This is the second of two related articles. Readers can access the first article here.
There was a time when open source and decentralisation looked like the answer to the concentration of power and money in the hands of Big Tech, whether in networks or the wider web; a coexistence model where free, community-built software sat alongside commercial products, good enough to keep the market honest. For Europe this was a lifeline compared to the amounts of money being spent in China and the USA on technologies which could – and have – compromised European technology sovereignty and the security of its inhabitants.
“It was going well, right? It became more and more acceptable, and it was a proper coexistence model,” Oleg Volpin told TelcoForge, speaking in a personal capacity rather than in his role as Group President for Amdocs International.
“And then something happened; something happened that meant they cannot compete any more.”
What happened, in his account, is a change in the size of the opponent, not the quality of the alternative.
Whale Hunting
Volpin’s argument rests on an observation about how recently corporate wealth reached its current scale.
“We never had any person in the world that had a trillion dollars, right?” he said. “I think it was Microsoft [that] became the first trillion-dollar company. I forget when, but it’s not 20 years ago. And suddenly you have, what, eight or ten trillion-dollar companies? So this gap between normal people and the people or companies with literally unlimited resources became wider and wider.”
The first company to cross a $1 trillion market capitalisation was Apple, in August 2018, with Microsoft following roughly a year later. And Volpin’s estimate of “eight or ten” trillion-dollar companies is on the low side. By mid-2026 there were 17 companies worth over $1 trillion, with the five largest US technology companies alone worth a combined sum in excess of $20 trillion. Nvidia recently became the first company in history to cross $5 trillion on its own. Whatever else has changed in the seven years since Apple’s milestone, the pattern Volpin describes – a small number of companies commanding resources that dwarf entire national economies – has only intensified.
That matters for open source specifically, Volpin argues, because free software’s original competitive advantage was never really about licensing philosophy.
“In the past, those free software projects competed with people who were very rich, but way less rich than today,” he said.
“In the past, the majority of those organisations relied on the goodwill of people, and that was good enough. I don’t think it can be any more. There’s just not enough critical mass.”
The asymmetry Volpin is describing is not hard to find in the public record, whether that’s investments in particular hyped stocks or in open versus closed business models. Microsoft has invested roughly €15 million for a stake of under 1% in French AI company Mistral, which offers a range of open-source and open-weight models; a stake that, however meaningful strategically, amounts to a rounding error against the roughly $13 billion Microsoft has committed to OpenAI over the same period.
His conclusion is not that individuals or small teams can no longer contribute to open technology. On the contrary, it’s absolutely necessary. However, this bottom-up activity isn’t sufficient on its own to fund programmes or infrastructure at the scale modern AI now requires.
“If you’re talking about Europe, you need a pan-European fund: somebody could apply for it, get money, and use it for this,” he said. “They can’t do it on their own. The people cannot do it on their own.”
That’s a difficult call, but one governments have already started answering in other technology areas by providing various funding streams. This would be, arguably, something on a different scale but recognisable to the bodies involved.
More Haste, Less Work
With all this, though, is there actually scope for Europe to develop sovereign digital capabilities or services? Volpin’s answer is nuanced – “yes, if you stop trying to build everything.”
“I think it’s feasible,” he said of sovereign AI, drawing an analogy to an early piece of cloud testing Amdocs carried out.
“We worked with a small cloud provider in Austria, and for years we’d only worked with AWS, Microsoft, Google. We did a test on how our software behaved on this cloud, and actually the results were quite encouraging. It was probably doing only 20% of what AWS is doing. But for our needs at that time, this was enough.”
Volpin generalises the point beyond cloud infrastructure to GPTs and language models:
“You’re not going to use the LLM that can do everything in the world. It will be unique for your purpose. If that’s what you need, that’s what you need.”
He draws a comparison to his own habits as an AI user: “I use all of them, just to see how they are. But I’m not going to use the very latest model of Claude, right? I just don’t want to pay for it. I’m using an older version. I don’t see that big a difference for my personal use, if at all.”
It’s a small, human illustration of a genuinely important strategic idea: sovereignty doesn’t require matching the capability of the market leader, only exceeding the threshold your own use case actually needs.
In other words, it’s not a question of “can Europe out-build OpenAI or Anthropic?” which on current capital trends looks absurd any time soon, but “can Europe build systems good enough for its own regulated industries to depend on?”
(We should also point out here that speculation about how long companies like OpenAI and Anthropic can continue to make billions in losses is rife. Arguably they’re not good models for businesses countries want as strategic assets anyway).
Mistral, still the only European large language model developer operating at meaningful scale, and Germany’s Aleph Alpha, now focused more on enterprise and public-sector deployment than frontier competition, are the clearest existing evidence that the scoped-down version of the strategy is viable. Whether it is viable at the pace European industry actually needs is a separate question.
As we saw in the previous article, current European mechanisms for funding at scale work painfully slowly; it’s funding at the pace of consensus rather than, in the US financial markets, funding at the pace of hype. There does need to be another way, and we know from the COVID years that governments can move fast when they see a need.
Put together, Volpin’s position amounts to this:
- Open-source development cannot, on its own, close a capital gap that has widened by trillions of dollars in under a decade. But
- Narrowly scoped, purpose-built sovereign systems, such as cloud infrastructure at a fraction of hyperscaler capacity or language models tuned to specific regulated use cases, are genuinely achievable,
- Provided the money to build them is pooled at a scale no single European country can muster alone.
“Maybe Germany, France and the UK could do it on their own. Maybe,” he said.
“If you look at the size of the EU, it’s the same population as the US. You really have to have the entire EU looking into it, rather than doing it country by country.”
Image by Hans from Pixabay
