What We’re Reading (Week Ending 04 February 2024)

What We’re Reading (Week Ending 04 February 2024) - 04 Feb 2024

Posted at 07:45h in What We're Reading by Ser Jing & Jeremy 0 Comments

0 Likes

Reading helps us learn about the world and it is a really important aspect of investing. The late Charlie Munger even went so far as to say that “I don’t think you can get to be a really good investor over a broad range without doing a massive amount of reading.” We (the co-founders of Compounder Fund) read widely across a range of topics, including investing, business, technology, and the world in general. We want to regularly share the best articles we’ve come across recently. Here they are (for the week ending 04 February 2024):

1. China Wants To Ditch The Dollar – Zongyuan Zoe Liu

Ganzhou also hosts the Ganzhou Rare Metal Exchange, where China’s renminbi currency is used to quote prices for spot trading of tungsten, rare earth products and critical minerals like cobalt that are essential to the clean energy transition.

The metal exchange, established in 2019 with the approval of the State Council, now operates as a subsidiary of China Rare Earth Group. It is China’s second mineral exchange, which was established to use the renminbi to price and trade minerals and rare earth products.

The first such exchange, the Baotou Rare Earth Products Exchange, which started operating in 2014, is jointly owned by 14 major Chinese rare earth suppliers and was explicitly set up, at least in part, to increase China’s overall role in pricing rare earth products. To that end, China also launched two renminbi-denominated exchanges — oil futures in 2018 and copper futures in 2020 — on the Shanghai International Energy Exchange.

By establishing commodities exchanges across its industrial cities, China aims to boost the use and power of the renminbi in global commodities pricing to establish an alternative global financial system that is less reliant on the almighty dollar. This effort also involves regional cooperation with China’s neighbors and non-Western multilateral partnerships to develop regional currency arrangements and enhance the use of local currencies in international trade and investment.

In China’s telling, these strategies are less about offense — trying to dethrone the U.S. dollar or replacing it in the global system with the renminbi — and more about defense: strengthening China’s financial security and reducing its geo-economic vulnerabilities within the existing dollar-dominated global economic and financial system. Beijing wants to minimize its exposure to a potential dollar liquidity crunch and ensure its continued access to global capital markets even during times of geopolitical crisis.

No Chinese leaders have publicly expressed an intention to dethrone the dollar despite escalating geopolitical and trade tensions between the U.S. and China beginning in 2018. However, as those tensions persist, Chinese financial regulators and scholars have explicitly expressed concerns about Beijing’s vulnerabilities and urged government officials to step up efforts to protect the financial system…

…Since President Xi Jinping came to power in 2013, he has repeatedly emphasized worst-case scenario thinking to “prevent macro-risks that may delay or interrupt the process of the great rejuvenation of the Chinese nation.”

From Xi’s vantage point, China’s state-owned financial institutions and enterprises must inoculate themselves against potential international sanctions in the event of a military conflict with the West over Taiwan. That concern has only grown more urgent after China witnessed the collective sanctions imposed by the West on Russian entities and individuals to punish President Vladimir Putin for his war against Ukraine.

The West’s decision to freeze Russian foreign exchange reserves has caused particular consternation in Chinese policy circles. Chinese economist Yu Yongding described such a move as “a blatant breach of…trust” and proof of the United States’ “willingness to stop playing by the rules.”…

…At the September 2022 SCO Summit, Xi explicitly proposed expanding the use of local currencies in trade settlement to promote regional integration, strengthening the development of local-currency cross-border payment and settlement systems and promoting the establishment of an SCO development bank to help shepherd such changes. SCO members agreed on a “roadmap” to accomplish these goals.

In a December 2022 address to the China-Gulf Cooperation Council (GCC) Summit, Xi emphasized his hope that China and members of the GCC should increase the use of renminbi for oil and natural gas trading and settlement through the Shanghai Petroleum and Natural Gas Exchange (SHPGX) in the next three to five years.

Since Xi’s speech, Chinese national oil and gas companies have accelerated initiatives to use the renminbi, instead of the U.S. dollar, in their international fossil fuels transactions through SHPGX. In March 2023, an important step towards the de-dollarization of energy trading occurred when China National Offshore Oil Corporation — known as CNOOC, China’s largest offshore oil and gas field operator — used the renminbi to complete the transaction of importing 65,000 metric tons of liquefied natural gas (LNG) from TotalEnergies SE, a French multinational oil and gas company, through SHPGX. The LNG was produced in the United Arab Emirates, a member of the GCC, carried by a Liberian-flagged LNG tanker Mraweh, and finished unloading in May at the CNOOC Guangdong Dapeng LNG receiving station.

This transaction was the world’s first cross-border LNG trade settled using the renminbi. Since then, CNOOC has executed more renminbi-settled transactions using the renminbi through SHPGX. In October, PetroChina, the largest oil and gas producer and distributor in China, settled a purchase of one million barrels of crude oil using the digital renminbi through SHPGX, marking the first cross-border oil transaction using the country’s central bank’s digital currency…

…Among SCO members, China has since signed bilateral currency swap agreements with Uzbekistan, Kazakhstan, Russia, Tajikistan and Pakistan. China also now has swap agreements with SCO observer and dialogue partner countries Mongolia, Turkey and Armenia; last March, Saudi Arabia committed to joining as a dialogue partner and a full member in the near future. While China doesn’t have an agreement with Kyrgyzstan, which is in the SCO, Kyrgyzstan’s national bank signed a letter of intent in September 2015, stating it aims to work with the PBoC toward establishing a bilateral currency swap.

China’s support for the expansion of SCO and BRICS over the last two years to include major commodities-exporting countries like Iran, Saudi Arabia, the United Arab Emirates, among others, suggests it is eyeing new opportunities to accelerate renminbi use in commodities trading.

The expansion has also given SCO and BRICS added significance as political forces in the shaping of commodity markets. SCO members include major hydrocarbon and minerals exporters in Central Asia like Kazakhstan and Uzbekistan, Russia and its newest member as of last year, Iran.

SCO also includes major commodities importers like China and India. In this context, as a non-Western group of countries, SCO potentially represents a potent coalition of exporters and importers of commodities centered around using the renminbi to finance the entire commodities lifecycle from production to trade to consumption…

…Chinese economists have argued that the ultimate goal of renminbi internationalization should be to have central banks and major international financial institutions worldwide willingly hold large amounts of renminbi for international transactions so that China’s currency can become an international reserve currency alongside the U.S. dollar and the euro.

Since then, the Chinese government has put resources into developing a renminbi-based financial infrastructure for cross-border settlement. In 2015, it launched the Cross-Border Interbank Payment System (CIPS) to improve the convenience of using the renminbi in international transactions by providing onshore renminbi clearance and settlement services.

CIPS allows global banks to clear cross-border renminbi transactions onshore instead of through offshore renminbi clearing banks, providing a one-stop alternative to the combination of the SWIFT system — a secure messaging system used by major banks to send financial information to one another — and the New York-based Clearing House Interbank Payments System.

However, CIPS is not a complete departure from SWIFT and still uses SWIFT’s standards to connect with the global system. It has adopted the ISO 20022 international payments messaging standard to be interoperable with other payment systems as well as with correspondent banks around the world.

By adopting existing cross-border messaging standards, China aims to make CIPS a critical piece of the world’s existing financial infrastructure to promote international use of the renminbi. By 2023, CIPS’s annual business transaction volume reached 123 trillion renminbi (roughly $17.3 trillion), according to data on the CIPS website. CIPS now has 139 direct participants and 1,345 indirect participants worldwide, most of which are foreign branches of Chinese banks.

The Chinese government has also used subtle but strategic initiatives to increase the global appeal of its currency and deepen the market depth of renminbi-denominated assets. Despite hesitations around liberalizing China’s capital account to allow capital to move freely in and out of the country, Chinese authorities have worked to broaden international acceptance of renminbi bonds as collateral.

In March 2021, the International Swaps and Derivatives Association (ISDA), a New York-based group composed primarily of the world’s largest banks, together with the China Central Depository and Clearing Corporation, the Beijing-based central depository for all Chinese government bonds, released a whitepaper detailing the usage of Chinese government bonds as an initial margin in derivatives contracts.

This past September, the Hong Kong Exchanges and Clearing (HKEX) and London Stock Exchange started to study the use of Chinese government bonds as eligible collateral for derivatives contracts as a way to reduce Asia’s heavy reliance on cash for margins on derivatives trades. Chinese institutions have also teamed up with leading resource-rich economies to make renminbi-denominated assets more attractive for international investors…

…China’s promotion of an alternative financial system is not about cheering on the demise of the U.S. dollar, but rather about creating an alternative financial system without a dominant currency in which the renminbi is accepted without bias. China has a strong incentive to prevent the dollar’s collapse because it would likely be the largest financial loser should the dollar depreciate. The majority of China’s over $3 trillion in foreign exchange reserves are invested in U.S. bonds and the lion’s share of Chinese sovereign fund portfolios are tied to dollar-based Western markets.

2. TIP602: Same As Ever w/ Morgan Housel – Clay Finck and Morgan Housel

[00:09:27] Clay Finck: He states, risk is what’s left over after you’ve thought of everything. And I just absolutely love this chapter. It’s Everyone wants to know what’s going to happen. What’s the stock market going to do? Interest rates, the Fed. You state the biggest risk and the most important news story of the next 10 years will be something nobody’s talking about today.

[00:09:45] Clay Finck: No matter what year you’re reading this book, that truth will remain.

[00:09:49] Morgan Housel: Yeah one way I think about this is I wrote Psychology of Money, my first book. I wrote most of it in late 2019. So obviously that was weeks or months from COVID completely throwing our life upside down, everybody’s life upside down.

[00:10:02] Morgan Housel: And I and everybody else had no clue about it. We were completely oblivious to what was going on. Staring at us in the face of that at that point, and I think what you can say, what is the biggest news story globally of 2023 of last year? It was, I think most people would say it was Israel Hamas, which is another thing.

[00:10:17] Morgan Housel: If you go to January of 2023, nobody was talking about that. No one was putting that on their radar. Even the day before it happened, virtually no one was talking about it, thinking about it, forecasting it. So it’s always been like that. You can say that for this year. The biggest news story of 2024 is something that you and I are not talking about today that we cannot see coming.

[00:10:35] Morgan Housel: Someone, I just saw this on Twitter just a couple hours ago. I thought it was really good. I’m paraphrasing it, but it was like, if you are making a decision tree or like a list of probabilities and you say, there’s a 20 percent chance of this happening and a 30 percent chance of that happening. If you’re just going through probabilities like that seems like a smart thing to do.

[00:10:51] Morgan Housel: But if all of your probabilities add up to a hundred, then you’re doing it wrong. Because what you are implicitly saying is that every potential possible outcome that there’s going to be. So I think the best you can do in any of these is if your known probabilities and you can think of should add up to 80 or something like that.

[00:11:07] Morgan Housel: Maybe it’s 90. You should always, you always have to leave a percentage chance for something could happen that I cannot even fathom. That I can’t even, no matter how creative I try to get, there can be a risk out there that I cannot even envision. And of course you should do that because that’s how it’s always been.

[00:11:21] Morgan Housel: The biggest news stories of modern times are things like the Great Depression, Pearl Harbor, World War II, 9 11, Lehman Brothers going bankrupt, COVID of course. And the common denominator of all of those is that you could not have seen them coming, at least in their specific nature of how they arrived and what they did, until they happened.

[00:11:37] Morgan Housel: And so it’s always going to be like that. It’s very uncomfortable to come to terms with that, to come to terms with how uncertain and unpredictable the world can be. But I think if you study history, you can’t come to any other conclusion.

[00:11:48] Clay Finck: What I find so fascinating about this is the biggest sort of disasters are those that no one expects, no one forecasted, no one projected.

[00:11:57] Clay Finck: You mentioned in your book that it seems that zero economists predicted the Great Depression. It’s no wonder it was so bad. No one was prepared for it. No one expected it to come. And COVID is very similar. And you’ve lived through the great financial crisis, likely an investor at that time.

[00:12:13] Clay Finck: Did it feel like no one was saw it coming and it was just a total disaster and much worse than anyone could have ever imagined.

[00:12:20] Morgan Housel: See, that’s a little bit different. That’s different than 9 11 because as recently, as re as early as 2003, there were people who are ringing alarm bells about how fragile the economy was and over leverage and whatnot.

[00:12:32] Morgan Housel: So that it’s not, nobody saw it coming. That’s not quite, that’s not quite true, but a lot of the people who quote unquote, saw it coming. When it did happen, it happened for reasons that they could not fathom. So for example, a lot of people, I won’t name names, but as in 2005, 6, 7, they said a giant recession is coming and it’s going to be caused by hyper, it’s going to lead to hyperinflation and interest rates are going to go to double digits.

[00:12:53] Morgan Housel: The exact opposite happened. So what do you do in that situation where they saw trouble coming, but it happened for the exact opposite reason than they saw it coming? That they envisioned. It’s there’s all these weird nuances there where it’s not black and white. There’s also what really sent the financial crisis into hyperdrive was Lehman Brothers going bankrupt.

[00:13:09] Morgan Housel: But there’s all these alternative histories of a lot of people forget that as Lehman Brothers is going down, Barclays was like hours away from buying it. And that deal fell through and Lehman Brothers went bankrupt. But there’s this alternative history of what if Barclays had bought Lehman Brothers and we escaped all of that.

[00:13:23] Morgan Housel: And the economy just zoomed to recovery after that. There’s all these different possibilities. And I think the takeaway from that is you couldn’t have seen it coming. Even if you saw trouble, you saw brewing in the financial crisis. Nobody in their right mind could have known exactly how it was going to play out.

[00:13:35] Morgan Housel: And I think that’s true, even not only during, but after the financial crisis, it was so common if you were an investor in 2009. To say, look, stocks are still overvalued. We’re in the quote unquote, new normal of low growth. That was a phrase that was always thrown around. The CAPE ratio is still just still too high.

[00:13:52] Morgan Housel: Expect lower returns. That was what virtually everybody was saying. I’m not going to say everybody. Of course, there are some people who saw it differently, but that was the very common narrative. And it made sense. If people were saying that you’re like, yeah, that makes a lot of sense. But what happened?

[00:14:04] Morgan Housel: The stock market tripled over the next three years. It was a, ended up being like the best three year period to be an investor in modern times. And so that, that’s a very common story throughout history too, is that the narrative at the moment, that makes sense that the majority of people cling to in hindsight looks ridiculous.

[00:14:19] Morgan Housel: And so we see that a lot across and it’ll be like that going forward whenever the next recession is, of course…

…[00:39:37] Clay Finck: I also wanted to tie in here, the emotional side of investing. The Buffett quote is be greedy when others are fearful and fearful when others are greedy. But you talk about in your book how this is much easier.

[00:39:51] Clay Finck: said than done. And it’s just so hard to put ourselves mentally fast forward in that type of situation when stocks have fallen, it’s the time to buy. And another problem is that when stocks fall, there’s usually a good reason why they’re falling. And I go back to March, 2020, and I had friends calling me at work, telling me how much money they’re making by shorting the market.

[00:40:10] Clay Finck: And yeah, this coronavirus going around and just the emotions just flood in. And it’s just so hard to act rational when those emotions are at play. You write in your book, hard times make people do and think things they’d never imagine when things are calm. And March 2020 was the complete opposite of calm.

[00:40:29] Clay Finck: Talk to us about how our views and goals can quickly change when our environment’s changing.

[00:40:35] Morgan Housel: I think if I today, right now, when the economy and the stock market are pretty strong and prosperous, if I said, Clay, how would you feel if the market fell 30%? Most people would say, I’d view that as an opportunity.

[00:40:45] Morgan Housel: That’d be great. The stocks that I love would be cheaper. I’d be a buying opportunity. That’d be great. Okay. And for some people that really is the case. But then if I said, Hey, Clay, the market falls 30 percent because there’s a pandemic that might kill you and your family and your kids school to shut down and you have to work from home and the government’s a mess, it’s going to run a 6 trillion deficit to try to figure this out.

[00:41:04] Morgan Housel: How do you feel in that situation? You might be like, most people will say, Oh, in that world, or once they experienced that world feels very different. Or if I said, Hey, the market fell 30 percent because there was a terrorist attack on nine 11. And all the experts think that was just scratching the surface of what’s to come.

[00:41:19] Morgan Housel: Do you feel bullish now? A lot of people will say no, they don’t feel. So once you add in the context of why the market fell, most people will realize that it’s much easier to quote Buffett than it is to actually be somebody like Buffett. I experienced this myself. I had some of the smartest people who I knew in March and April of 2020.

[00:41:35] Morgan Housel: Some of the conver I remember two specific conversations. One was somebody who said, Hey, look, there’s about 2 trillion of capital in the entire banking industry. You do not need to be creative to imagine how all of that’s going to be wiped out. The entire capital of the entire banking industry is going to be wiped out.

[00:41:48] Morgan Housel: And I remember thinking that and being like, yeah, no, you don’t. If the entire economy is shut down for three months, all of that capital is gone. The entire banking sector is insolvent. Obviously that did not happen. But when I heard that, I was like, no, that actually makes sense. I don’t know if that’s the base case scenario, but that’s not far fetched.

[00:42:03] Morgan Housel: I also remember during that period of COVID when it was like, no one’s really going to be making their mortgage payments when everyone is on lockdown that people are like, look, the entire non banking lending sector, non bank lending mortgage sector is all going to collapse. And that’s 80 percent of the originations market.

[00:42:17] Morgan Housel: 80 percent of mortgage originations are going to be out of business in two weeks. And I remember piecing that together and be like, that makes sense too. That didn’t happen either. But during this period, Which was in hindsight, and even at the time, you could have seen look, this is the opportunity of a lifetime.

[00:42:30] Morgan Housel: The market fell 50 percent in a short period of time. This is gonna be a great opportunity. When you add in the context, both the health consequences and the potential economic consequences. It’s a much different situation. I would even say to finish this up, this is maybe the most important part about, I think it was in early February, 2020, Warren Buffett went on CNBC and they’re talking about, Hey, there’s all these rumbles about a virus.

[00:42:51] Morgan Housel: Like, where do you get it? Like the market’s starting to fall. What’s going on? And Buffett said, I don’t want to, I’m paraphrasing here. This is not a direct quote, but He said, I don’t know how I’m going to invest in the next month, but I guarantee you, I’m not going to be selling. That’s what he said.

[00:43:03] Morgan Housel: Two weeks later, he dumped every airline stock that he owned. So even in this situation, somebody like Buffett, the originator of the phrase be greedy when others are fearful. When he added in the context of what was happening to the airline industry during the lockdown, he determined, and I think in hindsight, it was probably the right decision to sell those stocks.

[00:43:19] Morgan Housel: And some people have pointed out too that part of the reason that he sold him is that the government could not have bailed out the airlines if he was the largest shareholder, so people asked him to sell those stocks. It is a complicated thing, but once you add in the context of why the market’s falling, most people realize that their risk tolerance is actually much less than they thought.

3. What I Learned When I Stopped Watching the Stock Market – Jason Zweig

I’m back at my regular post at The Wall Street Journal after being away on book leave. That long hiatus disengaged me from the daily hubbub of markets so I could frame investing ideas in a longer historical and broader psychological perspective…

…When my last regular column ran last May 26, the S&P 500 was already up 10.3% in 2023—right in line with the long-term average annual return of U.S. stocks. “Let’s just call it a year right here,” I recall muttering to myself.

That was the last thing I remember. From that day to this week, I tuned out the daily noise of fluctuations in stocks, bonds, commodities and economic indicators…

…It’s a good thing the market gods ignored me, as they always do. Even though I thought a 10.3% return in five months was plenty for an entire year, the S&P 500 finished 2023 up more than 26%, including dividends.

When you don’t watch the market every day, you can finally see with unquestionable clarity that what you would have expected to happen didn’t. The unexpected did.

Had you told me war would break out in the Middle East in October and last for months, I would have been sad but unsurprised. Had you added that crude oil would—after a fleeting surge—finish 2023 at a lower price than the day I left, I would have been amazed…

…You probably can’t disappear for seven months, but you can pretend you did. Hal Hershfield, a psychologist at the University of California, Los Angeles and author of “Your Future Self: How to Make Tomorrow Better Today,” urges investors to “use the tools of mental time travel to escape the tyranny of the present.”

He means that envisioning how you will feel about your actions tomorrow can help prevent you from overreacting today…

…For general templates of such letters, see the “Future Self Tool” at consumerfinance.gov.

Research suggests this technique can help you avoid making decisions you might later regret—and can reduce the anxiety stirred up by negative news.

I’ve long thought financial advisers should encourage this approach to help clients make deliberate and durable decisions. Now I think it’s worth trying on yourself, too.

4. An Interview with Arm CEO Rene Haas – Ben Thompson and Rene Haas

RH: Yeah, two things are key. It’s very, very small, it takes longer to fabricate. So what may have taken you 12 weeks to put a product into production now may take you 26 weeks. That’s a big, big jump in terms of the lead required.

Then you look at these SoCs that are using Arm, back in the day, if we were putting 12 to 16 CPUs into an SoC, that was considered a lot. You now look at some of the recent chips being announced, just look at the Microsoft Cobalt, their recent CPU that they announced using Arm, 128 CPU cores into that SoC. That is a lot of work for someone building an SoC to figure out how those 128 CPUs are going to work together. What’s the cache coherent network look like? What does the interconnect look like? What does the mesh look like?

So what we felt was if we can provide more of that solution, i.e., stitching together all of the system IP, the GPU, the CPU, an NPU, anything around the processor fabric, we could fundamentally allow the customers to get to market much, much faster.

So we started this initiative towards what we call compute subsystems, which was really about developing the overall platform, which not only helps us in terms of getting an SoC to market faster, but it also allows us to work more quickly in terms of the software ecosystem. We can start to think about what gets product in the hands of developers sooner, or what gets the products in the hands of people who are developing the application software, people who are doing the OS work.

So for a myriad of reasons, it just made a ton of sense for us to go up and do that and we’ve started that with our hyperscalers, with our cloud compute, but we see it applicable to almost all the markets, whether it’s a cell phone, whether it is a laptop, whether it is an automotive ADAS [Advanced Driver-Assistance System] system. The same rules of apply, these are complex compute subsystems. The chips take a long time to build. If you can shave off any amount of development time that helps the people get the chips out faster, that’s huge value. What we are seeing is in some cases where it may have taken two years to get a chip to tape out, we’ve cut that in half. One customer came back and said, “Look, you’ve saved us 80 man-years in terms of efforts.” So across the board, we’ve seen pretty strong validation that this is the right thing to do.

Just to get to the nuts and bolts a little bit, you mentioned the savings in terms of design time and things getting smaller and how long it takes to fab a chip. Is this also just a matter of, there’s a lot of reports about interference and stuff like that, particularly when you’re getting even down to 3 nanometers or 2 nanometers in particular. Is that a real driver as well? Would this opportunity be presenting itself absent the real challenges that are coming along in terms of smaller and smaller size chips and the increased design challenges that are coming with that, particularly around interference?

RH: You follow technology for a living, so you know this well. Like everything with these type of things, a number of things need to come together at once. When you have these long cycle times to build the chips, the complexity in closing timing loops, you’re trying to drive the maximum power efficiency, you’re trying to maximize the ultimate work you’re doing with the libraries. Again, with these subsystems, we will not only handle everything in terms of validation and verification, but we’ll do the tuning for the process. So if folks want to make sure they can get that ultimate last mile of performance, that’s just a lot of work that needs to be done that if Arm is doing it with a platform that we control — because it’s our IP, right?

Right.

RH: At the end of the day, it is around the compute subsystem and computer architecture that we’ve delivered, it’s highly beneficial. So again, in the old days you could kind of throw all this stuff over the wall and people could just pull it together and make it work, but the world has changed a lot in terms of just the complexity of these chips, and one thing that’s not relenting is people want to get products out fast. The markets will move really, really quickly and I think actually we’re seeing them moving even faster now. When you look what’s going on with generative AI and everything relative to these multimodal models and large language models, you have so many moving parts relative to what it takes to develop a product. It’s really, really critical to maximize on efficiency of time to market…

… I’m curious because on one respect, a lot of consolidation at one part of the value chain would potentially increase the opportunity for competition in others. You could see how consolidation would play well to say, RISC-V prospects in that regard, open source is as modular and open as you could get. On the other hand, is that sort of drive in a value chain of consolidation at one point, driving modulation the other, is that just overcome by the complexity involved, such that there’s rooms for multiple highly consolidated aspects of this chain? TSMC is pretty centralized as far as things go, and you just feel optimistic that at this point in time, the ecosystem from a software perspective, Arm is x86, 15 years ago on the PC side, and even if you theoretically want to do something different, there’s so much software to build, your lead is just going to be much larger.

RH: I think so, but it’s a very, very good question because it’s a little difficult I think to look backwards to kind of predict the future. One of the things we’re seeing around the future innovation that’s really a gate to innovation is this massive capital investment required, and it’s not just in building a chip, not just in building a fab, it’s not who has enough money to go off and buy new ASML EUV machines.

Let’s take a look, for example, at foundation models and everything going on with generative AI and training. Right now, Nvidia is an amazing position because of just the pure access to GPU technology and how expensive it is now, how scarce it is. That in and of itself, starts to lead to an area of, “Well, you’ve got all kinds of interesting open source models and people in the open source community working with things like Llama. But if people can’t get access to the GPUs and actually training, then who wins?” So, right now you’ve got the Big King.

We saw an excellent example of that in the last few months. He who controls the GPUs controls the world, at least for now.

RH: Correct. So, then when you start to think about — I like to think that Arm is an amazing place because any one of these application areas, we’ve got a huge, huge installed software base and we’ve got a very, very powerful position on power efficiency. So, when I think about where the puck is going relative to an alternative architecture, let’s say, you’ve got to look at either, “Do I have a 10x advantage in performance or a 10x advantage in terms of cost?”. And right now, I think in the areas where Arm is really good at, people would have to look at it really hard and say, “Is it worth the investment to go port everything I’ve got to an alternative architecture? What is the ultimate benefit that I get to the application space?”.

I think what’s really fascinating about everything going on with generative AI right now is I think you’re just seeing huge amount of resources coming into all kinds of development around training and inference, that will drive the growth here, so I think that’s actually where the growth is going to come. I think Arm is in a great place there. Obviously I’m biased but I think when you think about everything that’s going on with generative AI training, all those inference workloads are pretty good for the CPU, and history has sort of shown us that over time, as you add, and we saw this, whether it’s floating point heading into the CPU, or vector extensions, the CPU start to add more and more of the base functionality it allows with some of the workloads and I think you’ll see that in this space…

…Fast forward to where I am now, I don’t spend a lot of time when I talk to people inside of engineering or product groups about, “Hey, who’s catching us from behind?”, I try to think far more about where the world’s going to be in five to ten years. If you think about where the world’s going to be in five to ten years and you focus as much as you can, you’re not going to get it specifically right, obviously, but you want to be directionally investing in that area so when things land in your space, you’re going to be in a good spot.

Take case in point, predicting what the mobile phone is going to look like in 2034, ten years from now, and trying to make sure I do everything defensively to make sure that we’re in a great position is kind of nutty because if you go back to 2008 when the smartphone was invented, folks who were trying to think about protecting what the future phone looked like would have been out of position. Where I really focus on, Ben, is just where are things going and where do we need to invest?

Again, I know this AI drum that gets, people at times try to think, “Oh my God, how many times you got to hear the word ‘AI’?” — obviously, on one level AI is not new, anything that was going on relative to voice recognition or data translation, obviously that was all AI. I think AGI and everything around generative AI that can think and reason, that’s a pretty compelling place, and whether that takes place in five years, ten years, fifteen years, I don’t think anyone can argue that an investment in that space isn’t going to provide huge benefit down the road. I think Arm is a compute platform, I want to be sure that we’ve got everything correct from either an infrastructure standpoint, instructions at architecture standpoint, everything around the subsystem to be able to capture that.

Do you feel pretty confident? I think that you’re trying to sort of tie all of your stuff together to a greater extent, but companies could bring their own neural processor. Google obviously does that at a very sort of small-scale example, scales as far as terms of numbers, not scale in terms of importance of AI obviously, but is this really core to your thesis that as opposed to you needing to bring up a super competitive sort of NPU that, to your point, about the extensions and floating point, which I think is a great analogy, this is all going to be built into the CPU, so regardless, you’re going to need — even if the mobile phone market doesn’t grow just because of the number or it’s limited to this number of humans on earth, that is still going to be a significant opportunity or do you think you have to bring up additional separate IP? Or is this idea of it all being separate meaningless in the long run?

RH: It’s not going to be a one-size-fits-all kind of situation. Today, there’s a lot of investment going on training these very, very large models on highly networked GPUs. Even when you start talking about inference in the cloud, what matters is compute, but less around the interconnect between all of those systems, which is why CPUs over time in the cloud may find themselves to be very, very good solutions without having a GPU necessarily connected to it.You’re going to have a CPU in the cloud no matter what, so at some point it probably makes sense to consolidate.

RH: While Grace Hopper is a fantastic design from Nvidia, there’s a lot of people who I know are asking for, “Just give me Grace and don’t give me the Hopper when I go off and run inference.”…

…But when you see announcements like Microsoft Cobalt — after Graviton was announced, we had a lot of folks saying, “Well, they did it for their own reasons, and there’s not going to be much level of scale.” I would say continue to launch these very, very significant product announcements from company moving to the Arm architecture and I think you’ll see more and more of those over the next 12 to 18 months. And really, those are indicators, whether it’s in the automotive space, in the AI space, look for things, and I can’t tease this out too much, but my example of floating point instructions moving into the CPU, watch for those things on the AI front because that’ll tell you the direction of travel that says, “Yeah, this is moving that way.” I can assure you that we’re not going to stop doing these subsystems, and you’re going to see more and more announcements coming out on those.

5. A beginner’s guide to accounting fraud (and how to get away with it): Part IV – Leo Perry

In September 2011 Quindell (at the time trading as Quindell Portfolio PLC) acquired a business called Quindell Solutions Limited (QSL).

QSL had previously been a subsidiary of Quindell Limited, which in turn reverse merged into Mission Capital in May 2011, to form the then listed business called Quindell. All clear?

In 2009 Quindell sold QSL to its CEO Rob Terry in for a pound. Companies House filings show it had a slightly negative book value at the end of 2009 and again in 2010; sales, costs and cashflow were all nil. So QSL was an empty shell.

In 2011 Quindell re-acquired QSL for two hundred and fifty grand in cash (£251,000 to be precise). At the time it was wholly owned by Quob Park Limited, which was in turn wholly owned by Rob Terry. Quob Park’s previous name was Quindell Portfolio Limited.

If you’re still following, Rob and his wife Louise Terry were both Directors of Quindell (the PLC) and of QLS / Quob Park when the acquisition took place. But it wasn’t disclosed as a related party transaction. I mean this was AIM after all.

So it looks a lot like Quindell made a large undisclosed payment to its CEO, for an empty shell company that it had sold to the same CEO two and a half years earlier – for a pound. Perhaps, surely, there was some great innovation at QSL in the mean time that justified the cost. But there didn’t have to be. And in our case there won’t be…

…I’ve seen a few listed companies stretching the limits of credulity between actual and maintenance capex. But none ever topped a marketing firm I was short about 15 years ago. Beginning in 2005, a little before it listed in London (yes, on AIM) the company went on an investment binge. Capitalised spending rose from only a few percent of sales to over half, and stayed around 20% for the next 5 years.

There are a few ways an investor can get some insight into that number, without even looking to see what it’s spent on. One is to compare it with similar companies. If you looked at the kind of business this management wanted you to believe they were competing with – mobile ad networks like, say, Millennial Media – you were in for a surprise. Millenial spent about 3% of sales on capex.

But you didn’t even need to get that specific. A fifth of sales going into capex is a lot for almost any established business. So another way of making sense of it is to ask what kind of operation needs that level of investment? Before you go and search for the answer, have a guess at what the most capital intensive companies are (I went for transport infrastructure, things like toll roads and airports). If you did screen for companies that had capex over 20% of sales back then you ended up with a pretty short list (leaving out start-ups with little or no revenue, which were mostly biotechs and junior miners). The list was more utilities and telcos than transport as it turned out, but there were a few airports (and some airlines as well). Shockingly no advertising agencies made the cut.

The best clue that the investment was bogus, though, was what the company stated it was spending it on. Software, sure. But not code they developed themselves, this was programming bought off the shelf. When I asked management to break it down the best example they could give was Microsoft licenses!

Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. We currently have a vested interest in Microsoft. Holdings are subject to change at any time.

Ser Jing & Jeremy

thegoodinvestors@gmail.com

What We’re Reading (Week Ending 04 February 2024)