DeepSeek: How China's 'AI heroes' overcame US curbs to stun Silicon Valley
When ChatGPT made waves in the world of artificial intelligence (AI), it quickly sparked a pressing question: did it pose a threat to China, America’s top tech rival?
Fast forward two years, and a new AI model from China has turned that question on its head: can the US stifle Chinese innovation?
For a time, Beijing struggled to find a solid answer to ChatGPT, especially since the platform isn’t available in China.
Users initially ridiculed Ernie, the chatbot from Baidu, with many calling it underwhelming. Following that, versions from Tencent and ByteDance appeared, but they were quickly dismissed as imitators, not quite measuring up to ChatGPT.
Washington, meanwhile, remained confident it was ahead and worked to ensure it stayed that way by imposing restrictions on the export of advanced chips and technology to China.
That’s why the launch of DeepSeek has taken Silicon Valley—and the world—by surprise. The company claims its powerful model is far more affordable than the billions US firms have spent developing AI.
So how did this relatively unknown company, led by a founder now being celebrated as an "AI hero" in China, manage to pull off such an impressive feat?
DeepSeek: The Chinese AI App Making Waves Globally
China's DeepSeek AI is challenging the status quo and taking a serious bite out of America's dominance in AI.
The Challenge
When the U.S. imposed a ban on top chipmakers like Nvidia, preventing them from selling advanced technology to China, it dealt a significant blow to the country’s AI ambitions. These chips are vital for building sophisticated AI models capable of handling everything from simple queries to complex problem-solving.
DeepSeek’s founder, Liang Wenfeng, referred to this chip ban as the company’s “main challenge” in interviews with local media. However, DeepSeek had a head start: long before the ban, they had already stockpiled a "substantial" number of Nvidia A100 chips—estimates range from 10,000 to 50,000 units, according to the MIT Technology Review.
While leading AI models in the West typically require around 16,000 specialized chips, DeepSeek claims to have trained its AI model with just 2,000 A100 chips, supplemented with thousands of lower-grade chips. This strategy, they say, enables their product to be more affordable.
Some skeptics, including tech mogul Elon Musk, have questioned the accuracy of DeepSeek's claims, suggesting that the company may not be transparent about how many high-end chips it truly possesses given the restrictions.
Nevertheless, experts argue that the U.S. ban has provided both challenges and opportunities for the Chinese AI sector. According to Marina Zhang, an associate professor at the University of Technology Sydney, it has “forced Chinese companies like DeepSeek to innovate,” enabling them to do more with fewer resources.
While these restrictions present challenges, they've also driven innovation and resilience, aligning with China's broader goals of achieving technological independence.
As the world's second-largest economy, China has heavily invested in big tech, from the batteries powering electric vehicles and solar panels to artificial intelligence (AI).
President Xi Jinping's long-standing ambition has been to transform China into a tech superpower, and Washington's restrictions became a challenge that Beijing was determined to overcome.
The release of DeepSeek's new model on January 20, coinciding with Donald Trump's inauguration, was a strategic move, according to Gregory C. Allen, an AI expert at the Center for Strategic and International Studies.
"The timing and messaging are exactly what the Chinese government wants everyone to believe—that export controls are ineffective and that America is no longer the global leader in AI," says Mr. Allen, a former director of strategy and policy at the US Department of Defense Joint Artificial Intelligence Center.
In recent years, the Chinese government has actively fostered AI talent, providing scholarships, research grants, and encouraging collaborations between universities and industry.
Initiatives like the National Engineering Laboratory for Deep Learning have trained thousands of AI specialists, according to Ms. Zhang.
And with an abundance of brilliant engineers available, China had plenty of talent to recruit.
The talent
Take DeepSeek's team for instance - Chinese media says it comprises fewer than 140 people, most of whom are what the internet has proudly declared as "home-grown talent" from elite Chinese universities.
Western observers missed the emergence of "a new generation of entrepreneurs who prioritise foundational research and long-term technological advancement over quick profits", Ms Zhang says.
China's top universities are creating a "rapidly growing AI talent pool" where even managers are often under the age of 35.
"Having grown up during China's rapid technological ascent, they are deeply motivated by a drive for self-reliance in innovation," she adds.
Liang Wenfeng, the 40-year-old founder of DeepSeek, exemplifies the new wave of tech innovators. A graduate in AI from Zhejiang University, he's described by those close to him as more of a "geek" than a traditional boss. Chinese media have labeled him a "technical idealist" for his commitment to maintaining DeepSeek as an open-source platform, a choice experts believe has fostered a thriving culture that enables young startups to pool resources and innovate faster.
Unlike larger Chinese tech companies, DeepSeek has focused heavily on research, allowing more room for experimentation. "The top 50 talents in this field might not be in China, but we can build people like that here," Liang said in an interview with 36Kr.
However, there are concerns about DeepSeek's long-term potential. Experts caution that new US restrictions could limit access to American user data, potentially hindering DeepSeek’s global expansion. Furthermore, the US still holds a significant edge, especially in terms of vast computing resources, and questions remain about how DeepSeek will sustain its progress with access to advanced chips.
For now, DeepSeek is riding high, with many in China only becoming aware of the company recently.Mr. Liang's sudden rise to fame has made him a sensation on China's social media, where he’s being hailed as one of the "three AI heroes" from southern Guangdong province, which borders Hong Kong.
The other two honorees are Zhilin Yang, a leading expert at Tsinghua University, and Kaiming He, a professor at MIT in the US.
DeepSeek has captured the attention of the Chinese internet, just ahead of Lunar New Year, the country's biggest holiday. This news is a bright spot for an economy facing challenges, especially with ongoing tariff concerns and the potential sale of TikTok's US operations.
"DeepSeek shows us that only the real deal can stand the test of time," reads one popular comment on Weibo.
Another comment says, "This is the best New Year gift. May our motherland prosper and grow strong."
Wei Sun, principal AI analyst at Counterpoint Research, described the reaction within China’s open-source community as a mix of "shock and excitement."Fiona Zhou, a tech worker based in Shenzhen, noticed her social media feed was suddenly flooded with posts about DeepSeek yesterday.
"People are calling it 'the glory of made-in-China' and saying it shocked Silicon Valley, so I decided to try it out and see how good it is," she said.
She asked DeepSeek to provide her "four pillars of destiny" (ba-zi), a personalized horoscope based on the date and time of birth. However, she was disappointed when the results were incorrect. Although the app offered a detailed explanation of its "thinking process," it didn’t align with her actual ba-zi.
Despite this, Fiona plans to give DeepSeek another try at work, believing it might be more useful for professional tasks.
Comments
Post a Comment