This article is the second installment in our four-part series, "The Rise and Impact of Open-Source AI Initiatives." Through this series, we explore how open-source AI is reshaping the global landscape—unlocking access, accelerating innovation, and redefining strategic dynamics for businesses and governments alike.

In the first article – Democratization and Acceleration - how open AI models are transforming access and speeding up development – we examined how open AI models are expanding access and speeding up development cycles across industries. In this second part – Geopolitics and the China Factor – we turn our attention to China’s growing role in the AI race. By diving into the story of DeepSeek, we challenge common narratives and explore the nuanced reality behind China’s AI ecosystem, its ambitions, and its position in the global open-source movement.

‍

Part 2 of 4

The Geopolitical AI Race: Hype, Fear, and Competition

AI has become a theater of geopolitical competition, often framed as a U.S.–China “AI race.” The rise of DeepSeek and other Chinese AI efforts has intensified this discourse, raising questions about ideological differences and real capabilities.

On one hand, Western observers have viewed China’s AI developments with a mix of awe and concern. The “Sputnik moment” narrative – invoked after DeepSeek’s R1 matched an American model – suggests a fear of being outpaced​ Political figures have sounded alarms (e.g., a former U.S. president calling DeepSeek a “wake-up call” theguardian.com], and there are suspicions: is a model like DeepSeek’s truly a civilian effort, or does it mask state or military involvement? The U.S. government’s export controls (ban on advanced chips to China) were motivated by concerns of Chinese AI being used for military or authoritarian purposes​. In that context, some ideologues jumped to label DeepSeek’s breakthrough as a failure of export controls or even speculate if the Chinese obtained secret sauce from Western labs (an espionage narrative)​ [lawfaremedia.org]. Moreover, because China mandates censorship and alignment with “core socialist values” for AI systems deployed domestically, Western critics worry Chinese models might inherently carry propaganda or biases aligned with the government. This leads to an ideological caricature: that any Chinese AI will be a tool of state control or a trojan horse globally.

A More Nuanced Technical and Market Reality

However, the grounded technical and market reality provides a more nuanced picture. DeepSeek is a private initiative by a maverick entrepreneur, not a state-owned enterprise or military lab. Liang Wenfeng explicitly designed it as a research outfit with an ethos closer to open science than propaganda​ theguardian.com. Its funding came from a hedge fund’s profits, not government grants​ theguardian.com. And its aim – evidenced by open publishing and open-sourcing – is to push technology forward, not to embed ideology (if anything, the decision to open-source might conflict with a government’s desire to control AI). DeepSeek’s model, when inspected by outsiders, did not show obvious signs of censorship in the raw weights; any ideological filters for China would be a layer on top when deployed locally. In fact, by open-sourcing, DeepSeek gave up control over how the model is used or fine-tuned – anyone can adapt it for any ideology or purpose. This runs contrary to the notion of it as a pure propaganda tool.

‍

Regulations and Market Adaptations

That said, China’s government does play a role in the AI ecosystem: setting regulations and fostering an environment for AI growth. Beijing in 2023 introduced rules requiring generative AI to avoid prohibited content. So, Chinese companies do build compliant versions (e.g. Baidu’s ERNIE chatbot filters politically sensitive queries). But these are market adaptations similar to how Western models avoid certain content due to ethical guidelines – the difference is the definition of “sensitive” content. For global competitiveness, Chinese AI firms know that to attract international users, their models must not be overtly biased or censored. DeepSeek’s open release can be seen as an attempt to gain global credibility (it’s hard to accuse an openly inspectable model of hidden agenda if anyone can test it).

‍

The reality is that Chinese AI efforts are driven by technical and economic goals very much like those in the West. Chinese tech giants (Baidu, Alibaba, Tencent) and startups (like Zhipu, MiniMax, and DeepSeek) are racing to capture a booming domestic market for AI services and possibly export these services abroad. They compete on metrics like accuracy, speed, and cost – for example, Alibaba’s latest model Qwen 2.5-Max claims to outperform DeepSeek and Meta’s models on benchmarks​ [theguardian.com]. This competition is spurring rapid improvements. ByteDance’s aggressive pricing of Doubao at 9 ¥/million tokens was a business move to gain adoption​, analogous to a Silicon Valley startup undercutting a competitor. In short, Chinese companies are acting as competitors first, ideological actors second. As Matt Sheehan (Carnegie Endowment fellow) pointed out, focusing only on DeepSeek or trying to “crush” it would be misguided, because many Chinese players are innovating in parallel. The AI boom in China has a momentum that goes beyond any one model or company.

Innovation Despite Restrictions

From a geopolitical lens, one impact of DeepSeek’s rise is that it undermines the narrative that export controls can decisively stall China’s progress. Despite restricted access to top-end Nvidia chips, DeepSeek trained R1 on alternatives and clever algorithms. It highlights that innovation finds a way – either through more efficient methods or gray-market hardware acquisitions. This suggests the race in AI is not simply who has the best hardware, but who has the best ideas and talent. And talent is globally distributed. The U.S. still leads in many areas (fundamental research, chip design, overall number of top-tier AI experts), but China’s pool of engineers and researchers is massive and growing, often educated at the same institutions or using the same open research. Therefore, the ideological framing of a monolithic East vs West AI war is less useful than a practical view of a highly interconnected competition.

‍

One ideological difference does persist: the openness vs control approach. U.S. companies initially were more open (OpenAI published papers/models early on) and Chinese labs more closed. Now there’s a bit of a role reversal – OpenAI/Anthropic have gone relatively closed, whereas a Chinese startup (DeepSeek) open-sourced a leading model. This is somewhat ironic and complicates ideological narratives. It also might influence policy: Western regulators contemplating restrictions might note that if they clamp down too hard on open source AI, innovation could simply shift elsewhere (possibly to China, if it embraces open models more). Indeed, China’s tech community historically benefited from open-source (China was a major user of open-source software); now they are actively contributing to open-source AI, which could endear Chinese models to the global open-source community, softening geopolitical biases in some circles.

‍

In summary, the geopolitical discourse around China’s AI needs to separate fear-driven ideology from the facts on the ground. Yes, China’s government will ensure AI aligns with its laws domestically, and yes, Chinese AI companies are now formidable competitors. However, DeepSeek’s story is one of independent innovation and the pursuit of AI advancement, more than any grand political scheme. As Lawfare concluded, DeepSeek R1 is “not the omen of American decline” some feared, but it does herald a new era where AI breakthroughs can come from anywhere and spread everywhere quickly​ (Ball 2025) The grounded reality is that we have a more multipolar AI world – and that might actually reduce the chance of any single ideology dominating the technology.

‍

The DeepSeek story serves as a reminder that breakthrough innovation can emerge from anywhere—and in today’s increasingly multipolar AI landscape, it's essential to distinguish between fear-driven ideology and grounded technical reality.

At Xantage, we help organizations make sense of these shifts by providing strategic guidance on how to integrate AI effectively, identify real opportunities, and turn uncertainty into competitive advantage. Reach out to us to learn how we can support your business in navigating the evolving world of artificial intelligence.

‍

👉 Follow us to stay updated as we continue this four-part series unpacking the future of AI—one insight at a time.

👉 Follow us on LinkedIk and subscribe to our newsletter Forward-Thinking Enterprises