ChinAI #275: What does China's government procurement market tell us about large model diffusion?

Greetings from a world where…

The giveaway for paying subscribers of ChinAI is still going on! If you financially support ChinAI and you’d like a free copy of my book Technology and the Rise of Great Powers (coming out August 20), please let me know via email, and I’ll make sure to mail you a copy once it’s out

…As always, the searchable archive of all past issues is here. Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors).

Feature Translation: Who took all the bidding orders for large models?

Context: At a time when the consumer-facing commercialization prospects of large models are still unclear, it is helpful to evaluate the extent to which these AI systems have landed with large firms and government entities. One important reference point: China’s multi-trillion-dollar public procurement market, which consists of purchases by not just the Chinese government but also central government-owned and operated state-owned enterprises (SOEs). In fact, according to a US-China Business Council report, central SOE procurement expenses may be around four times the size of the central government procurement market. Compiling public data on the winning bids for large model projects in the procurement market, this Gui Xingren [硅星人] article (link to original Chinese) sheds light on the diffusion of large models to business-facing and government-facing applications.

Key Takeaways: In the first half of 2024, the procurement market generated 498 bids related to large models, with a total bid amount of over 1.34 billion RMB, which more than doubled the total procurement amount for the entire 2023 year.

• This self-media account, Gui Xingren (recommended by the IT Juzi channel), got data on bids from Zhiliaobiaoxun [知了标讯].

• In 2024, the largest project so far is the Guizhou Eastern Data and Western Computing large model construction project, which seeks to boost data center capacity in western regions, where energy is cheaper than in eastern regions.

• Zhipu AI, Baidu, iFlytek, and Huawei ranked at the top in terms of the number of bids that they won.

Let’s take a look at 17 of Baidu’s large model-related projects secured through the public procurement market. The table below is my translation of the article’s compilation of Baidu’s successful bids in 2024, which includes the name of the project, the purchasing party, the bid amount, and additional notes.

• As you’ll see, the purchasing parties for all these projects are government entities or SOEs. For instance, Xiamen University is a comprehensive university but it is directly administered by China’s Ministry of Education.

• The article recaps, “Baidu cooperated with many large SOEs and industry leaders, such as China Southern Power Grid, China Merchants Group and China Unicom. According to incomplete figures, in the field of intelligent Internet of Vehicles, Baidu is the only company that has won bids for large model-related projects in transportation.”

One interesting finding is that, aside from Zhipu AI, China’s large model startups have made relatively few moves in the public procurement market. Based on this article’s data, MiniMax, Baichuan, and 01.AI have not secured any bids in 2024.

• The article reasons that this may be due to “the complex and diverse needs of the Chinese enterprise market, the fragmented industry market, and the large number of small and medium-sized customers. These customers have a strong demand for customized solutions, making it difficult to quickly standardize and modularize AI products. In the To-Government market, although the competition for intelligent projects is fierce, the profits are meager, and government and enterprise institutions prefer to introduce packaged AI, cloud, IoT and other software and hardware solutions. Therefore, more companies choose To-Consumer as a breakthrough point.”

• Last week’s feature translation compared the financing raised by this generation of Chinese large model startups to the fundraising of an earlier generation of computer vision start-ups. This week’s feature text continues that comparison: “The AI ​​wave that began around 2015, the AI ​​Four Little Dragons that grew out of computer vision technology, mainly made profits by providing customized AI solutions to enterprises and governments. Enterprise and government bidding and centralized procurement became an important source of income for them but also a root cause of their inability for successive follow-ups.” In sum, as we await the future diffusion of large models, the public procurement market may be an important client base, but these models will also need to diversify beyond this revenue source.

FULL TRANSLATION: Who took all the bidding orders for large models?

ChinAI Links (Four to Forward)

Should-read: Book review of Technology and the Rise of Great Powers in Survival

Thanks to Melissa K. Griffith, lecturer in Technology and National Security at Johns Hopkins University SAIS, for her generous and thoughtful review of my book in the Survival journal. Listen to her on why you should get your copy!

“To observe that Technology and the Rise of Great Powers has had a timely release would be an understatement. Anyone with an interest in the often misunderstood history of industrial revolutions, the evolution of disruptive technologies, the potential for artificial intelligence to reshape global economic power and the future of US–China relations should pick up a copy.”

Should-read: Cloud Computing White Paper (2024)

A regular source of in-depth white papers on emerging and digital technologies, the China Academy of Information and Communications Technology (CAICT) has published the 10th edition of its cloud computing white paper. Some good details on the interaction between cloud computing and AI development.

Should-read: Practice and Prospects for the Application of Large Models in the Government Domain

Caught up on some CSET translations this past week. This translation of a research report by CAICT’s policy and economy institute describes how the Chinese government is using large AI models. From CSET’s summary:

The authors list more than 50 large models developed by Chinese tech companies that provincial and local governments in China have deployed for various purposes, and provide four brief case studies. While these AI systems have the potential to improve Chinese local governments’ provision of public services, the focus of many of these governments on early detection and suppression of social disturbances suggests that these AI models are being used to enhance and intensify China’s surveillance state.

Should-read: Podcast narrations of ChinAI newsletter

I listened to some narrations of the ChinAI newsletter recently, and they’re getting really strong. There’s even accurate pronunciations of Chinese characters. Good way to keep up-to-date with ChinAI if you’re on the move!

Thank you for reading and engaging.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.

Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).

Also! Listen to narrations of the ChinAI Newsletter in podcast format here.

Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99