Open this photo in gallery:

DeepSeek had its roots in a hedge fund company called High-Flyer, created in 2015 by Liang Wenfang, a 40-year-old trained engineer.Dado Ruvic/Reuters

Owen Guo is a freelance writer in Toronto.

This week, a Chinese company took the tech world by storm. It wasn’t Alibaba, Baidu or TikTok. A tech start-up named DeepSeek stunned the world with its ability to create advanced AI models rivalling those developed by OpenAI and Google – at a fraction of the cost.

DeepSeek’s breakthrough is remarkable in many ways.

At its core, this is a story about an ambitious firm striving to reach new heights. But it also marks the first time a Chinese tech company has directly challenged Western technological supremacy in the midst of a global race for AI dominance.

Not only has DeepSeek managed to do impressive AI work with less computing power – and thus lower costs – it has matched or even surpassed its rivals on several benchmarks. On Monday, DeepSeek dethroned ChatGPT as the most-downloaded free app on the Apple Store.

Amid the frenzy, prominent American tech investor Marc Andreessen has called DeepSeek “AI’s Sputnik moment.” It was surely a moment that took many by surprise. In recent years, Washington has imposed export controls restricting the sale of advanced semiconductors to China, a move Beijing has protested against loudly.

Explainer: What is DeepSeek and why is it disrupting the AI sector?

How exactly did DeepSeek manage to overcome hardware bottlenecks to not only survive but also thrive?

Its foray into AI almost happened by accident. DeepSeek had its roots in a hedge fund company called High-Flyer, created in 2015 by Liang Wenfang, a 40-year-old trained engineer. At High-Flyer, Mr. Liang used AI to trade stocks – until Chinese financial regulators started to crack down on speculative trading.

And so High-Flyer became DeepSeek in 2023. It was then when Mr. Liang turned his attention to AI research.

He is as ambitious as he is nerdy. In an interview he gave in 2024, Mr. Liang was portrayed as a modest leader who spends his time reading papers, writing code and participating in group discussions.

In that interview, Mr. Liang spoke passionately about his deep curiosity and commitment to innovation and research. He also took a jab at big Chinese tech firms – flush with cash but deficient in innovation. Hamstrung by the lack of advanced chips, DeepSeek has turned to open-source methods, pioneering data analysis techniques and novel technical designs to make inroads.

“For many years, Chinese companies are used to others doing technological innovation, while we focused on application monetization,” Mr. Liang said in the interview. “Our starting point is not to take advantage of the opportunity to make a quick profit, but rather to reach the technical frontier and drive the development of the entire ecosystem.”

He added, “China should gradually become a contributor instead of freeriding.”

DeepSeek’s singular focus on research – and its quest for AI sophistication – has a powerful ally: the Chinese state. In recent years, Beijing has not only made science and technological innovation a key plank of its economic policy, it has acknowledged, in President Xi Jinping’s own words, that the competition among nations today is ultimately a competition over technological prowess.

Already, spending on research and development has tripled in the past decade in China, now second only to the United States, the world’s top spender in R&D. In response to calls for cultivating more AI talent, many Chinese universities are revamping their science and engineering programs. With the world’s largest pool of STEM graduates, China also leads globally in publishing widely cited papers in 57 of 64 critical technologies, based on tracking by the Australian Strategic Policy Institute.

DeepSeek is determined to capitalize on those gains. So far, it has only hired coders and engineers from top Chinese universities, dangling big salaries. And just this month, Mr. Liang met with the Chinese premier to discuss home-grown innovation.

For all the attention DeepSeek has garnered, some difficult questions remain. Can the company continue to perfect its AI models given that the Chinese government may keep certain information from the public? Put another way, is there a point at which DeepSeek’s professed drive to innovate runs afoul of government censorship?

Early signs suggest so. When asked about sensitive topics such as the Tiananmen Square massacre or Taiwan, DeepSeek declined to respond.

Nevertheless, the West has reacted with a mix of awe and consternation to DeepSeek. U.S. President Donald Trump described the DeepSeek moment as “a wake-up call” for American tech companies. Nvidia, whose stocks nosedived earlier this week, praised DeepSeek’s ingenuity. Meanwhile, Facebook has reportedly set up “war rooms” of engineers to figure out how a scrappy Chinese start-up managed to advance so rapidly – and so cheaply.

This is AI’s watershed moment. And it’s just the beginning.