Whilst India might have replaced China as the fastest growing large economy in the world, China is in a totally different race altogether – the technology race with the world’s super-power, the US. Here’s a blog by a venture capitalist, taking stock of where the US and China stand on three key aspects of the technology race today – AI, semiconductors and electric vehicles.
The author begins with an on the ground report on how bad is the economic slowdown in China:
“With public equities AND real estate prices in free fall, consumers are rightly wary of depleting their cash reserves despite China’s current deflationary environment. This was probably the first time since living in/visiting China (between 2005-2020) that I could almost “feel” the downturn…These “macro trends” are quite visible on the ground. The malls in China are more empty than prior years.. Restaurants, even high end ones, are consistently offering set menus to entice diners… Durable goods, meanwhile, have also deflated quite a bit, and domestic giants like Pinduoduo and ByteDance are locked in brutal price wars.
These are all indications that China is running out of time to transition into a full fledged advanced economy before they fall into the infamous middle-income trap…”
This sets the context for why China needs to win the technology war. On low end tech (such as e-commerce, social media, fintech), the author claims China has eclipsed the west with the likes of Alibaba, TikTok, Shein, etc. But the real race is on the high-end (AI, EV and chips).
On AI, the author reckons the US is ahead, not by much though:
“In terms of model capabilities, China has demonstrated that it’s also able to train quite competitive models (Qwen 1.5-72B-Chat, Yi-34B-Chat), though they are a step behind the leading edge. Looking at the performance of leading open models (e.g. Databricks’ DBRX, which China can also use) relative to closed source models from OpenAI/Anthropic, the West is ahead by 2-3 years at best, and potentially less in areas like multi-modal models.”
Where China is at a disadvantage is on compute capability for AI: “….sanctions on both import of chips, as well as the tooling to make them ensures that Chinese companies and academic labs will have punitive TCO (total cost of ownership) for the same FLOPS”
[For those interested, the author gives a useful four pronged framework to assess model building capabilities].
That leads us to the war on chips – on the one hand, the sanctions make it imperative that China becomes independent on chip fabrication capabilities…
“China has become quite competent at chip design. Fabrication is another matter. Due to import restrictions on EUV (extreme ultraviolet) lithography machines from ASML (required to create leading edge chips), China only has access to older DUV (deep ultraviolet) machines, and there are even some restrictions on these! As a result, China’s national champion SMIC needs to aggressively use multi-patterning to get to 7nm/5nm, and has been rumored to only achieve quite uneconomical ~50% yields, vs. an industry standard of 90% yields. The key question here is whether China can replicate an entire semiconductor fabrication stack in-house, or find workarounds to catch up to the West (which will be on the 2nm process soon). Another risk is here is that the next level of sanctions could result in ASML withholding support engineers and spare parts, forcing China to draw down its existing stockpiles.”
.. .whilst on the other hand, the US is still heavily reliant on Taiwan for fabrication and hence needs to reshore those capabilities quickly.
“…while the first dollars from the CHIPS Act are starting to be deployed, we’re already seeing why the world of atoms is so hard. TSMC’s $40B factories in Arizona will already be a step behind the leading edge by the time they come online (N4 in 2024/2025, N3 in 2026, with possible delays). The US essentially needs multiple CHIPS Acts to continue to incentivize TSMC’s leading-edge fab investments stateside. Therefore, in any potential “hot war”, the US will also lose access to its most advanced chips.”
On EVs, the author contends that China has a formidable lead over the west despite Tesla:
“Companies like Ford, GM, Mercedes and Volkswagen have already scaled back or delayed their plans to roll out EV-only fleets. Here, we see a clear example where the US ability to iterate on the design and manufacturing of hardware have slipped significantly. In North America, Tesla is essentially the only “scaled” US-based EV manufacturer, and the only vehicle one can readily purchase….
…in China, competition between local manufacturers is fierce, and Chinese vehicles come with significantly more features and at lower cost relative to the West. Just this week, Xiaomi threw its had in the ring with its SU7 vehicle which costs $4K less than a Model 3! I’ve test driven around 10 different vehicles at this point in the 150K-300K RMB price point (~$21-42K USD), and these vehicles are quite capable. Many “premium” features come in standard—long range (500-800km/~300-500 miles), snappy infotainment systems, quality of life features (360-degree cameras, parking assist, built-in voice assistants, etc.).”
Just last week, Apple (considered the flag bearer for western tech) announced that it will no longer be pursuing its car project – the author reckons Apple might have burnt $10bn over the last decade. So, its not like the west can spend its way to catch up.
If you want to read our other published material, please visit https://marcellus.in/blog/
Note: the above material is neither investment research, nor financial advice. Marcellus does not seek payment for or business from this publication in any shape or form. Marcellus Investment Managers is regulated by the Securities and Exchange Board of India as a provider of Portfolio Management Services. Marcellus Investment Managers is also regulated in the United States as an Investment Advisor.
Copyright © 2022 Marcellus Investment Managers Pvt Ltd, All rights reserved.
