AI openness: From success to sovereignty

During a keynote at "Civo Navigate India," I shared new data that  India has now surpassed the US in number of new open source contributors and contributions to open source, taking the number one global position; and if that level of growth in India's contributor numbers continue to grow at its current pace, it will also be number one by contributor numbers globally by 2030 too. This update was greeted by a spontaneous round of applause.

The applause reminded me of being in Paris in January. Pre-"Global AI Action Summit", I keynoted an event where I was able to share that the Tortoise Index identified France as Europe's number one position for "open source AI".  It's nice to deliver good news, and people are so happy to hear it. But the reality of open source and AI today is much more complex.

Despite open source entering mainstream conversations, openness faces a grave challenge today and it's one that unfortunately very few people really understand, let alone applaud.

The success of open source software has come at a price…the pace of adoption hasn't been matched by a scale of understanding. Many who are part of the contributing community don't understand the licensing subtleties and requirements. They have engaged as a second generation, attracted to an innovation methodology, rather than a social movement that has underlying basic principles that must be met for its continued success.

Distilled to its most basic, open source licenses enable anyone to use the code for any purpose, subject to applicable laws. This is at the heart of its success, as it enables a free flow of innovation between contributors, community and users. Users can rely on that. They know that they may use, iteratively innovate and share.  

As we shift to a world of AI, whichever of the evolving approaches to categorising AI openness we follow – definition or disaggregation – the bottom line is that to consider AI to be "open source" it must also follow this approach. It must meet that same standard. That means in a business context, whether we are talking software or AI, open source requires that we enable our competitors with our innovation because anyone can use it for any purpose. That's the price we pay for the benefits of open source – which include building on others' work, iterative innovation, collaboration, pace of adoption, and ultimately the potential to become the de facto standard in a tech space.

I have been struck, talking to people in India from the open tech communities, just how parallel our universes are. Of course, the global south faces different challenges, in the scale of population and demands on  its infrastructure. That aside, the challenges are the same and surprisingly so are the conversations.  

Ultimately we drilled down on two areas, "Sovereignty" and "data access" with a healthcare slant. The Sovereignty conversation is made all the more relevant by our report being released in Bengaluru, on the same day as the "Summit for More Digital Sovereignty"  in Berlin. This resulted in headlines like "Europe wants to go it alone on tech." An ironic decision when Europe also released the Digital Omnibus seeking to simplify its overly convoluted and bureaucratic regulations to enable innovation. The two approaches feel like they contradict each other.

With somewhat varied thinking across the contributors to our report, the real debate has been about sovereignty – what do we mean by the word sovereignty in this context?  Is it local tech or localisation? Does it exclude the US and Big Tech or collaborate to localise? The EU appears to want to go it alone. 

Anyone talking about sovereignty will describe open source as its cornerstone. It is not possible to build a local stack or AI without open source. Even China's DeepSeek wouldn't have been able to create R1 had it not distilled other open LLMs. The public cloud of the hyperscalers is built on open source. The AI DeepSeek distilled was in fact American. Key to the sovereignty discussions, whether in Berlin or Bengaluru, is the simple fact that to build sovereignty in digital today – whether we are talking AI or software-based infrastructure – open source is the cornerstone of that possibility. I doubt many would argue this fact today.  

In this digitised world, governments are rightly concerned with dependency on tech from other countries. They need certainty that they can access  the tech on which national services depend today over the long term. The real issue isn't where the tech comes from, but whether it can be accessed and used without that international partner's permission. Open source has the power to enable that access and remove these underlying concerns.

Yet whilst open source can offer this support, if these solutions are to have longevity, the question then becomes how this software is built, implemented and maintained? Is it real open source or a 'sovereignty-driven' local source? Real open source benefits from building on existing contributions from global participants and these dependencies enable sharing and iterative innovation.

This international collaborative nature is critical to any long-term success of a sovereign activity. Bifurcation, or "going it alone", is doomed to fail. Some local companies will get rich along the way and a form of "local source" will be created. Few nation states could afford to fork, maintain and match the pace of innovation of the globally collaborative open source community. A pooling of resources saves costs and allows teams to build.

If we wish to see success in our approach to sovereignty, then it requires two things:

• Using open source on a globally collaborative basis to create those de facto standards accessible by all, with no bifurcation or local source.
• A non-isolationist approach to sovereignty, enabling nation states to bring their people the best global innovation, and to localise these solutions for what Nvidia's Jensen Hung calls "national intelligence" made up of a country's security, culture and language.

If the real concern in a conversation around sovereignty is that companies haven't scaled locally then this is the actual challenge we must tackle. A false boost in sovereign investment in local companies by governments will create short term wealth for a few, who are sadly encouraging this thinking. This local infrastructure will however not scale businesses for the future, nor will it deliver in its local source the best global technology for its people.

The true power of open source is its ability to enable global innovation that benefits all people. It has the power and potential to be a Digital Public Good available to all people which can be localised to enable sovereignty and meet the 3 Sutras of People, Planet and Power set out by the Global AI Impact Summit.

How businesses are built in a country is a totally different matter. That should not be allowed to undermine the globally collaborative power of open source, whether by the EU or anywhere else. Rather than looking at local source as a route to protecting markets for sovereignty reasons, open source should look to global markets supported by local opportunity and understanding.