Deepsek: How did China’s ‘AI Heroes’ make the US a Stun Silicon Valley

When the chopping gave a storm to the world of Artificial Intelligence (AI), an indispensable question was followed: does it face trouble for China, America’s biggest technical rival?

Two years later, a new Chinese AI model has left that question: Can US stop Chinese innovation?

For some time, Beijing seemed to be a thunder with his answer to the spat, which is not available in China.

Uncontrolled users made fun of Erney, chatbot by search engine legend Baidu. Then the versions came by Tech firms TENCENT and bydentance, which were rejected as followers of Chatgpt – but not as good.

Washington was confident that it was ahead and wanted to keep it in this way. Therefore, the Biden administration increased the restrictions that ban advanced chips and technology exports to China.

That is why the launch of Deepsek has surprised Silicon Valley and the world. The firm says its powerful model is cheaper than the billions of American firms spent on AI.

So a small -known company – whose founder is being included on Chinese social media as “AI Hero” – close it?

This was certainly a blow when the US stopped China from selling advanced technology to the world’s major chip-makers such as NVidia from selling advanced technology.

They are necessary to create a powerful AI model that can demonstrate many human functions, from answering basic questions to solving complex mathematics problems.

Deepsek founder Liang Wenfeng described the chip ban as his “main challenge” in an interview with local media.

Long before the ban, Dipsec has an MIT technology review, NVDia A100 chips – estimates are estimated from 10,000 to 50,000.

The leading AI models in the West use estimated 16,000 special chips. But Deepsek says that it trained its AI model using 2,000 such chips, and thousands of lower -grade chips – which makes its product cheaper.

Some people, including the US Tech billionaire Elon Musk, have questioned the claim, arguing that the company cannot reveal how many advanced chips actually used restrictions.

But experts say Washington’s ban brought both challenges and opportunities to the Chinese AI industry.

This has forced Chinese companies like “Deepsek to Innovate”, so they can do more with less, Marina Zhang, an associate professor of the University of Technology, called Marina Zhang.

“While these restrictions face challenges, they have also increased creativity and flexibility, aligning with China’s comprehensive policy goals of achieving technical freedom.”

The second largest economy in the world has invested heavily in Big Tech – from the battery to electric vehicles and solar panels, to AI.

Converting China into a technical superpower has long been the ambition of President Xi Jinping, so Washington’s ban was also a challenge that Beijing took.

The release of the new model of Deepsac on January 20, when Donald Trump was sworn in as an US President, was deliberately performed, according to Gregory Sea Allen, AI expert from the Center for Strategic and International Studies.

Former director of strategy and policy, Mr. Alan says, “Time and the way it is being messed up – exactly what the Chinese government wants to think to everyone – export control does not work and America AI does not have a global leader in AI Is.” At the Joint Artificial Intelligence Center of the US Department of Defense.

In recent years, the Chinese government has nurtured AI talent, offered scholarships and research grants, and has encouraged partnerships between universities and industry.

According to Ms. Zhang, the National Engineering Laboratory has helped to train thousands of AI experts for deep learning and other state -backed initiatives.

And China had a lot of bright engineers for recruitment.

For example, take the team of Dipsek – Chinese media says it consists of less than 140 people, most of which proudly declared “home -developed talent” from aristocratic Chinese universities.

Western observers “recalled a new generation of entrepreneurs, which prefer basic research and long -term technological advancement on quick profits”, called Ms. Zhang.

China’s top universities are constructing “rapidly growing AI talent pools”, where the managers are often under the age of 35.

“After China grew up during a rapid technological climb of China, they are inspired by a campaign for self -sufficiency in innovation,” she says.

An example of this is a report of Deepsek founder Liang Wenfeng-40-year-old studied AI at the prestigious Zhejiang University. In an article on the Tech Outlet 36KR, people familiar with them say that he is “more like a geek instead of the boss”.

And the Chinese media described him as “technical idealistic” – he insists on keeping Dipsek as an open -source platform. In fact, experts also believe that a prosperous open-source culture has allowed young start-ups for pool resources and extended rapidly.

Unlike large Chinese tech firms, Deepsek preferred, which, according to experts and people working in the company, has allowed more use.

“Top 50 talents in the region cannot be in China, but we can build such people here,” Mr. Leiang said in an interview with 36kr.

But experts wonder how far the deepsek can go. Ms. Zhang says that “new American restrictions can limit access to American user data, potentially affect how Chinese models such as Deepsek can be global”.

And other people say that America still has a major advantage, such as in Mr. Alan’s words, “their huge amounts of their computing resources” – and it is not clear that the use of advanced chips to improve the model How will it continue.

But for now, Deepsek is enjoying its moment in the sun, given that most people in China had never heard of it by the end of this week.

His sudden fame has seen Mr. Liang becoming a sensation on China’s social media, where he is being appreciated as one of the “three AI heroes” of Southern Guangdong province, which is the border of Hong Kong.

The other two are Yang, a major expert at the University of Tsinghua, and Caiming, who teaches at MIT in the US.

Deepsek has pleased the Chinese Internet before the lunar New Year, the country’s biggest holiday. It is a good news for a unhappy economy and a technical industry that is ready for the potential sale of ahead tariff and Tikok’s American business.

“Deepsek shows us that only if you have a real deal then you will stand on the test of time,” written in a top-looking Weibo comment.

“This is the best new year gift. Our motherland is rich and strong,” reads another.

A “mixture of shock and enthusiasm, especially within the open-source community,” how Wei Sun, Principal of Counterpoint Research AI analyst, described the response in China.

Fiona Zhou, a technical activist in the southern city of Shenzhen, says his social media feed was “suddenly filled with deep-related positions”.

“People call it ‘Made-in-China glory’, and say that it jerks the silicon Valley, so I downloaded it to see how good it is.”

He asked it like “four columns of (his) destiny”, or Ba -Zaji – like a personal horoscope, based on the date and time of birth.

But for his disappointment, Deepsek was wrong. While he was fully clarified about its “thinking process”, it was not “four columns” from his real Ba-ja.

She says she will still let it go to work, as it will probably be more useful for such tasks.