How China is Poised to Dethrone the U.S. AI Monopoly

Deepseek on the rise © Dia Radu

The unveiling of Deepseek, a highly anticipated Artificial Intelligence (AI) platform by a Chinese start-up, has sent ripples across Silicon Valley and Wall Street. The launch has not only disrupted the U.S.- dominated market, sparking a fierce debate about the future of AI, but also captivated everyone from industry experts to everyday consumers. But why is its arrival causing such unease?

Deepseek’s Energy Advantage

For years, the AI market, especially in the generative space, has been controlled by a handful of players, with Open AI’s ChatGPT at the top. OpenAI quickly became the gold standard for conversational AI, establishing a near monopoly that spanned businesses and consumers alike.

However, this dominance has come with significant cost: ChatGPT’s vast computational demands, requiring thousands of powerful servers across data centers, have drawn criticism over high operational expenses. Despite efforts to cut down its environmental impact, AI’s carbon footprint remains a growing concern, as many data centers still rely on fossil fuels.

While training ChatGPT consumes enough energy to power thousands of homes for months, Deepseek’s engineers have optimized their algorithms, achieving similar results with a fraction of the energy.

 “The real game changer seems to be the reduction in resource consumption,” Mihai Rusu, a professor at the University of Sibiu, explains for The Glass Room. “But there’s still no clear and transparent data on its actual energy usage. Let’s not forget this is a symbolic competition with Chinese technology claiming to surpass the American one.”

Yet, the stakes are high, and investors have started to panic. Unlike OpenAI, Deepseek’s technology doesn’t rely on Nvidia’s expensive hardware for training. The result? Just two days after Deepseek’s launch, Nvidia, Open AI’s main chip supplier, saw its market value drop by $593 billions.

The Hidden Costs

One of the major concerns regarding Deepseek is data privacy. While ChatGPT adheres to international standards, like General Data Protection Regulation, and has pledged not to sell personal data to third parties, models like Deepseek pose greater risks, as they are subject to the Cybersecurity Law, that gives government the power to access data under certain conditions or even censor some inquiries (as seen in The Guardian’s experiment.)

“Just look at what happened with Facebook and Cambridge Analytica to understand that nowhere is truly safe, not even in the West,” Stefan Baghiu, a researcher on platform capitalism, tells The Glass Room. “When ChatGPT first launched there were no clear agreements regarding data usage. OpenAI still retains user conversations to improve its model. We’re not in an unregulated space, but one where deregulation seems to be the ultimate goal.”

The final blow came on Wednesday, two days after the launching, when cloud security firm Wiz revealed in a report, that DeepSeek had inadvertently exposed a critical database online, leaving sensitive data vulnerable to anyone who stumbled upon it. The exposed database contained over a million records, including system logs, user prompt submissions, and authentication tokens. Despite repeated attempts by journalists to reach out for comment, DeepSeek has yet to respond to the allegations.

AI for the Masses

On the upside, Deepseek has lowered the barrier to entry for AI development. With less reliance on specialized hardware and lower maintenance costs, the platform has paved the way for other startups to enter the AI arena. In fact, just one day after Deepseek’s announcement, Alibaba released Qwen 2.5- VL, another Chinese AI model, further democratizing access to these powerful tools.

“There will always be cultural sensitivities and geopolitical stakes, which is why people can be cautious about Chinese tech,” says Rusu. “But we can’t ignore the obvious benefit: democratization. The new Chinese AI has an incredibly low access cost, making it a tool for everyone.”

Being fully open-source means that anyone can build their models on top of Deepseek. “And that is great for society,” says Bart Brouwers, a professor at University of Groningen. “But it's also extremely dangerous. We shouldn't expect much from American Big Tech when it comes to ethics and legal frameworks, but at least there are some governance structures in place. It terrifies me to imagine what could happen, especially if these models become more intelligent than us humans.”

A brief comparison © Dia Radu

In the midst of the new Cold War between superpowers, the recent rise of Deepseek signals a shift with far-reaching consequences. “The real issue isn’t whether Deepseek is better than ChatGPT - the fact that they are so similar is enough”, says Baghiu. “What’s truly at stake is how Deepseek undermines an entire system of AI management, one built on pouring massive amounts of money into corporations. Tech companies in the U.S. rely heavily on investor support, often inflating their market value. OpenAI grew this way, too. Now China is saying, ‘Fine, throw your money at it! We’ll give you something 1,000 times cheaper. It doesn’t have to be the best – it just has to blow your market.”