What We’re Reading (Week Ending 20 October 2024) - 20 Oct 2024
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 20 October 2024):
1. Actual Reform has materialised – Leonid Mironov
The Ministry of Justice and the NDRC have put out the draft of the Law on Private companies, or to give it’s full name, People’s Republic of China Private Economy Promotion Law. And it’s really good one…
…It emphasizes innovation, technological advancement, and participation in strategic industries, while also providing improved legal protections and equal treatment to address longstanding concerns. In return, private businesses are expected to follow Party leadership, contribute to national development, and operate in compliance with laws and regulations…
…The takeaway is that the private enterprises are now not discriminated against in the key project deployment. They will have similar cost of capital to the SOEs and will be able to supply most major national projects.
Chinese SOEs have been told to get more competitive earlier in the year, now the playing field is being somewhat levelled. The government, to my mind, is taking onboard the idea that employment is employment, whether SOE or not, and if a private enterprise can provide it, its fine.
I honestly think that this is the most consequential announcement, as it’s an example of a long-term reform that the government has committed to, and it is carrying out. This gives us hope for land and hukou reforms, as well as pension reform eventually. But also, this is a sign that there no decision to increase direct state participation in the economy but rather, assuming that companies follow guidance form the CCP, the more efficient actors, whether private or SOE, will drive the new policies.
2. Becoming Berkshire: 1969 – Illinois National Bank – The Weekend Investor
Around this time, Buffett and Munger sought a bank to purchase and found a candidate in Rockford, Illinois.
On April 3, 1969, Berskhire Hathaway, Inc. acquired 81,989 shares, out of a total of 100,000 shares outstanding, of the common stock of the Illinois National Bank and Trust Co. of Rockford, Illinois, at a cash price of $190.00 per share. They also have made a tender offer to acquire the remaining outstanding shares at the same cash price.
Buffett considered Rockford Bank one of the most well-run and profitable he had ever seen. It was managed by Eugene Abegg, who was 71 years old.
Abegg, who owned one-quarter of the shares, had been negotiating to sell the business to someone else before Buffett came along. The potential buyer had started criticizing the deal and wanted an audit. This affected Abegg, and he decided not to go ahead with the deal. Meanwhile, Buffett worked out what he was willing to pay, which turned out to be about $1 million less than the other buyers.
Abegg was so fed up with the other bidders that he pressured his fellow shareholders to accept Buffett’s offer, threatening to resign if they did not.
The crusty Abegg was just the type of fellow that Buffett liked.
In 1931, Eugene Abegg, a young man with only $250,000 of capital, formed a bank in Rockford, Illinois… It had $400,000 of deposits. Since then, no new capital had been added to the bank by its owners. Nevertheless, by 1969, Abegg had built, piece by piece, a bank with a net worth of $17m and $100m of deposits.
He carried thousands of dollars of cash in his pocket and cashed checks for people on the weekends. He carried a list of the number of unrented safe deposit boxes with him everywhere and would try to rent you a safe deposit box at a cocktail party. Mind you, this is the biggest bank in the second-largest city in Illinois at that time…
…The Illinois National Bank, which Buffett soon came to refer to by its colloquial name of Rockford Bank, had been chartered in the days before the U.S. Treasury assumed the exclusive right to coin money. Buffett was fascinated to discover that it still issued its own currency. The ten-dollar bills featured Abegg’s picture. Buffett, whose net worth was now more than $26 million, could have bought almost anything he wanted, but not this. Gene Abegg had done him one better. He and the United States Treasury had the privilege of issuing their own currency, but not the Buffett Partnership or Berkshire Hathaway. The idea of legal tender with your own picture on it captivated him. He began carrying a Rockford bill in his wallet…
…Berkshire paid $190 per share to acquire Illinois National, plus $2 per share to an investment bank for services rendered in the transaction…
…With total Assets of $117.3 million, shareholder equity of $16.8 million, and a net profit of $1.7 million in 1968, the bank had an ROE of 10% and ROA of 1.4%.
3. Jigar Shah on the Nuclear Power Revival in the US – Tracy Alloway, Joe Weisenthal, and Jigar Shah
Joe (04:30):
Thank you so much. So I’m going to start off with this question, which is: Okay, we went for a long time basically without building new nuclear power plants. It’s starting to pick up again. How much is it because something has changed policy-wise with subsidies and tax credits, et cetera, versus demand is back, therefore the economics of nuclear makes sense? Or would you say it’s not binary?
Jigar (04:53):
Well, look, I think that when you think about what happened through a historical context in the 1970s, we had high inflation and nuclear power was subject to high inflation. And so part of this is people were already worried about building new nuclear plants before the incident occurred because things were just getting more expensive. And when you think about the utility bankruptcies that occurred way back when, it was because they had cost overruns on nuclear power. And so I think that in general, it goes to when America stopped believing in itself and its ability to do big things and infrastructure. And I think this moment, with load growth and with the president saying we are going to build big things here, has gotten people thinking again, “Hey, what would it take to actually figure this out this time around?”
Tracy (05:48):
This is actually exactly what I wanted to ask you about because I was reading that the initial construction cost for Unit One of Three Mile was about $400 million. And I guess today the cost of building a nuclear plant would be like $5 billion, $10 billion. Obviously the $400 million isn’t adjusted for 1960s prices, but it does seem in general like it’s more expensive to build nuclear plants, certainly since the 1960s. Where did that additional cost increase actually come from?
Jigar (06:23):
So when you think about building things… like if you were to build multifamily housing and you would build one multifamily housing building versus building 12 throughout the city. You can imagine if you’re using the same design, it would be cheaper. The workers would get better. The first one would cost more, the second one would cost less, the third one would be even less.
You get faster. I mean, you see that when you go into a new home construction place. The first home takes it seems a lot longer and then suddenly the homes start popping up every week. This is the same with nuclear power. We trained 13,000 people to build the Vogtle nuclear plant in Georgia, and then we were done. And where did all those workers go? To other jobs. So now if we wanted to build Units Five and Six — we wanted to rebuild V.C. Summer [Nuclear Station] in South Carolina, which is like a hundred miles away — we’d have to go out and find another 13,000 workers. And so one of the things that we have to figure out how to do is to figure out how to build 10, right? And have those same workers that we trained, all those same EPC [engineering, procurement, and construction] contractors, all of those same suppliers, not have to stop and start, but we continue to do these one-off things…
…Jigar (09:42):
So when you restart a nuclear plant, the nuclear plant is viewed as new additional capacity, right? Because it was shut down. And so as a result, this technology agnostic credit that was created by Senator Wyden, right?
Because remember we always had the solar tax credit and the wind tax credit and all these other things. So over time the IRA moves us to a technology-neutral tax credit so that everything that is clean gets this technology-neutral tax credit. It’s a pretty lucrative tax credit. Depends on the technology, but let’s say 3 cents a kilowatt hour. And so now you’re in this place where you actually have a bonus production credit. Now you separately can choose to get an investment tax credit, but it happens to be that the production tax credit is more lucrative for these restarts of nuclear plants. But if you decide to do the investment tax credit, then you get the 30% tax credit, then there’s bonus tax credit.
So if it’s part of an energy community, you get an extra 10%, right? If you have a lot of domestic content, you get another 10%, right? So you could imagine that some of the folks who are building brand new nuclear plants might go that direction, but as a result of these incentives, nuclear power is now very cost effective.
Then the question becomes who actually wants to buy this power? Because wholesale market prices have been low. And so then the question becomes who wants to buy it? And it happened to be that two different utility groups in Michigan competed over wanting to buy all the output out of the Palisades restart. And so he picked one of the groups to buy that power and then that led to the project becoming financeable, right? And so once that succeeded, then Constellation was like, hell, maybe we could do this…
…Jigar (11:49):
So for a restart, you generally choose a production tax credit, not the investment tax credit. And that’s because the cost of restarting a reactor is a lot lower than the cost of building a brand new reactor. So you make more money by getting that extra 3 cents a kilowatt hour for the next 20 years. So the math there is you put up, it depends on where the final cost runs out, but let’s call it $1 [billion] to $2 billion to do the restart. And then you get this 3 cents a kilowatt hour multiplied by the number of kilowatt hours that plant creates. And remember, a nuclear power plant runs on average, in the United States, 92% of the time. So that’s a lot of kilowatt hours that comes out of that plant. Whereas with a solar farm, you might get 25% of the time production, with a tracking system. The math means that you could get almost all of your money back on the $1-to-$2 billion from the tax credits.
Then you’ve got the sale of the power that you’re signing a long-term contract for, and that’s where you make your return…
…Jigar (14:42):
Into the PJM [Interconnection.] And Microsoft says, depending on what happens with this power, we will make you whole on the payment. So if we said that we’re going to pay 9 cents a kilowatt hour and you end up getting 7 cents a kilowatt hour, we’ll pay that 2-cent difference. And that includes not just the kilowatt hour price, but also includes the capacity payment. So you may have heard that the PJM had a very large increase in the price that the capacity payment cleared and the capacity payment is essential, because it convinces the coal plants or the natural gas plants or others who are sort of at the end of life to make investments to last a little longer because they got paid a capacity payment to stay open. So the pieces that come here are both a capacity payment and the energy payment, and Microsoft is saying that we get all the attributes, so we get to call our usage green, but separately, if for whatever reason the wholesale market value for what the nuclear power plant is creating is less than the strike price that we agreed to, then we will make you whole…
…Jigar (19:34):
It really is an extraordinary thing. I think that most people view electricity like water. So you just put a bigger pump in, you put in more pipe, it gets to your house, you got hot water, that’s great. It’s not like that at all. There is this complex physics equation that you have to solve for.
Joe (19:53):
Because the grid has to be in perfect balance all the time, right?
Jigar (19:56):
Well, so there’s the perfect balance between supply and demand. But then there’s also figuring out what the constraints are of each individual segment on the transmission line.
So if you’re using power in New York City and you’re creating a lot of extra power out of the nuclear plants in Illinois, then that power has to go via Indiana and Ohio and then through Pennsylvania to New York City, and they may or may not be able to carry that much. And so they have to do these studies. So every time you try to add something to the grid, they have to do a study and they have to figure out whether that capacity is there, how often it’s there, whether it would continue to be balanced or whether it would be imbalanced.
And so the big fight there is that… so in Texas what they do is they just look at the safety part of it, but they don’t look at the capacity part of it. They just say, “You connect at your own risk and if we’re clogged, we’re just going to tell you to shut down, and that’s on you.” That’s why they’re approving people super fast. Whereas with the PJM and others, they’re saying, “Not only are we’re going to do a safety study, we’re also going to do a capacity study and we’re not going to let you connect until this other generator shuts down and frees up capacity for your generator.” And so that then makes the wait time much longer.
4. Invest Local? – Victaurs
Well, a community bank in the U.S. is generally defined as a depository or lending institution that primarily serves businesses and individuals in a small geographic area. These banks emphasize personal relationships with their customers and often have specialized knowledge of their local community and customers. They tend to base credit decisions on local knowledge and nonstandard data obtained through long-term relationships, rather than relying solely on models-based underwriting used by larger banks…
…As of a year or two ago there were roughly $25 trillion in assets in the entire U.S. Banking system.
And the entire amount of assets in the Community Banking system is … drumroll please … $4.8 trillion for a grand total of 19.2%. Only 1/5 of all the assets in the system are controlled and managed by these smaller banks…
...When people don’t bank locally, they inadvertently contribute to a cycle that can harm their local economies:
Capital Drain: Deposits in non-local banks are often invested in national or international ventures, rather than being reinvested in the local community
Reduced Access to Credit: As community banks disappear, so does their deep understanding of local economic conditions and business opportunities.
Loss of Personalized Service: Large banks often use standardized lending criteria that may not account for local economic conditions or individual circumstances.
Economic Homogenization: As local banks disappear, communities lose a key institution that helps maintain their unique economic character.
Decreased Local Decision-Making: When banking decisions are made in distant headquarters, local economic needs and opportunities may be overlooked.
I don’t want to over dramatize the situation, but do any of these things sound good to you? And given lots of us grew up in small towns, love where we came from, owe our position in life to the kindness of a HS coach or the first job at a local restaurant, do you want capital to move away from these people? I don’t think I do.
Banking is numbers, so here are some numbers because they paint a stark picture:
- For every $100 deposited in a local bank, $58 is reinvested locally. For large banks, that number drops to just $36. This isn’t to demonize big banks, only to point out the facts.
- Community banks make 60% of small business loans, despite holding only 12% of all banking assets. (I know their 12% doesn’t jive with my 19%).
- When a community bank closes, the local area experiences an average 33% reduction in small business lending for several years. I highly recommend checking out this study. This is an awful second level impact of losing community banks…
…As of 2023, the United States is home to a staggering 33.2 million small businesses. These enterprises employ 61.7 million people – that’s 46.4% of all U.S. employees. To put it in perspective, if small business employees formed a country, it would be the 23rd most populous nation on Earth, just behind Italy. That was pretty crazy to me. Imagine if all of the small businesses went away?
But it doesn’t stop there. Small businesses are the dynamos of American innovation and economic activity:
- They generate 44% of U.S. economic activity.
- They create 1.5 million jobs annually (64% of new jobs created) – that’s like creating a new city the size of Philadelphia every year, filled entirely with new job holders. And even for those of us who aren’t Eagles or Phillies fans, we can agree this is a massive deal.
- They contribute to 33.6% of known export value and represent 97.5% of all exporters in the United States.
“Small businesses are more than just economic units,” says Dr. Emily Chen, economist at the Small Business Administration. “They’re the innovation labs of America, constantly adapting and evolving to meet new challenges and opportunities.”…
…This is a repeat stat, but worth mentioning again. Community banks provide 60% of all small business loans, despite holding only 12% of all banking industry assets. It’s as if the local high school football team was outscoring all the pro teams combined!
They make 80% of agricultural loans, forming the financial backbone of rural America.
During the COVID-19 pandemic, community banks processed 57.5% of all Paycheck Protection Program (PPP) loans, saving countless small businesses.
Community banks have over 50,000 locations nationwide, compared to about 18,000 locations for the largest banks. That’s like having a friendly neighbor on every block, compared to a distant acquaintance every few neighborhoods…
…Community banks have consistently demonstrated resilience in the face of economic challenges:
During the 2008 financial crisis, community banks continued to lend when larger banks pulled back, increasing their small business lending by 5.2%.
In the recent COVID-19 pandemic, community banks were often the first to step up, offering forbearance and emergency loans to struggling local businesses.
66% of small businesses that received PPP loans from community banks said the process was “easy,” compared to 51% for large banks…
…In 2005, Hamdi Ulukaya bought a defunct yogurt factory in New Berlin, New York, with the help of a Small Business Administration loan backed by a local bank. From this modest beginning, Chobani has grown into a billion-dollar company, employing thousands and revolutionizing the yogurt industry.
“Without that initial loan and the trust of our local bank, Chobani might never have existed,” Ulukaya has said. “They believed in us when no one else would.”
5. The next tectonic shift in AI: Inference – Rihard Jarc
To simplify it, the o1 model has a backtracking ability. The model predicts something, realizes it did something wrong, goes back, erases that, and comes back and predicts again from that point.
The most significant implication of this kind of model is that inference workloads should grow substantially more than we were expecting in the pre-o1 period.
The calculation for Inference is now not just the number of users using it multiplied by the number of times they use it. The model can now take 10x or even more time on inference compute to come up with an answer. So inference also becomes part of the accuracy process.
The second big implication for investors is that inference computing is now becoming a new scaling paradigm. So, you not only scale the model with what is now known as data and training compute, but you can also scale them with more inference.
Noam Brown, an OpenAI researcher, has said that a study on the board game Hex using AI found that if you have 15x the inference compute, it equals 10x the training compute.
The fact that you can now scale LLMs via inference means that:
A. You can have smaller models that you dedicate more inference compute that can be as good as bigger parameter models with less inference compute
B. Inference computing is much cheaper than training computing, but the market for inference will be vastly bigger than training computing. In my discussion with Sunny, I asked Sunny how big he thinks, as an industry insider, the Inference market will be; Sunny revealed that he had the chance to preview an interview with Jensen Huang, the CEO of Nvidia, where Jensen said that Inference will be 1 billion times larger than Training. Sunny added that it makes sense to think that a model is going to be used billions of times before it is updated (trained) again.
It is also important to note that the Inference chip market has much more competition than the training market, where Nvidia dominates. From an industry expert:
»Training also is notoriously hard because you need special architectures and special cards and interconnects between the clusterand RDUs and stuff like that. It’s mostly dominated by NVIDIA because they’ve done the best work there. Inference is interesting because inference can be done anywhere. Inference is very, very easy to do on any hardware. Training is harder.«
This means that other companies will be able to reap the benefits of inference chips besides Nvidia. It also means margins on inference chips are not going close to Nvidia’s margins on its training GPUs, where it basically has a monopoly.
It also opens a path for some companies to lower some of their costs, and instead of going heavy on training GPUs and scaling there, they can split some of that on inference chips and still scale the models. Inference for customers is vastly cheaper than training…
…The thing that I also didn’t mention, but because of the o1 model release and the fact that we are coming to the start of Big Tech reporting earnings, I believe there is a high chance that the hyperscalers and companies like Meta, who are building these LLMs will increase their CapEx expectations now even more in the short-term than what they did before and much higher than analysts expect. The reason is that they now have to account for spending on Inference compute to improve these models. Inference was before a cost that they could gradually introduce and control more with users getting limited access to AI features, etc. This has changed, and you can use inference to scale the model. This might not be what investors will like in the short term. Still, it is something that, in the long run, brings us even more capable models, possibilities of easier agentic AI use cases, and SLMs that have a good enough accuracy to be used more often compared to bigger LLMs. There are already estimates on how much the inference costs are more expensive with an o1 model than with »pre o1 models«. This industry expert quantifies how much more expensive it is:
» Analyst: Strawberry o1, I’ve been told it’s 4X-5X more expensive than ChatGPT?
Industry Expert: Yeah. That’s the right level. That adds up given that it will essentially use 4X-5X more tokens on average. In the worst case, it will be 10X possibly. The 4X-5X is an average number of how much more expensive it is.«
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 Meta Platforms, and Microsoft. Holdings are subject to change at any time.
