In a geopolitical context characterized by growing technological tensions between East and West, China with Liang Wenfeng has launched DeepSeek R1, an open source artificial intelligence model developed with limited resources but capable of competing with the proprietary systems of OpenAI, Google and Meta. This milestone, achieved in just two months and with limited investments compared to Western standards, represents not only a technical advance, but a clever strategic move to redraw the balance of power in the critical sector of AI.
Beijing’s adoption of an open source model overturns the traditional paradigm of centralized control, typical of US corporations. The public accessibility of DeepSeek’s code allows states and organizations to study, modify and implement solutions without depending on foreign actors, a crucial factor for nations like Italy that aim to strengthen their digital autonomy. In an era in which data is a strategic weapon, this openness reduces the risk of exposure to foreign technological monopolies and facilitates the development of national critical infrastructures, aligning with the security objectives outlined by the European PNRR and the NATO doctrine on cyber-resilience.
DeepSeek shows that technological excellence does not necessarily require unlimited capital. With optimized training and limited hardware, the Chinese model challenges the Western narrative that only billion-dollar investments guarantee superiority in AI. This low-cost/high-yield approach could inspire US allies, such as European countries, to rethink their innovation strategies, privileging efficiency and transnational collaboration over hyper-competitive and fragmented models.
The launch of DeepSeek coincides with a moment of acute geopolitical friction, marked by the return of protectionist policies in the US and restrictions on the export of advanced semiconductors to China. Beijing, however, has circumvented these limitations by focusing on algorithmic optimization and an open-source ecosystem that mobilizes global resources. This model not only weakens the Western advantage based on closed technologies, but transforms AI into a tool of soft power, attracting developing countries looking for accessible alternatives to Western systems.
The recent cyber attacks against DeepSeek, which have blocked access to Western users, highlight a new front of competition. While there is no direct evidence, it is plausible that these attacks reflect sabotage attempts by entities with competing interests, in line with the hybrid warfare tactics increasingly common in cyberspace. DeepSeek’s resilience to these threats will be a test of the credibility of the Chinese model as a safe and reliable alternative.
The rivalry between open and closed models has its roots in the Cold War, when the Soviet bloc and NATO pursued divergent strategies in computing. In the 80s, Richard Stallman's free software was an ideological response to Microsoft's privatization of code. Today, DeepSeek re-enacts this clash, positioning China as the champion of "democratized" AI, in contrast to the Western oligopoly. However, behind the collaborative rhetoric, Beijing pursues clear objectives: reducing dependence on US technologies and consolidating its technological influence in strategic areas such as Africa and Southeast Asia.
DeepSeek’s success has already triggered a market earthquake: the collapse of Western Big Tech stocks reflects fears of a reduction in their dominance. For the United States and NATO allies, the challenge is twofold:
- Preserving technological leadership without falling back into an unsustainable economic arms race.
- Balancing security and openness, ensuring that export controls on AI do not stifle domestic innovation.
The Western response will have to be coordinated and multifactorial: strengthening public research centers, partnering with emerging companies, and reviewing cybersecurity policies to counter hybrid threats.
DeepSeek is not just an algorithmic model, but a symbol of the realignment of global technological power. China has shown that open source can be an effective weapon to erode Western primacy, combining economic pragmatism and geopolitical vision. For liberal democracies, the risk is not so much China's technical superiority, but its inability to adapt to a world in which AI is increasingly decentralized, accessible and weaponized. The game is open, but one lesson is already clear: in the age of artificial intelligence, the real power lies not in chips, but in the ability to shape the ecosystems that govern them.
