannot—and should not—hand over to a centralized third party. Project Tapestry’s federated architecture is the key that unlocks this vast, untapped reservoir of human knowledge. By allowing participants to contribute to a collective intelligence without sacrificing control, it creates a data advantage that no single company can replicate.

Of course, the path ahead is fraught with challenges. The technical complexities of harmonizing model updates from heterogeneous sources are immense. The logistics of coordinating compute, funding, and talent across dozens of countries are daunting. And the political risks, from intellectual property disputes to the ever-present friction of geopolitics, are very real.

Project Tapestry is a grand, fragile experiment in technological federalism. It is an attempt to weave together the fraying fabric of global innovation with threads of mutual interest and architectural trust. Whether this intricate tapestry can hold in a world pulling itself apart remains the critical question.