China vs. USA– The Long Game

For once, Jensen is not long-term enough.

• Jensen has bemoaned the potential loss of the Chinese market which he thinks will be worth $50bn.
• Even though he is famed for thinking way ahead of everyone else, this time he is too short-term focused, as what US and Western companies lose in China, I think they will get back elsewhere in time.
• In an interview with CNBC, Jensen Huang, Founder & CEO of Nvidia, stated that he thought that the Chinese AI chip market could be worth $50bn over the next few years and that it would be a “tremendous loss” if Nvidia were to be cut out of that market.
• This was a thinly veiled pitch to the Trump administration in an attempt to limit any further restrictions on its ability to sell into China, but I don’t think that it is with the USA where the issue now lies.
• Despite the claims and the propaganda, China remains years behind the West on semiconductors, and it is extremely unlikely that it will ever catch up.
• This has forced China to spend billions on developing its own semiconductor technology and billions more on subsidising chips and services that are uncompetitive when it comes to cost.
• As a result, I think that China is absolutely determined that it will never suffer another dependency, and so, regardless of whether Nvidia is restricted or not, I suspect that China is looking to stop buying data centre chips from Nvidia, AMD, and everyone else that is not Chinese.
• In general, if China can replace a piece of foreign technology with a home-grown variant, then it is likely to do so as soon as it is viable.
• This is why there are at least 12 fabs under construction that will build 7nm chips using the multi-patterning technique developed by TSMC and Intel, and now being evolved by Huawei and SMIC.
• This allows China to make 7nm chips without using EUV, but this comes at the cost of a much more complicated manufacturing procedure and lower yields.
• This is crucial because this means that when competing against Nvidia and AMD at 4nm and beyond, the homegrown chips will be much more expensive when one considers total token output per dollar spent.
• In the Chinese market, this matters less because, as far as the Chinese state is concerned, this is a matter of national security, and so it has been subsidising 7nm chip production and is showing every sign of continuing to do so.
• The problem is that as usage of the 7nm process grows, subsidisation becomes less and less affordable.
• Combine this with the lacklustre Chinese economy, demographic issues, the huge debt pile and the dreadful state of the Chinese real estate market, and this rapidly becomes unaffordable.
• It is impossible to tell how long this can continue, but I don’t think that the Chinese state is in any condition to continue this subsidisation outside of China.
• In the long term, this means that the competitive dynamic for 3^rd party countries who will be forced to choose between adopting Western technology or the Chinese variant will have a much easier choice.
• Historically, Chinese technology has been much cheaper but came with strings attached to Beijing, but now it looks to me like it will be more expensive and still have those strings attached.
• I think it will be a decade or more before China manages to get EUV working properly, and so I don’t think that this competitive dynamic is going to reverse itself anytime soon, if ever.
• Hence, the Western variant will be cheaper as it uses more advanced silicon, making it a better choice for all 3^rd party countries.
• For example, no one I have spoken to in the telecom industry who used Huawei for 4 and 5G had any intention of using Huawei for 6G when it finally makes an appearance.
• This is how I think that whatever Western companies lose in China, they will more than make up in market share gains in unaffiliated countries, although this will take some time.
• This is also crucial in determining the outcome of the ideological struggle between the two sides that is currently being fought in the technology sector.
• This is the one advantage of the accelerating balkanisation of the technology sector that RFM Research and Alavan Independent have been predicting for the last 6 years, which we continue to believe will materially reduce the size of the opportunity for everyone.
• Hence, I think Jensen will get more than the $50bn back, but he will have to wait for a while to get it.