ChinAI #328: The Cold Reality for Chinese AI Start-ups

Greetings from a world where…

Key & Peele sketches never get old

…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: The Realities of AI Start-ups in 2025

Context: A branch of TMTpost (a Chinese tech, media, and telecommunication outlet) interviewed twelve developers, five AI community mods, and seven investors to get their take on the current state of the AI startups ecosystem in China. Much thanks to Lily Li for contributing this week’s feature translation and commentary. A recent graduate of University of Toronto with a PhD in theoretical computer science, she is making the transition to the AI safety field through projects in technical and governance research.

What follows is Lily’s analysis (lightly edited by me):

---

Key Takeaways: For software-based AI startups, the general consensus is that Chinese consumer and businesses are unwilling to pay for AI products and services. The article postulates that:

• Chinese consumers are unwilling to pay because they have grown accustomed to browsing the web from their phones and complex/standalone features do not integrate well into their viewing habits.

  • The article quotes one developer saying, “The general consensus is: launch products overseas. Develop it in secret and make money on the down low. Launching products domestically is a loss-making endeavor. If it's truly unfeasible, relocate everyone to Singapore or elsewhere.” Chinese video tool Kling AI, for instance, generates 70% of its $100 million USD in annual revenue from overseas.

• Chinese businesses are unwilling to pay because labor is relatively cheap and most industries are not fully digitized and thus cannot take advantage productivity software. From the article: “The ARR (Annual Recurring Revenue) of leading AI companies is five to one hundred times higher in North America and the average annual procurement budget for North American businesses is nearly 10 times higher.” (see image below)

The interviewees were far more optimistic about hardware-based Chinese AI startups. China already has many tech hardware companies with robust supply chains and access to vast pools of talent. Taking advantage of these strengths may allow these startup to overtake their competition.

• The secondary markets have reflected this sentiment as opportunities to invest in hardware projects become a hot commodity in the past six months.

AI startup have a high barrier of entry and VCs find themselves competing for the same list of about 200 top talent with the technical background and powerful network necessary to launch a successful AI product. While some investors court “second-tier developers” with competitions, developer communities, boot camps, and free compute credits, the path for these developers is necessarily an arduous one.

Large Chinese tech companies have been caught flat-footed when it comes to AI. Their models are much weaker than those of frontier US companies and they are still mostly focused on their usual revenue metrics (traffic on their existing platforms) rather than making investments to AI infrastructure and developing the China AI ecosystem. The result is the loss of AI talent to opportunities abroad and their ecosystem is generally unattractive to domestic developers.

• The article cites this blunt assessment from a media outlet, “The US is using trillions of dollars of 'asset-heavy' investments to turn AI into the next generation’s water and electricity, while China is pursuing AI with an 'asset-light' mindset, still focused on monetizing views.”

• Another section: “Several leading developers remarked that if they were to start a new project, they most likely would not choose the AI ​​ecosystem of a major domestic company, nor would capital from market-driven funds be their top choice. Instead, they prefer Open AI’s incubator because it offers both financial support and a robust AI ecosystem.”

FULL TRANSLATION: The Realities of AI Start-ups in 2025

ChinAI Links (Four to Forward)

Must-read: China's Big AI Diffusion Plan is Here. Will it Work?

Matt Sheehan launched his Substack newsletter with a meticulous breakdown of China’s AI Plus initiative. He extensively builds out both the bearish and bullish case against AI Plus and then concludes with his own judgement. As just one example of how Matt does his homework on these issues, check out his comments on last week’s ChinAI full translation: Matt takes the extra step to poll multiple contacts to verify whether the 90% target applies to the whole economy or six priority sectors.

Should-read: Taiwan’s “silicon shield” could be weakening

I try to keep up-to-date on Johanna Costigan’s work, and I’m late to this MIT Tech Review article she published last month on Taiwanese semiconductor champion TSMC. It’s a detailed analysis of how this firm is navigating China’s ramped up pressure and U.S. demands for friendshoring.

Should-read: Working with US CAISI and UK AISI to build more secure AI systems

This was a really interesting OpenAI update on their collaborations with the U.S. and UK AI security institutes, which gave these institutes access to non-public prototypes of their systems. As they write, “These collaborations represent some of the deepest public-private collaborations on evaluating real-world frontier AI systems for security and against misuse, and we hope they serve as promising models for the field.”

Should-read: Silicon Valley enabled brutal mass detention and surveillance in China, internal documents show

This Associated Press investigation, based on independent reporting by Yael Grauer, finds that U.S. tech companies have supplied technologies that directly support Chinese surveillance systems. The evidence is sourced from tens of thousands of leaked emails from a Chinese surveillance company as well as thousands of procurement records.

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