Telegram group chat for paid/trial subscribers

Get daily updates on Value Investing, ASEAN stock markets & Global Macro on Telegram (link at the bottom of this article)!

Invite link to join the Telegram group chat is located at the bottom of this article.

In light of adding even more value to my paid subscribers, I have created a new Telegram group chat exclusively for paid subscribers. This includes both trial and fully-paid subscribers… so if you’d like to see what we have to offer, hop onto a trial real quick!

Click this button to sign up for a free 30-day trial

It is my intention to cultivate this group as a home for value investors who are interested in investing the Warren Buffett way in ASEAN - the next China. While there aren’t any limits around discussion topics, the vision is to stimulate discussion around three main topics: 1) Value Investing, 2) ASEAN stocks, and 3) global macro. So while everyone is free to discuss Apple or Tencent, you’re also probably not going to find a lot of discussion about them here.

Here’s how I would describe the main benefits of joining a private group chat like this:

• Personalized Feedback - You can ask personalized questions regarding any topic I’ve discussed on Value Investing Substack before, be it about stocks or macroeconomics.

• Swifter Updates - As I can respond via text much, much faster than I can respond to a comment, you can get an answer to your particular concern in a far more expedited manner. This can be especially helpful for those of you who happen to be living outside of ASEAN - who may benefit from receiving on-the-ground updates on regional macroeconomic developments as they happen in real-time.

• Broader Research Scope - While I’ve mostly only written about the most interesting things happening in this part of the world so far, my actual research scope actually expands far beyond what ends up being published on my blog. If you want to know my thoughts on more granular stuff which may impact markets here - e.g. the progress of the currently ongoing elections in the Philippines, how the political winds are blowing in Malaysia, or the latest developments in Myanmar - I’ll be happy to share what I know.

• Make New Friends - One great thing about being value investors is that our friendships know no borders. I’ve had the immense privilege of making friends with so many people from different countries all across the world - simply by virtue of the fact that we all identify as value investors. Hopefully, this Telegram group can help foster such new friendships as well.

For a glimpse of the quality of the discussions that we have in the Telegram group chat, here’s one of my responses when a subscriber sought clarification over Intel’s past decade of underperformance (in the context of my AEM Holdings stock reports):

(…) As all of you are familiar with, Intel has been underperforming its peers both the likes of AMD (in the fabless space) and TSMC (in the fab space). Let’s talk about the fabless space first. As I’ve described in my article, the subsector involved here is the Chip Design subsector. So what they do is they only design the chip, and then outsource manufacturing to the fabs like TSMC.

What AMD did prior to 2016 when their share price began skyrocketing was introduce what is known as “chiplet” technology at commercial scale. This idea was introduced by the relatively recently appointed CEO, Lisa Su, who joined AMD in 2014. Basically, the idea behind chiplets is to make “normal” chips modular – i.e. rather than having all CPU functions on one die, you divide the CPU functions into separate dies and mix-and-match them as necessary. This gives you 1) yield advantages as they are smaller (dies per wafer) and also 2) allowed different arrangements of chiplets vs one monolithic die, which enabled higher performance. This culminated in the Zen architecture, which was miles ahead of Intel’s approach at the time. If you want to know more, I highly recommend reading this phenomenal article which paints AMD’s history from beginning to the recent past: https://www.techspot.com/article/2043-amd-rise-fall-revival-history/

Intel had fallen behind because it was stuck on the Broadwell architecture. If you might recall, there was a period of about 4-5 years in the mid-2010’s where Intel’s CPUs did not increase their clock speed (e.g. 2 Ghz dual-core). This was mainly because 1) they were still relying on the old monolithic die approach (i.e. non-chiplet), and 2) they were hitting issues on the Fabrication side with achieving smaller process node (e.g. 10nm). We’ve already covered the 1st issue, so let’s discuss the 2nd one.

As you might know, TSMC only really begin using EUV technology for manufacturing chips at scale around 2018. Prior to that, the main approach was called lithography, which is basically using a regular laser as compared to EUV light. The main difference here is that lithography techniques rely on laser at 193 nm wavelength, while EUV rays have 13 nm wavelength – so lithography is basically a lot fatter than EUV, and therefore can’t etch higher transistor designs. The way Intel tried to overcome this was through a process called “multi-patterning”. As the name implies, multi-patterning involves etching transistor designs onto the semiconductor wafer multiple times – with each pass being “in between” the first pass to simulate a laser of smaller wavelengths. So you might etch half of the transistor design one time on a semiconductor wafer, then etch the second half of it in between the gaps of the first half. As you can imagine, this was relatively inefficient (i.e. lower yields) due to the extra time requirement; and also involved higher complexity and thus higher chance for errors.

So why didn’t Intel use EUV earlier to solve this stumbling block? Recall that TSMC (who AMD outsources manufacturing to) started using EUV at scale around 2018, while Intel only started using it several process cycles later around 2020-2021 – which is an eternity in the semiconductor world. The main reason for this is because EUV was still highly experimental at the time AMD started leapfrogging Intel in the mid-2010’s, and Intel didn’t want to take the risk of an experimental approach, and decided instead to continue relying on multi-patterning. In contrast, TSMC made a huge bet on EUV and funded ASML’s research into the technology in the face of high levels of uncertainty whether they would even complete the R&D according to expected timelines. This was mainly because ASML’s engineers were having trouble discovering the current process of creating EUV rays – by pulsing two lasers at a falling drop of tin to output a sufficiently strong EUV ray. They were actually quite lucky to have discovered it at the time that they did; because otherwise the process node increase would have to be delayed yet again, and the industry would have to fallback to multi-patterning.

Recall also that at the time, the readiness timeline for EUV at commercial scale had already been delayed and pushed back several times – so Intel’s decision not to bet the horse on EUV was actually quite rational. TSMC took a risk, and because it was generously funding ASML’s research, they got first dibs on orders. Since ASML only makes about 30 machines per year, Intel had to wait in line in order to get their hands on them.

Okay, so this is the history of why Intel paled in comparison to AMD and TSMC, and underperformed both the fabless and fab side of the industry over the past half-decade. Now the question becomes, can Intel overcome these hurdles going forward in light of its past failures?

Fortunately, the answer is a very affirmative yes. Firstly, on the technology side, Intel is already well on the way to catching up with TSMC, as it now has similar access to ASML’s EUV equipment. If you’ve read my article, you’ll also know that the US Govt has a huge vested interest to see Intel succeed at its IFS enterprise – so you can only imagine how much lobbying is going on right now to accelerate that timeline. Intel has also already copied most of AMD’s current design processes (e.g. chiplets with Foveros, vertical stacking, etc), so it shouldn’t be too far behind AMD right now. The only real threat on the horizon is Apple, which is a totally separate topic as long as this one.

Secondly, on the culture side, Intel’s new CEO Pat Gelsinger is somewhat of a legend in the sector – he’s literally been called the “Steve Jobs of the semiconductor industry”. With his stature he brings a lot of credibility to Intel’s revival, and has already been successful at poaching several star figures in the sector from rivals since he joined last year (where previously the reverse was true). I won’t go into the details, but imagine if Steve Jobs joined Dell – that’s kinda the analogy we’re drawing here. Or another analogy is Nadella’s joining Microsoft - where previously Microsoft’s corporate culture was characterized as fiefdoms protecting their own turfs under Ballmer, Nadella apparently has managed to harmonize them into one focused group. And Microsoft’s share price since Nadella became CEO in 2015 reflects that cultural turnaround success – something I’m expecting Gelsinger to do at Intel in the coming years as well. If you’d like more granular detail on the cultural revolution at Intel thus far, here’s a very good piece which discusses it at length: https://semianalysis.com/intel-is-throwing-the-kitchen-sink-but-is-the-turn-around-plan-reasonable-deep-dive-on-tower-semiconductor-fabs-and-ip-intel-culture-shift-future-product-and-roadmap-competitiveness-by-business/

If you think discussions like this add value to your investment process, do feel free to sign up for a free 30-day trial - and join our Telegram group chat to see what else we have to offer!

Click this button to sign up for a free 30-day trial

If you’re a paid subscriber, please click the invite link below to join this exclusive Telegram group!