ChinAI #323: The AI Deflation of China's Tech Giants

Why are Chinese tech giants spending so little on AI Capex?

Greetings from a world where…

the feeling of watching True Detective season 1 for the first time must be what it was like for the first person to eat peanut butter or soup noodles

…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: In this US-China AI race, Chinese internet giants are rapidly becoming marginalized

Context: Why are Chinese tech giants spending 10x less on capital expenditures than their American competitors? In this week’s feature translation (link to original Chinese), Jinduan Research Institute [锦缎研究院] argues: “in the five years since the start of the AI race, Chinese internet companies have consistently lagged behind their American counterparts, at least in terms of capital expenditures and AI infrastructure development, resulting in significant ‘technical debt.’” The piece sorts through the factors behind this “AI deflation” [通缩] among Chinese tech companies.

Source: Jinduan Research Institute. Past five years of capital expenditures by Chinese (left bar) and American tech giants.

Back in the day, we used to do a lot of translations of “techlore” articles by “development bloggers”. Here’s how I characterized this style (ChinAI #98):

Longform pieces that read, at times, like epic poems in which the heroes (tech company leaders) wage battle over the commanding heights of the economy. “Development Bloggers” or the “Industrial Party,” usually people who have experience working in the tech industry and espouse techno-nationalist views, are emerging as a formidable force in Chinese media and the semiconductor industry is especially fertile ground for techlore.

I’ve shied away from these type of pieces in recent years, as they often do not provide sourcing for their claims (as we’ll see with this piece and as some of the WeChat comments point out). Still, this piece was aggregated by a well-known platform Huxiu, and has racked up 80k views in a few days, so I thought it would be useful to see how a Chinese development blogger views the US-China AI race — especially since it represents such a contrast with what we hear from the DC blob.

Key Takeaways: The capex gap between Chinese and U.S. tech giants is large and growing. From the article: “The combined capital expenditures of the four US tech giants (Google, Microsoft, Meta, Amazon) over the past five years reached 5.36 trillion RMB, while the combined capital expenditures of China's seven leading internet companies—Tencent, Alibaba, Baidu, JD.com, Kuaishou¹, Meituan, and NetEase—were only 630 billion RMB.”

• In 2020, this capital expenditure ratio (of Chinese tech giant capex vs. US tech giant capex) was 1:6, and by 2024, it had reached 1:10.

• Of course, not all capital expenditures by tech giants covers AI-related investments, but a substantial share does go toward data centers.

Why does this capex gap matter? To the author, investments in data centers will bring cumulative effects in the years to come: “For AI data centers, the impact of network effects is no less significant than that of search engines in the internet era and operating systems in the PC era. The more people use AI, the harder it becomes for competitors to catch up. There's little low-hanging fruit for Chinese software companies going global today. Every competition track is crowded.”

• Jinduan blames this capex gap for the U.S.’s clear lead in AI adoption. It relays a stat that the AI adoption rate has reached 78% among U.S. companies, and that the AI adoption rate will not exceed 15% for Chinese companies.

  • I was able to confirm the U.S. indicator, which comes from this McKinsey report.² I was not able to confirm the China indicator, and other WeChat comments questioned the lack of a supporting source.

  • Jinduan also provides a weekly active user statistic: US AI models have passed over 1 billion WAU; Chinese AI models have about 70 million WAU (again, no source for these claims).

• These two factors are intermingled. Does low AI capex drive low AI adoption investment, or is it the other way around? From the article: “Even without considering chip performance issues, the lower the AI adoption rate, the lower the return on AI capital expenditures. And the lower the return on AI capital expenditures, the further China will lose distance from the United States in the AI race.”

The main reason I wanted to flag this article is how it justifies this capex gap. It discounts the U.S. export controls, which it calls “the excuse many Chinese internet companies” use to justify their lack of capital expenditures.

• To be sure, there is some logic behind this so-called excuse: Chinese tech giants can’t purchase the most advanced chips and supply is not reliable, which makes it difficult to invest in long-term infrastructure.

• However, the author also calls out Chinese tech giants for using their profits for share buybacks and dividends, instead of capital expenditures. Specifically, according to Jinduan, “In 2024, the net total amount of Tencent's share buybacks, dividends, and debt repayments reached 168.1 billion RMB, more than double its capital expenditure for that year.”

• The author concludes, “U.S. export restrictions are only a small factor in Chinese tech companies’ ‘AI deflation’, as evidenced by the sudden emergence of Deepseek. Cynicism about technological growth and the use of propaganda (or, a public opinion battle) to mask their own inaction in AI are the real inner obstacles (“thief within” [心中贼]) facing Chinese tech companies.”

FULL TRANSLATION: In this US-China AI race, Chinese internet giants are rapidly becoming marginalized

ChinAI Links (Four to Forward)

Must-read: China Watching in Chinese - A Guide to Chinese-Language Analysis of Chinese Politics

This is really exciting and worth of your time! Shengyu Wang compiled a list of resources for insightful Chinese-language analysis of Chinese politics, including top ten Twitter accounts and WeChat public accounts. From the list, I’m particularly excited to check out veteran New York Times journalist Li Yuan’s Bumingbai [不明白播客] podcast.

Should-read: Tencent Research Institute and Factchecking Platform Jiao Zhen Analyzed 100 Cases of AI-generated Disinformation

I added an additional section to last week’s translation, in which Tencent Research Institute and JiaoZhen sort through the top 10 cases of AI disinformation in 2025. One of the top examples of AI-enabled misinformation occurred in February, when the following headline spread through the Chinese web: “Elon Musk tweets at 3 a.m.: It's hard to understand that WHO Director-General Tedros Adhanom Ghebreyesus earns $9.5 million annually.”

Should-read: Chinese scientists do a comprehensive safety study of ~20 LLMs – and they find similar things to Western researchers

From Jack Clark’s essential ImportAI newsletter: “Researchers with the Shanghai Artificial Intelligence Laboratory have conducted a thorough (~100 page) assessment of the safety properties of ~20 LLMs spanning Chinese and Western models. Their findings rhyme with those that come out of Western labs...”

Should-apply: Horizon Fellowship: bring your China expertise to Tech Policy

The Horizon Fellowship provides a direct path into the world of emerging tech policy: a fully-funded placement at a federal agency, congressional office, or think tank for up to two years. A deep understanding of China's technology ecosystem and governance approaches is essential for effective US policy on these issues.

For its 2026 cohort, Horizon is actively seeking candidates with expertise on China's technology ecosystem, policy landscape, and strategic goals. Prior technical or policy experience is not required. This is an excellent opportunity to work on some of today’s most pressing issues. The application deadline is August 28.

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.

1. I found it weird that Jinduan included Kuaishou in this list but not ByteDance (a few WeChat comments also pointed this out).

2. The report goes on to state, “However, this enthusiasm has yet to translate into tangible economic results. More than 80 percent of companies still report no material contribution to earnings from their gen AI initiatives.”