Artificial intelligence is one scalea game that revolves around scale: more data, more computing power, more money to train AI models, more energy to power the data centers. That was, until this week, the broad -divided view of artificial intelligence (AI). The announcement last week by the new Trump government from Stargate, a huge investment in ‘super intelligent AI’, added that extra strength. With $ 500 billion and an unprecedented expansion of data and energy capacity, the US is fully committed to scale. But then suddenly there was deep key.
A new version of this Chinese AI model, strategically launched on the day of Trumps inauguration, appears to perform better on crucial indicators than the largest, most expensive models of American competitors such as OpenAi’s Chatgpt and Anthropic’s Claude. It led to a shock golf: never before did a company lose so much market value in one day as AI chip maker Nvidia, nearly $ 600 billion.
What this means in the long term for tech companies is uncertain. In any case, a scale remains a factor of interest in AI development. But Deepseek in any case shows that there are other ways to win the AI race: with much less computing power, with much fewer investments, with a Source-Approach instead of a closed approach such as (the very misleading) OpenAi. Efficiency, cleverness and cooperation instead of just scale.
The wrinkle effects of this can be considerable. For the monopoly power of tech giants such as Google and Meta for example. You have to invest enormously for building search engines and social networks. But once you have a search engine, every extra search will cost almost nothing. This, together with the fact that many web technologies work better if many people use them (Zogetehen network effects), ensures that there is often only one big winner left. Deepseek shows that it Dynamics do not necessarily apply to AI.
It is hoped that this will finally shake in Europe. Also the old continent, so far almost nowhere to be seen in the AI race, can still participate for the marbles. And there is a lot at stake with AI: the future of work, the speed of innovation and scientific breakthroughs, but also geopolitical influence. Deepseek offers hope that European companies also have a chance to participate seriously. But that must happen now.
In that respect, the new ‘competitive compass’ of the European Commission shows in any case that it recognizes the urgency of this issue. On that new European innovation race, announced on Wednesday by committee chairman Ursula von der Leyen, there is a lot to be said, but there are meaningful ideas and investment plans for accelerating innovation and reducing regulatory pressure. That raises pressing questions, not in the least about AI. In the US, the large AI investments are accompanied by the abolition of ethical guidelines. That route does not have to go on Europe.
If artificial intelligence is not only about scale, there is more room for various flavors AI. European artificial intelligence can then distinguish itself by building in ethical crash barriers from the first moment, for example, about responsible applications in the workplace, ethical data use, lower energy consumption, fairer distribution of yields, and more.
That’s not easy. The development of ethical European AI will in the coming period require an extremely precarious balancing exercise between on the one hand sufficient innovation and experimental space, and on the other hand the monitoring of social frameworks. Europe has no other choice than developing its own European taste AI that can handle the competition.
At the inauguration of President Trump, with tech chief executives at honor places, it was clearly visible how Big Tech will take on Trumps political agenda. American Super-AI does not have to be good news for democracy and personal freedom in Europe. No naivety fits about the Chinese taste of AI either. Deepseek, for example, refuses to answer questions about sensitive Chinese political topics.
Without solid own AI development, Europe makes itself completely dependent on the technology and the ethical compass of other power blocks. That is a huge, not to say existential, risk for the strategic autonomy and EU freedom. The sudden success of Deepseek shows that Europe can not only participate in the AI race, but also has to participate in securing a democratic future in which European values continue to count.