USA vs. China – The Grind

Pressure is making life very hard for Chinese semiconductors.

• SMIC reported difficult results as the cash machine that is supposed to be lagging edge is not delivering meaning that the losses that are being incurred at its most advanced, but uneconomic fabs are increasingly being felt in its financial performance.
• Q1 24 revenue / net income was $1.75bn / $71.8m broadly in line with forecasts of $1.69bn / $76.8m but this is not where the real issue lies.
• The real story lies in the gross margin which once again weakened and there is no sign that this is going to improve for as long as SMIC pursues “leading edge” technologies.
• Q1 24 gross margin was 13.7% down 11.0% in Q4 23 and 20.8% in Q1 23.
• SMIC put this down to domestic competition, but I suspect that behind the scenes, life is more difficult than the company is making out.
• SMIC blamed domestic rivals were cutting prices to poach customers, but this explanation does not quite make sense.
• This is because SMIC is the biggest of the Chinese domestic manufacturers which means that when run properly, it should have a significant scale advantage over its peers.
• Hence, it should be SMIC leading prices down and being able to make money where all of its peers are in the red.
• Instead, it seems to be smaller competitors leading prices down with SMIC having to follow suit and suffering as a result.
• This leads me to conclude that the real issue with SMIC is to be found at the “leading edge” where it is using a multi-patterning technique to manufacture at 7nm and 5nm because it has no access to advanced EUV equipment.
• Both Intel and TSMC tried to do multi-patterning for 7nm and abandoned it because the yields were so low that it was not economical to continue.
• However, we are expected to believe that SMIC has got this to work well enough so that its yields are economic and that it is making money.
• I think that the reverse is true and that a part of the weakness we have seen in gross margins over the last several quarters is due to increasing revenues coming from chips made using multi-patterning techniques.
• The expectation is that revenue from “leading edge” nodes will increase meaningfully this year which I think will result in sustained pressure on gross margin.
• I see this as a sign that the US-led strategy to slow China’s rise as a technological power through semiconductor restrictions is working.
• Semiconductors lie at the heart of everything, and this strategy has been expanded into AI where the last two generations of Nvidia and AMD AI training chips are blocked from being exported to China.
• None of this will stop China from making the products that it needs, but these products will cost much more to make and take more time to complete putting severe financial strain on the technology industry.
• The Cold War between the West and the Soviet Union was not won by ideology but by economics, as the USSR could no longer afford to keep subsidising its industries and its military.
• While China’s semiconductor industry can make products at much higher cost, the state will need to support the industry which is becoming increasingly difficult as China’s economy limps along.
• Furthermore, in African, South East Asian and Latin American countries, Chinese technology will no longer be the cheaper option making Western products and standards more competitive in those markets.
• As the technology balkanises into competing standards as opposed to one global standard, the more countries that adopt the Western variants over Chinese, the greater the balance of power will tilt in the West’s favour.
• However, in the long run, multiple and incompatible standards are bad news for everyone because the total opportunity for value creation will be orders of magnitude less.
• This means lower long-term growth for the entirety of the technology sector unless there is a rapprochement of some sort.
• At the moment, this is a very distant and dim hope.