DeepSeek's Success Changes the Game in U.S.-China Trade Rivalry -- Barrons.com

Dow Jones
01-30

By Reshma Kapadia

The sudden success of Chinese upstart DeepSeek's AI model is forcing a reassessment of the U.S.-China strategic rivalry and how the U.S. steps up its game.

DeepSeek says it created AI programs on par with OpenAI's oi reasoning model without using the most advanced chips. Exactly how much it cost is up for debate, but its impressive feat after years of U.S. efforts to restrict China's access to critical technology is shining a light on how the U.S. tackles this competition.

Adding to the AI fever, China's Alibaba Group released a new AI model, describing it as better than DeepSeek's.

"The world needs to get more comfortable with a China that is highly competitive in all tech domains," says James Goodrich, a senior adviser for technology analysis to the Rand Corp., who previously led policy at the Semiconductor Industry Association.

Technology has been a big part of the shifting relationship between the two economic powerhouses, with each increasingly viewing policy through a national security lens. The result: a spate of efforts to become more self-reliant for critical goods and to try to disentangle from each other while bolstering efforts to dominate tomorrow's technologies.

Over the past eight years, U.S. efforts have included measures to hobble Chinese telecom giant Huawei, place tariffs on $350 billion of Chinese imports, incentivize companies to make goods in the U.S. through the CHIPS & Science Act and the Inflation Reduction Act and restrict China's access to critical advanced technology -- like the chips needed for AI -- through a blanket of export controls.

The Quest for AI Leadership

Beijing sees AI leadership as synonymous with economic and military power. That is why it has invested hundreds of billions of dollars and prioritized efforts to become more self-reliant amid fears of further U.S. restrictions. It is also building up its manufacturing base to take market share globally in areas ranging from autos to aviation.

China has been trying to shift its economy away from debt-fueled property spending toward domestic demand -- a transition where it has stumbled in ways that have sapped confidence among entrepreneurs and consumers and hurt its economy.

DeepSeek offers a small but significant win that could help private sector sentiment going into the Lunar New Year. "This is hailed as proof that its new growth model is working. Xi Jinping has been 'tech, tech, tech' and moving up the value chain as they reallocate resources from real estate and property to technology and finance," says Rory Green, head of TS Lombard's China research.

Chinese companies have been using software to get more out of the hardware they can get their hands on in the wake of U.S. export restrictions on its access to technology. If access to the most advanced AI training chips becomes less of an edge, Chinese companies will have more of a shot in catching up to U.S. competitors, according to Gavekal Research analyst Tilly Zhang.

For longtime China watchers like Sinology's Andy Rothman, it's the latest example of the resilience of Chinese entrepreneurs to obstacles China faces, internally and externally. For investors worried about the impact of Xi's tighter control on China and the flagging confidence among the private sector, there could be glimpses of optimism in DeepSeek's success.

'A More Pragmatic Approach'

"It reflects a directional move toward a more pragmatic approach by the government in some sectors, like AI. It isn't enough but a move in the right direction," Rothman says of the lighter approach Beijing has taken in areas it deems as critical.

That could be good for contrarian investors who have been eyeing the sharp discounts in Chinese technology stocks like Alibaba and Tencent Holdings, which have their own AI programs.

"When the U.S. markets were ripping, there was no reason to take any of this China market risk," says Ben Harburg, founder of CoreValue Alpha, which is home to exchange-traded funds investing in China as well as India and a recent "America First" product. But with Chinese stocks like Alibaba and Tencent trading at half to a third of U.S. peers on a price/earnings basis, often growing at a double-digits pace, he sees the makings of a reassessment.

Policymakers may need their own reassessments. DeepSeek is a microcosm of China's broader changing competitive positioning. China accounts for 30% of global manufacturing share, helped by developed market capabilities, developing market production costs and enormous state support. No longer is China just the factory to the world for toys and cheap goods. It is the leading autos and electric battery manufacturer. It makes 80% of cranes used at ports around the world and is the biggest maker of small drones.

The West has anchored on the idea "we think, they sweat" -- that they could cede dominance in areas like auto manufacturing because they would lead in software and AI, says Louis Gave, co-founder of independent macro research firm Gavekal. DeepSeek shakes that view.

China is gaining market share from Japanese, German and U.S. rivals in areas such as aviation and robotics. The U.S.-China battle has revolved around trade and technology in the past couple of years. Analysts now see it expanding into other areas, including pharmaceuticals.

Tools for the Battle

DeepSeek shines a spotlight on U.S. policy to deal with this evolving competition. While DeepSeek may have had access to some of Nvidia's advanced chips, Gregory Allen, director of the Wadhwani AI Center at the Center for Strategic and International Studies who previously worked on AI strategy at the Defense Department, says these controls work on a lag. The first batch in 2023 was tightened further in 2024, which could make it harder for DeepSeek to iterate on its work from here.

Allen and others expect the Trump administration to further tighten export controls to control China's access to technology. That could include tackling issues like transshipment so that China can't access the technology through third countries. In his confirmation hearing Wednesday, Commerce Secretary nominee Howard Lutnick said that the U.S. needed to do more to "not help" China.

The Trump administration is prioritizing innovation and investing in energy infrastructure. In a Fox interview, AI czar and PayPal co-founder David Sacks reiterated Trump's comments that this was a "wake-up call" and stood by Trump's $100 billion AI infrastructure push announced last week, even as DeepSeek appeared to offer a more efficient AI model. Some policywatchers also stress the need to think bigger to combat China's "whole of government" approach that includes plans to develop 20 new science labs and dozens of new universities to cultivate millions of research hers and engineers focused on frontier technology.

Goodrich warns that the U.S. .is "incredibly weak or weakening" in investing in national labs and universities -- the basic research and education that have been the building blocks of American innovation. While $200 billion was earmarked for investing in research and development in the 2022 CHIPS & Science Act, that part of the funding was never appropriated. "I fear we are losing our edge and understanding of success that got us here," Goodrich says.

Another avenue of concern: The more the U.S. withdraws from multilateral organizations, delivering foreign aid and its broader role abroad, the bigger the vacuum for China to fill. That includes areas like setting global standards on data and privacy.

Write to Reshma Kapadia at reshma.kapadia@barrons.com

This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.

 

(END) Dow Jones Newswires

January 30, 2025 02:00 ET (07:00 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.

免責聲明:投資有風險,本文並非投資建議,以上內容不應被視為任何金融產品的購買或出售要約、建議或邀請,作者或其他用戶的任何相關討論、評論或帖子也不應被視為此類內容。本文僅供一般參考,不考慮您的個人投資目標、財務狀況或需求。TTM對信息的準確性和完整性不承擔任何責任或保證,投資者應自行研究並在投資前尋求專業建議。

熱議股票

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10