Open Source: The key to achieve Europe’s digital sovereignty in the AI era

Hi, it's Abel from Red River West, welcome to this first edition of Commit Pulse: our new newsletter about Open Source, and infra software. Today, I'll explain why we believe Open Source is the only lever for a competitive Europe in the AI Era 👇

Two weeks ago, DeepSeek made headlines by launching an Open Source AI model that rivals OpenAI's top-performing models, at just a fraction of the cost (27x cheaper, according to some sources). Earlier this week, Mistral (a French company) launched “Le Chat,” a competitor to ChatGPT based on their own Open Source* models that match the performance of OpenAI while running >10x faster. These breakthroughs align with the predictions we made in our previous article on Open Source AI: Open Source is winning the AI race.

By embracing an Open Source approach, organizations can compete with tech giants despite having fewer resources, leveraging the collective contributions, research, and expertise of thousands worldwide.

Here's where things get interesting for us Europeans. While we're watching from the sidelines as American and Chinese tech giants have dominated the tech race for years, we're sitting on an untapped goldmine: one of the world's most vibrant Open Source ecosystems 😃

Think about it: in an era where digital sovereignty equals geopolitical power, can Europe afford to remain dependent on proprietary software and AI models controlled by foreign corporations? The answer is a resounding no.

This isn't just about catching up; it's about jumping ahead by playing to our strengths. With Europe's rich tradition of collaboration, world-class academic institutions, and emerging tech leaders like Mistral and Hugging Face, we are uniquely positioned to drive an Open Source AI revolution.

So in this article, I’ll try to give you a clearer picture of why, at Red River West, we believe that betting on Open Source is the only way for Europe to be competitive in the AI race. Are we really behind? Why does Open Source strengthen European sovereignty? What elements do Europeans have to lead the Open Source AI race?

We are behind and lack the resources to face tomorrow’s AI revolutions

AI is revolutionizing industries worldwide. If we want to maintain some level of sovereignty against the U.S. and China while benefiting from the upcoming economic revolution, Europe must become a key player. However, we face significant resource constraints:

• Among the top 10 companies investing the most in R&D annually, only one is European (Volkswagen).

• Except for Spotify, no European company founded in the last 50 years has exceeded a valuation of $100 billion, while all major tech giants were created during this period (see the graph below).

• No cloud provider comes close to matching even 10% of the infrastructure capacity of the major tech giants.

A few weeks ago, the US started what we could call an “AI investment race” by pledging to invest $500 billion over the next four years in building new AI infrastructure in the United States. France and the European Union quickly followed this week announcing respectively a 109B$ and a 200B$ investment during Paris AI Summit.

We are going in the right direction, and a few weeks ago nobody would have thought that Europe would do that. So let’s not be too pessimistic!

But we need to be smarter because, in raw financial power, we simply don’t have what it takes to win this battle against the US and China (e.g. Sam Altman talked about a 5,000B$ investment for the US in the coming years).

The stakes have never been higher, whoever controls AI will shape the future. To prevent a few corporations in the U.S. or China from dominating the field, Europeans need to have models at their disposal that rival the ones created by these companies. But fear not, it's exactly what Open Source offers, and Europeans are damn pretty good at it 🔥

In history, open source always win

Even if you're joining the game late, adopting an Open Source strategy and utilizing Open Source technologies can still lead to success. History has repeatedly shown this to be true throughout the evolution of the Internet:

• Operating Systems: In the 1990s and 2000s, proprietary solutions like Windows and UNIX (AIX by IBM, HP-UX) dominated server operating systems. Today, Linux and Apache rule the market, with Linux powering 90% of cloud servers.

• Databases: In the 2000s, Oracle and Microsoft held a near-monopoly on databases. The emergence of Open Source solutions like MySQL and PostgreSQL ended that monopoly. Today, the top four most popular databases are Open Source.

• Smartphones: When Google launched Android (an Open Source OS), it was late to the smartphone revolution, dominated by players like Blackberry with RIM or Symbian (acquired by Nokia). Then, Windows Phone launched, but it never caught up with Android, and today, Android powers 70% of smartphones globally.

• Web Browsers: In the 2000s, Internet Explorer had over 90% market share. In 2008, Google launched Chrome, based on the Open Source Chromium project, which now holds over 65% of the market, while Internet Explorer is extinct.

And on and on ... So, as Europe is falling behind, why not use the only strategy that always worked through history and bet on Open Source?

Open SourceStrengthens European Sovereignty

“Europe has missed the boat on the Web, social networks and the cloud. But if we want to catch up in Europe, we can't do it without open source.” Jean-Baptiste Kempf (VLC founder)

Open Source brings significant advantages that promote European sovereignty:

• Technological Independence and Infrastructure Control: Open Source software empowers governments, businesses, and institutions to reduce reliance on proprietary technologies from the U.S. or China. By adopting Open Source solutions, they can retain full control over critical infrastructure, free from foreign regulations like the U.S. Cloud Act. Additionally, Open Source code is fully auditable, customizable, and can be hosted on the infrastructure of their choosing, ensuring greater security and autonomy (reducing drastically vendor locking).

• Security, Transparency, and Regulatory Compliance: Unlike proprietary software, Open Source solutions are fully auditable by the community, reducing risks of hidden vulnerabilities or backdoors. They also facilitate compliance with European data protection laws, ensuring sovereignty over data storage and usage. Notably, the EU’s AI Act exempts Open Source organizations from restrictions imposed on other companies.

• Boosting the European Technological and Economic Ecosystem: Open Source fosters the development of a local technological ecosystem by stimulating innovation and collaboration between startups, large enterprises, public institutions, and universities.

Investing in Open Source has a significant economic impact. A 2020 report from the European Commission revealed that in 2018, Open Source contributed between €65 billion and €95 billion to the European economy, despite an investment of just €1 billion. Additionally, a 10% increase in Open Source contributions could raise the EU’s GDP by 0.4% to 0.6% !!! 😮

Open Source Is Winning the AI Race

As we explained in our previous article, the Open Source path provides numerous advantages in the AI race:

• Transparency and Bias Reduction: Open Source allows a broad community to review and improve models, helping to identify and correct biases, making AI fairer and more reliable.

• Lower cost and better efficiency: Training and development expenses can be shared across a global network of contributors and users. By utilizing established Open Source technologies, you can reduce development time by years and save billions on GPU costs.

• Faster Innovation: A large contributor base accelerates advancements, reducing barriers to experimentation and optimization.

• Data Privacy and Control: Companies can host their models in-house without relying on external providers, ensuring greater data security and sovereignty.

• Customization and Flexibility: Unlike proprietary solutions, Open Source AI is not a “black box”, everything can be modified and optimized for specific needs.

AI has been built on Open Source collaboration since its very inception. Alan Turing’s groundbreaking 1950 paper Computing Machinery and Intelligence laid the theoretical foundation, and from there, global cooperation and shared research have propelled the field forward. Today, the most widely used AI frameworks including Scikit-Learn (developed by INRIA, with over 2.2 billion downloads), TensorFlow (by Google), PyTorch (by Meta), Keras, and Pandas are all Open Source, enabling researchers and developers worldwide to build on each other’s progress. This spirit of openness has driven some of AI’s most significant breakthroughs, such as Google’s BERT, which ignited the transformer revolution and paved the way for models like ChatGPT.

2023, a memo from Google engineers leaked predicting what is now happening "We Have No Moat, And Neither Does OpenAI [...] Open Source models are faster, more customizable, more private, and pound-for-pound more capable."

And now, this prediction has become a reality because Open Source models are surpassing every proprietary model.

Of course, we can talk about DeepSeek, which released a model that surpasses OpenAI’s best reasoning models on certain benchmarks while being 27 times cheaper. It surprised the world and was the first unconditional proof that Open Source is winning.

But Europe is not far behind, as Mistral (a French unicorn building Open Source* AI models) just launched a new version of their ChatGPT competitor, "Le Chat," which produces results as good as the best AI assistants from OpenAI, Anthropic, and Google at a much faster pace (see benchmark below). Le Chat is today the fastest AI assistant in the world, working at almost 20 times the pace of ChatGPT at a cost 10 times lower.

Europe has the elements to lead in Open Source but lacks funding

“Open Source AI can help European organizations make the most of this new technology by leveling the playing field... Europe is particularly well placed to make the most of this Open Source AI wave.” Mark Zuckerberg (CEO Meta) and Daniel Ek (CEO Spotify)

Europeans have produced some of the biggest Open Source success, especially in the AI field

Over half of the top 15 highest-valued Open Source companies have at least one European founder (Elastic, Mistral, Odoo, Sonar, Confluent, Hugging Face, Gitlab, Aiven, Suse).

50% of the 10 fastest-growing Open Source companies in the world in 2024 are Europeans according to our data platform Ramp which takes into account 10+ data points such as Github stars growth, Hugging Face downloads growth, Website visits growth, Docker / NPM / Pipy download growth, etc.

Europe, and France in particular is leading the Open Source AI race

"France is geared up to become the Open Source AI Capital. The French have long been big into Open Source (VLC, etc.)” Michael Jackson (Entrepreneur & VC)

"Is France the Next Open Source AI Capital ? Mais oui !" Yann LeCun (chief AI scientist at Meta)

We have all the elements to lead the Open Source AI race:

1. The most downloaded AI framework in the world ahead of PyTorch and Tensorflow with > 2B downloads is scikit-learn, a framework created by French researchers at INRIA, and now amplified by Probabl.ai, and it's Open-Source.

2. Hugging Face (a French company) is THE place where people and companies across the world share and download the best AI models. This French unicorn is becoming the Github of AI, and the trusted party that ranks and evaluates the best models. DeepSeek, Meta, OpenAI, Google, etc. All these companies rely on Hugging Face to distribute their best Open Source models.

2. Most of the big tech companies have well-known AI research labs in Paris such as Google with Deepmind, Microsoft, and Meta. The Facebook AI Research labs located in Paris for example are recognized for their contributions to both foundational research and applied AI. As a result, LLaMA, one of the leading Open Source* AI models was developed by Meta in Paris by mostly French nationals.

3. Mistral, one of the first leading Generative AI companies to take the Open Source* path was founded by French nationals, in Paris. Mistral is now valued at 6.2B$ with investors such as General Catalyst, a16z, Lightspeed, etc. And last week, they showed the world that Europeans could still produce better technologies than their American counterparts by releasing multiple great models, as well as their first mobile application.

2. Open Source AI companies from all over Europe dominate their verticals, such as Weaviate (Netherlands) and Qdrant (Germany), which are two of the leading vector databases, or Stability.ai (UK) and Black Forest Labs (Germany), which produced two of the leading image generation models or lead the way in Open Source AI for biology.

3. New leaders are emerging fast (like Dust, which raised >20M€ with Sequoia, or H, which raised >200M$ with Accel and others), and the number of Open Source companies created in Europe is growing exponentially. In France, we went from 10 Open Source early-stage rounds per year to >30 in the last six years. Great new companies are created every week, and as the year 2025 unfolds, we will see more incredible Open-Source technology emerge from European labs.

European infrastructure, education, and regulatory environment favor Open Source

European academic institutions have long been pioneers in open research and collaborative innovation. From the CERN's world-changing decision to make the World Wide Web freely available to everyone, to today's AI research labs at institutions like INRIA, ETH Zürich, Polytechnique, and KTH Royal Institute of Technology, Europe's educational ecosystem naturally aligns with Open Source principles. We have some of the best universities that produce some of the best AI researchers in the world, and they all mostly favor Open Source. We must play on that strength.

Regarding regulation, it kind of became a joke in the tech world: while Americans are the first to find innovative ways to build technology, Europeans are the first to find innovative ways to regulate it.

And let's face it, we Europeans love to regulate, and maybe a bit too much for innovation to strive. From cookie policies to AI ethics frameworks, we've never met a technology we didn't want to regulate. But there is a plot twist: these same regulations that make life challenging for traditional tech companies could create an unexpected tailwind for Open Source development.

Most of the regulations put in place by Europe demand transparency, accountability, and data protection things Open Source does naturally:

• Transparency: Proprietary companies struggle to balance openness and competitive advantage, while Open Source is transparent by design.

• Accountability: Public scrutiny ensures compliance is built into development.

• Data Protection: Open Source allows local deployment and data sovereignty which is ideal to comply with GDPR.

As regulations expand, and Open Source keeps being exempted from those (e.g. the AI act doesn't apply to Open Source companies) Open Source's cost advantage grows because compliance is built in.

I'm not saying that more regulation is good particularly good for Open Source. For example, the Cyber Resilience Act could unintentionally arm Open Source companies in Europe. But in general, it makes life harder for Closed-Source companies than for Open-Source ones, which favors Open-Source development and focus in Europe.

But European Open Source companies face a significant funding gap

“VCs [in Europe] are more sceptical and often do not recognise the future potential of a COSS model,” Andre Zayarni co-founder of Qdrant

“Unfortunately, few French and European investors understand this model, so we have to seek funds elsewhere.” Anh-Tho Chuong co-founder of Lago

After speaking with over 100 European Open Source entrepreneurs in recent months, a clear pattern has emerged: most European VCs do not understand Open Source. Even in firms that do, it’s usually just one or two individuals with the necessary expertise. This lack of understanding causes a significant funding gap: European VCs not only have fewer funds to deploy compared to their American counterparts, but they also tend to prioritize other sectors over Open Sourcee.

As a result, European Open Source startups are overwhelmingly funded by U.S. investors. Over 80% of seed investors in European Open Source unicorns are American, compared to just 27% for other types of companies (source: Dealroom). This figure does not even account for the many Open Source startups that have relocated to the U.S. in search of funding and a more supportive ecosystem. On average, US-based Open Source startups receive 4.2 times more funding than their European counterparts (source: Dealroom), further widening the gap and reinforcing the trend of European innovation being financed, and often owned, by foreign capital.

We need a new kind of actor with significant tech knowledge, a great understanding of Open Source business dynamics, and the capacity to support both financially and operationally Open Source business in Europe.

* small precision here, Misral and Llama are not completely Open Source as I explained in this article.

Conclusion

Of course, Open Source alone won't guarantee Europe's sovereignty. To achieve that, Europe must also secure efficient access to space, continue investing in AI infrastructure, and significantly boost its investments in energy (especially nuclear power).

That said, Open Source will serve as the backbone of future digital and AI advancements. It is the only way to maintain control over our data, and without it, we don’t have the ability to compete with Big Tech. Europe has a rare opportunity to leverage its strengths in Open Source and establish itself as a key player. So, let's turn our current challenges into our greatest advantage.

The future of tech is open, and Europe can lead the way.

Don’t hesitate to reach out to me at abel@redriverwest.com or Olivier at olivier@redriverwest.com if you want to discuss :)