ChinAI #191: AI Venture Capital Trends Report
Revisiting IT Juzi data
Greetings from a world where…
mosquitoes seem to target DC newcomers
…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: China AI Venture Capital Data Report (IT Juzi)
Context: Back when I first started researching China’s AI development at GovAI in 2017, I gained a lot from an IT Juzi (IT桔子) report on the scale of China’s AI ecosystem. My “Deciphering China’s AI Dream Report” frequently cited that report, which draws from IT Juzi’s database on financing and venture capital in technology fields. This week, let’s revisit IT Juzi and break down their recent report on venture capital trends in China’s AI landscape.
Key Takeaways: According to IT Juzi data, in 2021, only 57 AI companies were established in China. This number has fallen sharply from the peak in 2018, when Chinese entrepreneurs started 1089 new AI companies.
See image below for the full time distribution from 2010 through 2021. That first bar shows that there were 892 Chinese AI companies before 2010.
Note: this database collects “general AI companies” [泛人工智能] companies. Maybe another way of interpreting this concept is: “core AI companies.”
Based on number of transactions and total investment amounts, the capital market declined in 2019 and 2020, but it has since rebounded to peak levels. The IT Juzi summary reads, “It can be said that China's AI field has passed through the capital market’s cold winter period, and there is no obvious low-valley period.”
The types of financing events have changed over the years, as well. In 2014, seed/angel and A round financing accounted for 76% of investments in this domain. In 2021, that share had declined to 51%.
Per the IT Juzi interpretation, the rising proportion of investments going to B and C rounds, as well as IPO and post-IPO financing, shows that the AI industry is maturing.
We can also learn a lot from the distribution of investments across different AI subdomains. The top 10 subdomains, in terms of companies that received financing, are intelligent finance (575 companies), intelligent robots, intelligent transportation, computer vision and imaging…and smart logistics (251 companies). Image below shows the number of companies that received financing by subdomain. Each of the subdomains are organized into one of three layers: industry application, technical, and foundational.
The industry application layer (e.g., smart finance) is the most active area in China’s AI landscape, as the capital market favors companies that are closer to commercialization opportunities. In contrast, only one sub-domain (data platforms) in the foundational layer appears in the Top 10. Developments in other fields such as storage and AI chips, per IT Juzi’s assessment, are relatively weak.
IT Juzi does note that “Against the backdrop of Sino-US trade frictions, the proportion of investments going to the AI infrastructure layer centered on AI chips, computing power, and data platforms is increasing.”
For other details, including the geographical distribution of Chinese AI companies, see FULL TRANSLATION: Through the cold winter, China's AI industry will receive nearly 400 billion (RMB) in investments in 2021 | IT Juzi releases “China Artificial Intelligence Venture Capital Data Report”
ChinAI Links (Four to Forward)
Should-read: Chinese Authorities Announce $1.2B Fine in DiDi Case, Describe ‘Despicable’ Data Abuses
An important DigiChina translation by Graham Webster of the Cyberspace Administration’s announcement and Q&A on the investigation into ride-hailing company Didi Chuxing. One of the illegal activities highlighted was the “excessive collection 107 million pieces of passenger facial recognition information.” Didi was punished for violating the Cybersecurity Law, the Data Security Law, and the Personal Information Protection Law.
I learned a lot from this Foreign Policy piece by Selina Lee and Ramona Li:
By branding a country as an aesthetic, context and meaning gets lost. In most cases, these symbols carry little relevance to the topics discussed in contemporary news cycles. If journalists want to separate themselves from the echo chambers that define authoritarian state-run media, they can start by actually illustrating the topic of their journalism, instead of resorting to the same overused and insensitive visual tropes.
Cate Edmondson for The New York Times has a good read on “the most significant government intervention in industrial policy in decades,” which is interpreted as a rare bipartisan consensus for “forging a long-term strategy to address the nation’s intensifying geopolitical rivalry with Beijing.” The House voted to pass this bill last Thursday.
A short diverting piece by Daniel Flatley for Bloomberg on how the above bill lost its previous name: The Endless Frontier Act.
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 a postdoctoral fellow at Stanford's Center for International Security and Cooperation, sponsored by Stanford's Institute for Human-Centered Artificial Intelligence.
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).
Any suggestions or feedback? Let me know at email@example.com or on Twitter at @jjding99