RPT-BREAKINGVIEWS-China AI labs face growing open-source dilemma

Reuters
Yesterday
RPT-BREAKINGVIEWS-China AI labs face growing open-source dilemma

The author is a Reuters Breakingviews columnist. The opinions expressed are her own.

By Robyn Mak

HONG KONG, March 25 (Reuters Breakingviews) - To open-source or not may soon be the trickiest question facing China's top artificial intelligence companies. The $300 billion Alibaba 9988.HK and compatriots made most of their models free and available for anyone to use, which in turn has accelerated innovation and adoption across the country. Even U.S. firms have benefitted. But mounting investor pressure to generate a profit, as well as geopolitical pressures, may prompt a rethink to prioritise proprietary breakthroughs instead.

This week, buzzy San Francisco-based coding startup Cursor AI acknowledged its latest "frontier-level" model was built on an open-source one owned by up-and-coming Chinese AI lab Moonshot. It's an awkward situation: Cursor, backed by heavyweights including Andreessen Horowitz and Google GOOGL.O, is in talks to raise funds at a $50 billion valuation, Bloomberg reported this month, citing sources. That's nearly double the worth it secured in November and more than two and a half times that of Moonshot.

It also underscores a new reality in Silicon Valley and beyond: startups and more established companies alike are increasingly eschewing pricey proprietary models from OpenAI and others for free or low-cost Chinese versions which are rapidly closing the performance gap despite Washington's export curbs and sanctions. Rankings on OpenRouter, a platform that lets developers access and use different models, currently show 7 of the 10 most popular offerings are Chinese; companies from Airbnb ABNB.O to German industrial giant Siemens SIEGn.DE have made no secret about using models from the People's Republic.

Chinese tools, however, now have a target on their back. On Monday, the U.S.-China Economic and Security ⁠Review Commission published a report warning that China's open-source successes "reflect a more fundamental challenge to U.S. AI supremacy". And last month, OpenAI and Anthropic separately accused Chinese AI labs like DeepSeek of improperly using U.S. models to improve their own; both have called for tougher restrictions on U.S. tech.

There is a simple reason for Chinese companies to pivot away from open-source: Alibaba, Tencent 0700.HK and rivals are under increasing pressure to show they can monetise their AI models and applications. The e-commerce group last week even went as far as laying out a goal of surpassing $100 billion in annual revenue from cloud computing and AI customers over the next five years. Alibaba has reaffirmed its commitment to open-source, but reaching that lofty target will require devoting more resources to its high-end proprietary offerings - possibly at the expense of its open-source Qwen models.

Keeping more Chinese innovations behind closed doors may even suit officials in Beijing, who are increasingly anxious that the country is losing its best and brightest entrepreneurs. The world may soon see fewer free Chinese models.

Follow Robyn Mak on X.

CONTEXT NEWS

An executive from U.S. artificial intelligence coding startup Cursor acknowledged on March 21 that the company's latest model, the Composer 2, "started from" Chinese open-source model Kimi 2.5, after claims from a developer posting on X.

In November 2025, Cursor announced it had raised $2.3 billion at a post-money valuation of $29.3 billion. Its investors include Accel, Thrive, Andreessen Horowitz, Nvidia, Google and Coatue. The company is currently in talks to raise more funds at a $50 billion valuation, Bloomberg reported on March 12, citing sources.

Moonshot, which owns the Kimi model and chatbot, is raising funds at a target valuation of $10 billion, Bloomberg reported in February, citing sources. Its existing backers include Tencent and Alibaba.

Chinese low-cost AI models are the world's most popular https://www.reuters.com/graphics/BRV-BRV/xmpjynglbvr/chart.png

(Editing by Una Galani; Production by Aditya Srivastav)

((For previous columns by the author, Reuters customers can click on MAK/ robyn.mak@thomsonreuters.com))

At the request of the copyright holder, you need to log in to view this content

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

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