The emergence of ChatGPT in the artificial intelligence (AI) arena raised concerns about its implications for China, America’s primary tech competitor.
Two years later, China’s new AI model has reversed that narrative, prompting a new question: can the U.S. impede Chinese innovation?
Initially, Beijing’s response to ChatGPT seemed lackluster, with Ernie, Baidu’s chatbot, and models from Tencent and ByteDance drawing criticism.
Confident in its lead, Washington doubled down on chip and technology export bans, aiming to curtail China’s technological advancements.
Yet, DeepSeek’s groundbreaking launch has defied expectations. The company claims its powerful AI model is more cost-effective than U.S. counterparts.
How did DeepSeek, a relatively obscure firm led by a founder hailed as an “AI hero,” manage this feat?
The Challenge
The U.S. restrictions on advanced chip exports dealt a significant blow to China, given the necessity of such chips for AI development.
DeepSeek founder Liang Wenfeng identified the chip ban as their primary obstacle, yet the firm strategically stockpiled Nvidia A100 chips beforehand.
Reports suggest DeepSeek acquired between 10,000 and 50,000 A100 chips before restrictions. However, its AI was trained with just 2,000 of them.
The company supplemented these high-grade chips with thousands of lower-grade alternatives, achieving cost efficiency in its product.
Skeptics, including Elon Musk, question the transparency of DeepSeek’s chip usage, citing restrictions on advanced chip procurement.
Experts argue that the U.S. export ban, while challenging, forced Chinese firms to innovate, making them resourceful and efficient.
Marina Zhang from the University of Technology Sydney notes these restrictions spurred creativity, aligning with China’s push for technological self-reliance.
China has heavily invested in emerging technologies like AI, EV batteries, and solar panels to solidify its position as a tech powerhouse.
President Xi Jinping’s ambition to make China a global tech leader meant U.S. sanctions were a challenge Beijing was determined to tackle.
Gregory C. Allen of the Center for Strategic and International Studies believes DeepSeek’s release was strategically timed for maximum impact.
“The timing signals China’s defiance, asserting that U.S. export controls have failed to hinder their AI capabilities,” Allen explained.
The Talent
DeepSeek’s team comprises fewer than 140 people, many of whom are graduates from elite Chinese universities celebrated as “home-grown talent.”
This new wave of entrepreneurs prioritizes foundational research and long-term innovation over quick profits, shifting perceptions of Chinese tech firms.
China’s top universities are cultivating a robust AI talent pool, with young engineers motivated by self-reliance and technological independence.
Liang Wenfeng, DeepSeek’s 40-year-old founder, studied AI at Zhejiang University and is described as a “geek” and “technical idealist.”
Liang insists on maintaining DeepSeek as an open-source platform, leveraging collaboration to accelerate progress in the AI community.
DeepSeek’s focus on research rather than commercial gain allowed it to experiment and develop innovative approaches to AI.
Liang has emphasized cultivating talent locally, expressing confidence in building world-class expertise within China.
Despite DeepSeek’s success, concerns persist about how it will overcome challenges such as limited access to global datasets and advanced chips.
The U.S. retains advantages like vast computing resources, raising questions about DeepSeek’s ability to sustain its momentum.
For now, DeepSeek’s breakthrough has captured China’s imagination, bringing national pride and optimism to its beleaguered tech sector.
The New AI Heroes
Liang’s sudden fame has turned him into a social media sensation, celebrated alongside fellow “AI heroes” Zhilin Yang and Kaiming He.
DeepSeek’s launch coincided with Lunar New Year, providing a morale boost for China’s tech industry amid economic and political challenges.
A viral Weibo comment praised DeepSeek, stating, “Only real innovation stands the test of time.” Others called it “the glory of made-in-China.”
Wei Sun, an AI analyst, described the reaction as a mix of shock and excitement, especially among open-source enthusiasts in China.
In Shenzhen, tech worker Fiona Zhou noted her social media feed flooded with posts about DeepSeek’s potential to rival Silicon Valley.
Zhou tested the AI for a personalized horoscope but found it inaccurate, although she remains hopeful about its professional applications.
DeepSeek’s rise symbolizes China’s resilience and resourcefulness in the face of adversity, inspiring confidence in its future technological ambitions.