How is TSMC Still Growing?
Thoughts on TSMC Q3 2022 Earnings
Before we talk about TSMC, I want to share with you that I am doing a little bit of podcasting work. I am serving as a guest host for a few episodes in the Startup Island Taiwan podcast. In it, I interview members of Taiwan’s startup community.
The first episode just came out, featuring my interview with Clément H. Dieudonné of Ubiik. Go check it out. I think it’s a good way to get exposure to a startup ecosystem that most people aren’t really looking at.
Alright, let’s talk about TSMC.
Are there really words to describe what TSMC has done over the past few years? It took five years for TSMC to add $9 billion in revenue, 2014 to 2019. And then they added nearly $11 billion in revenue in 2020.
Amazingly enough, they did it again in 2021. And with their Q4 guidance being at about $20 billion, then they are going to do something like $76 billion in revenue in 2022.
As Doug from Fabricated Knowledge points out, over half of TSMC’s revenue comes from its two leading edge nodes, N7 and N5. TSMC is a bit top-heavy right now, but not unusually so.
Apple is a major contributor to this effect. They signed on in 2014, and make up some 25% of TSMC revenue. Which implies that Apple alone is the size of AMD, Qualcomm, Nvidia, and all other leading edge node users put together.
Gross margins however are higher than they have ever been before.
Hitting 60% gross margins is quite high. And I have been wondering why these margins have been rising so much. This is what TSMC cites as drivers for their gross margins:
As a reminder, 6 factors determine TSMC's profitability, leadership technology development and ramp-up, pricing, cost reduction, capacity utilization, technology mix and foreign exchange rates. Looking ahead to 2023. We faced challenges from N3 ramp dilution, higher year-over-year increase in depreciation costs, rising inflationary costs, semiconductor cyclicality and overseas fab expansions.
We also know that TSMC raised prices. There was a flurry of news in Taiwan media about price increases. First a piece in Sept 27 about Apple “resisting price increases” - apparently ranging like 6 to 9% and then another piece a week later about Apple “accepting the price increases”.
Apple ultimately accepted the price increases. Not like they had a real choice. TSMC tends to really hold the line when it comes to prices. If Apple breaks the line, then Nvidia and all the others do too. That is the kind of stuff that gets foundry CEOs fired.
There is a piece in Nikkei Asia saying that the A16 Bionic, which uses a N4 process, costs 2.4 times more than the A15 used last year.
The higher production cost is mainly due to the A16 Bionic chips used in the iPhone 14 Pro and Pro Max models. The proprietary chip costs $110 -- over 2.4 times more than the A15 version used in the iPhone 13 Pro Max released last year.
Does N4 really warrant a near 250% price increase over N5? It isn’t a full node step. TSMC showed quarter over quarter smartphone revenue growth of 25%. Smartphone is largely Apple. I feel like if N4 cost 250% more than N5, it would show up right? Something is weird about that one.
Alternatively, if we take this piece of information at face value, how much more will N3 cost?
Perhaps these high gross margins are due to TSMC simply getting better and better at N5 and N7. N5 has been out for 3 years now. They are probably really good at it right now, which is why they are making so much on it. A new node dilutes gross margin as when it first starts and as it ramps up. There is another side to that where a leading edge node collects massive revenue, but gets steadily more profitable as TSMC optimizes and optimizes it.
N3 & N7
Back in August, I read a piece in Taiwanese media that TSMC would start N3 high volume production in September. I waited through September for any further indication, and the call makes it seem like that N3 has still not hit high volume.
Now I will talk about our N3 and N3E status. Our N3 is on track for volume production later this quarter with good yield. We expect a smooth ramp in 2023, driven by both HPC and smartphone applications.
TSMC’s revenues are pretty dependent on its leading edge nodes. So it is important to maintain market leadership by onboarding new nodes. There are still 2 more months for N3 to ramp. I don’t doubt that it will do well once it gets started, but things are getting a bit tight.
People noticed that N7 went through a bit of a decline this quarter, going from 30% of revenue to 26%. Management said that this was cyclical and not structural.
It just happened that most of my smartphone and PC customers are using N7 and N6 node. And it just happened, the market weakness in the smartphone and PC happening at the same time.
I would still keep an eye on it though.
The China Thing
When asked about the recent moves by the Biden administration to restrict Chinese chipmaking, CC Wei said:
Okay, let me answer that. Bruce, it's based on our initial reading and feedback from our customers. The new regulation set the control threshold at very high-end specification, which is primarily used for AI or supercomputing applications. Therefore, our initial assessment is the impact to TSMC is limited and manageable.
There has been a lot of work done by Dylan and Doug on this. I don’t really have anything to add right now except I think China is going to get a lot better at semiconductor manufacturing after this - which I reckon is going to spill a lot of ink and headlines.
But at some point their progress will slow. I just wonder where that point is going to be.
The Industry
TSMC believes that the semiconductor industry - moving in line with the economy - is going to take a bath next year. Yet despite being 50% of the entire foundry industry, and all these other electronics companies reporting difficult earnings … yet they are going to still grow.
We expect probably 2023, the semiconductor industry will likely to decline. But TSMC also is not immune, but we believe our technology position, strong portfolio in HPC and longer-term strategic relationship with customer will enable our business to be more resilient than the overall semiconductor industry. And that's why we say in 2023, still a growth year for TSMC and the overall industry probably will decline.
I was thinking about this. TSMC is still adding a whole lot of capacity in the coming years. There are new fabs and expansions across Taiwan and the United States. The $35 billion plus of capital expenditures will bring returns in the next 1-2 years.
At the start of this year, Sebastian Hou pointed out that TSMC’s CapEx gives a hint at its revenues the next year - about 3 times ratio. Now, this ratio broke a bit this year. If TSMC hits the $76 billion revenue this year they are predicting, that is a 2.89 ratio from last year’s $26 billion CapEx.
But even with that and even with the CapEx spend down from $40 billion to $36 billion, that implies TSMC does in the neighborhood of $100 billion revenue in 2023. Hmmm.
Who is buying all these chips? I was a recent guest on the Analyse Asia podcast and near the end, I mused, “We don’t really know why they are growing so fast”.
I am still mulling it over.
HPC is server, AI, super computing, etc. Are the hyperscalers still adding that much capacity now that the pandemic is (largely) over? Are the ARM cloud computing chips really taking that much x86 market share? Is AI that big of a deal that it is going to suddenly make up for the dump in graphics cards from crypto?
I don’t know. I haven’t yet seen any solid data that makes me stand up and say, “That’s it!”
Perhaps it is Apple again. HPC presumably includes whatever Apple is bringing over from Intel as it transitions over. If so, then the growth story is done because the transition is also done.
I wonder if there is something in the packaging story. I know that TSMC has been dipping its toes more into packaging as a value-add. I know that a number of the new factories they are building in Taiwan are for packaging. I should look into that more.
I guess I just haven’t wrapped my head around it. So long as I feel that I can’t get a grasp on it, it still kind of makes me wondering how sustainable it is especially as the world economy continues to deteriorate.
I hear that Taiwan has lifted its quarantine for visitors. I am not sure how I am going to handle having tourists now after 3 years of peace and quiet, but I think Taipei is a tourist gem and everyone should visit. I will do a listicle some time of food places I personally have liked and enjoyed.