Discover more from ChinAI Newsletter
ChinAI #244: Can Chinese Automotive Chips Overtake on the Turn?
Examining factors behind China's low localization rates in automotive chips
Greetings from a world where…
I’m very jealous of my grandma going to eat crab at Yangcheng Lake
…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 localization rate of automotive chips is still less than 10%. How can Chinese automotive chips take opportunities to overtake?
Context: I used the Emerging Technology Observatory's Scout tool again to scout out this week’s feature translation, from Caijing’s transportation industry portal (link to original article). Here’s Scout’s summary:
Driven by the growth in demand for new energy vehicles, China's automotive chip sector is also undergoing rapid development. Nearly 300 companies are developing a range of chip products in areas such as cockpits and autonomous driving, and the automotive chip market is expected to reach $29 billion USD by 2030. However, while China's chip production localization rate has increased from 5% to 10%, China still lags behind the major chip-producing countries, such as the United States and Japan. Challenges remain due to foreign protectionist measures, notably including the U.S.'s 2022 CHIPS and Science Act, and technical weaknesses in the domestic industry.
Key Takeaways: Chinese automotive chips are gaining speed. In the Navigate on Autopilot (NOA) intelligent driving chip market, Chinese start-up Horizon Robotics (49% of market) has edged Nvidia to become the industry leader.
One semiconductor firm CEO details the case for optimism to Caijing: “The largest automobile market is in China, the fastest engine of technological innovation and change is in China, and the fastest growing downstream customers are also in China. Chinese chip manufacturers are based in this type of innovative soil in China. They have the best advantage to work closely with downstream partners to jointly create some independent and indigenous products in the future.”
On that point, China accounts for 60% of global new energy-vehicle production. This was a fun stat: average # of chips used in new-energy vehicles > 1,000; whereas, # of chips in consumer electronics like phones and PCs sits at ~dozens to hundreds.
As another chip company exec relates, “In the early days, Nvidia and Qualcomm chips basically monopolized more than 95% of the market, but currently domestic Huawei MDC chips, Horizon J3 and J5 chips and Black Sesame A1000 chips have begun to enter the market and have entered the global OEM and Tier1 supplier system.”
Still, the localization rate of automotive chips still remains below 10%. Why?
Accurately measuring localization rate requires a comprehensive examination of the supply chain. One market researcher states: “One of the biggest pain points in the entire industry is that the Tier 2 in automotive semiconductors (lower-level suppliers or service providers in the supply chain) is monopolized by foreign companies. Foreign companies account for about 50% of the market among the top five Tier 2 companies, and account for 70% of the top ten.”
The article also mentions the CHIPS and Science Act and export controls as major barriers. One other barrier that I found especially interesting: “In terms of testing and verification, China currently does not have its own standards, nor does it have an authoritative organization corresponding to international standards. This is an important link that restricts the development of automotive chips and the globalization of automotive products. In addition, China lacks an application system to verify chip adaptability at the actual vehicle level, making it impossible to achieve rapid iterative verification.”
Dig deeper: In ChinAI #123, I translated a Leiphone report on localized substitution in the “Xinchuang Industry): chips, servers, cloud, basic software, application software, and information security, etc.
ChinAI Links (Four to Forward)
Recently published in Comparative Political Studies, Yuen Yuen Ang, Nan Jia, Bo Yang, and Kenneth G. Huang question China’s innovation drive:
Can China catch-up with the United States technologically by mobilizing its bureaucracy and assigning ambitious targets to local governments? We analyzed an original dataset of 4.6 million patents filed in China from 1990 through 2014 and paired this with a new, rigorous measure of patent novelty that approximates the quality of innovation. In 2006, China’s central government launched a national campaign to promote indigenous innovation and introduced bureaucratic targets for patents. Our analysis finds evidence that these targets, combined with political competition, pushed local governments to “game the numbers” by channeling relatively more effort toward boosting non-novel—possibly junk—patents over novel patents. Nationally, this is reflected in a surge of aggregate patents paired with a falling ratio of novel patents. China’s innovation drive is susceptible to manipulation and waste—it is enormous in scale but low in productivity.
Considering making this the feature translation this week: a fun exploration of large language models’ understanding of classical Chinese. The piece examines a LLM, constructed by Beijing Normal University researchers, specifically trained for comprehending classical Chinese texts.
Should-read: President Biden’s new plan to regulate AI
This Vox explainer, by Sara Morrison, was very helpful for catching up on the Biden administration’s lengthy executive order on the safe, secure, and trustworthy development and use of AI.
Should-read: Can Sampling Survive in the Age of AI?
A really fun read, by Eric Ducker for my favorite website The Ringer, on the intersection of AI tools and “sample snitching”: With the spread of AI technology, there is a fear not only that uncleared samples from the past will be exposed, but also that current producers will feel creatively stifled since their most ingeniously disguised samples could still get called out.
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 email@example.com or on Twitter at @jjding99