What We’re Reading (Week Ending 26 June 2022)

What We’re Reading (Week Ending 26 June 2022) - 26 Jun 2022

Reading helps us learn about the world and it is a really important aspect of investing. The legendary Charlie Munger even goes 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 26 June 2022):

1. Josh Wolfe, Chris Power – Factories of the Future – Patrick O’Shaughnessy, Josh Wolfe, and Chris Power

[00:02:39] Patrick: Chris and Josh, this is going to be a totally different conversation about an area that I don’t think I’ve ever explored before, very keyed in on a certain kind of manufacturing. I’m sure we’ll hit bigger themes of onshoring of manufacturing in just the next generation of this part of the economy. We’ll spend a lot of time around precision parts, what Hadrian’s doing, why Lux is interested in this area, what Chris, you and your team, are building. To set the stage, Chris, it would be great if you could, as you did for me on the phone recently, give an overview of the recent past and what has happened in this world. It’s become a topic that everyone’s talking about a little bit, but probably doesn’t really fully understand the recent intermediate past of manufacturing, where it happens, why it’s happened that way. So a little bit of a history lesson would be a great place to start to frame our conversation.

[00:03:23] Chris: For advanced manufacturing, in general, which I describe as space, defense, semiconductor, eVTOL, energy, medical devices, basically everything in the Jetson’s flying car, future, all has to be domestically manufactured because of ALTAI requirements. It’s super high precision components. And basically, 80% of the manufacturing parts for those industries flows through a high precision network of machine shops. There’s 3,000 or 4,000 of them. Average size is 10 to 12 million in revenue. In aggregate, they do 40-50 billion in revenue, but it’s incredibly fragmented, super low NPS. It’s the most perfect Keith Rabois fragmented, low NPS, vertically integrated structure you could ever possibly think of. Historically, what happened is this was built off the defense primes needing a bunch of suppliers. All these machine shops got built in the first Space Race or the Cold War. They were businesses that got started 30 years ago by 30 year olds. And now they are 30-year old businesses run by 60 year olds. What’s happened in the last five years is there’s not a lot of slack in the system. And generally, a machine shop might be making some semiconductor parts, some parts for Boeing, and then some parts for Raytheon, for example. In the last five years, because of the boom in commercial space, which has been largely driven by lowered launch costs, the success of companies like SpaceX and Anduril, and then investors like Josh have been putting money into satellite companies, rocket companies, the whole thing. If the top level, you’ve got a bunch of net new spend in high precision components from commercial space and companies like Anduril that are flooding the same supply chain. That’s big problem number one.

And what you’re seeing for those customers is, “Hey, I’m trying to ship a satellite really quickly. I’m getting parts in 6 to 10 weeks. That’s insane. Because I’ve got an aerospace engineer sitting around for another part, wasting time, when I’m trying to get a launch up. I’m trying to get my startup goals.” So all these new entrants to the market are going way, way faster than your traditional primes. Now, that’s putting speed pressure on the supply chain. And basically, you’ve got this thing where customers want fast supply chain, huge opportunity to build a business meeting that need, with a bunch of net new spend in the supply chain. That’s phase one is Hadrian builds a better mouse trap for new space and new defense. The second phase, which is really scary for the country though, is all of those 60 year olds are going to retire in the next 5 to 10 years at an increasing rate. And 90% of them, historically, when they do retire, don’t transition to private equity, or sell or transition the business to a son or a daughter. They sunset the business, lock the door, sell a machine, and throw away the keys. There’s two bits that are really dangerous for the country about that. The first one is just purely capacity. In the decade where we’re trying to butt heads with the CCP and win Space Race 2, the capacity that feeds rocket satellite, drone companies is going to fall through the floor because of this capacity issue. They’re retiring, so you’ve got this huge supply and demand imbalance in the worst possible decade that that could be happening. And on top of that, it’s not as simple as, say, a Raytheon going, “Hey, Patrick’s Machine Shop, you’re retiring. Let’s take all the digital files that tell someone how to make those parts, and give it to another machine shop.” Most of them have been made for 20 years. There’s no CAD file. The drawing is in someone’s desk drawer. And we’ve just seen this where we shipped something like a third of all our Stinger and Javelin missiles to the Ukraine. This is on the defense side, but this happens across space semiconductor really. So we shipped all over to the Ukraine. The Biden administration went to Raytheon and said, “Hey, we need more Stingers and Javelins.

And then Raytheon came back and said, “Well, apart from the fact that supply chain’s super bottlenecked and we can’t ramp up production, we just don’t know how to make any of the parts anymore. And it might take a couple of years to figure it out.” So it’s a complete disaster, both on net new spends in secular growth and decline. But what people don’t realize is it’s not as simple as, “Hey, let’s raise your semiconductor. Let’s throw a 50 billion into an Intel plant in Arizona.” Because in the ’70s and ’80s, when we outsourced advanced manufacturing, what we lost was not just capacity or capability, it was the talent and the people. And what people don’t understand about manufacturing, it’s like software engineering. To get AI researchers, you have to have a base of backend software engineers. You’ve got a million software engineers, and it breeds the best. It breeds the best. And all of a sudden, you’ve got some top tier people in deep cloning and all that other stuff. It’s the same in manufacturing. You can’t really skip these training levels. So what we lost was not the knowhow to do a specific part, but the talent base that can produce better and better people that can work on things like semiconductors or advanced manufacturing. The slack in the system is not simply a capital problem. It’s this talent based problem. You can solve some of that by trying to grab some people from Taiwan, people who really know this, and rebuild all these industries. But it’s much, much slower than people think it is because it’s not as simple as turning a capital key, buying some machines, and ramping up production capacity. It’s incredibly difficult to do. It’s a huge commercial opportunity, but it’s incredibly important that we get this right for the country because space is basically a defense domain. Peace through strength is a huge deal. We’ve created this period of peace with Pax Americana. And I think in the next couple of years, maybe in the next 18 months, we’re going to really see that a lot of that is risked, and there’s going to be a huge wake up call when the average American consumer not just can’t buy an iPhone for less than $4,000, possibly can’t buy one at all because of all this global shifting of the advanced manufacturing supply chain…

...[00:12:57] Patrick: I actually just did one of these with Brian from Anduril. And it was really interesting to dive into the nature of the pieces of what they’re building and their goal for speed, simplicity, modularity the philosophy of how these things are built, whether it’s Ghost or whatever the product is at Anduril, is it’s very different from a Predator drone or something that Lockheed or Northrop would put out over the course of a decade. I’m curious, Chris, how much you think in the success case for Hadrian, where it’s everything you dreamed of and more, 10 years from now, how its existence changes the nature of the things that get built? What will this new manufacturing capability, just like Stripe and Twilio, people build stuff that they couldn’t have dreamed of before because they were able to go so fast with this new tooling, how do you think about that, Chris, in terms of what this might lead to? That even though there’s amazing things happening at SpaceX and everywhere else, it’s on the back of these 3,000 mom-and-pops. What will be different in the success case for Hadrian, for the people at the top of the chain?

[00:13:53] Chris: I think if you think about software engineering 10 years ago, maybe to start a SaaS company, it costs a million dollars, and you were spending more than 50% of your time on activity like running a server farm or building payments that every single software company had to deal with. When you see these platform infrastructure companies like AWS come out, or Stripe come out, or Twilio, you get to really interesting dynamics. One of which is the cost to start a company in this space goes through the floor. So now with all the tooling, you can start a software company for a couple hundred dollars. Secondly, the number of companies that get started because of that tooling goes through the roof. And then the third thing is the speed at which those companies can iterate, basically turn engineering time into a good product that the market wants, goes through the roof because their iteration cycle goes through the roof.

If we get this right, we should be able to drive three things. One is that existing companies can iterate on products in order of magnitude faster, which means that at the product layer, you just get better products. You’re not doing a year long cycle for a satellite. You’re doing a two month cycle for a satellite. As you’re getting feedback from the customer, your designs can change way, way faster. Secondly, by having Hadrian as a platform, we should be able to dramatically lower the cost of starting advanced manufacturing companies, which will drive a Cambrian explosion in both this evolutionary, who’s winning in the marketplace to build a satellite company or a drone company. The raw number of these companies that start will go through the roof. And that would be success for me.

[00:15:19] Josh: One of the things you just said, which I think is really interesting, there’s this old thought experiment, which was actually manifest in a physical experiment, where you took two different classrooms of people that were making some sort of pottery. They had a very specific end state of a pot that they had to make. And one was told, “Spend two hours or an hour or whatever it was making the pot as perfect as you can.” And the other was told, “Make as many pots as you can.” The latter, which was rapidly iterating, and trial and error, and trial and error, ended up making the more perfect pot. So that idea, which I think applies to industry, is if you make something and then you’re waiting forever to test it in the real world versus being able to rapidly iterate. The latter example is something that Hadrian’s going to enable, that in turn then, lets many more startups flourish for less capital. We can do more experiments, fund more companies. They can fail fast. Or they can come up with a product that is superior and competitive, and then build a platform from there…

…[00:22:50] Patrick: Chris, can you help us understand, going all the way to the beginning of the supply chain, the rare earth or base metal component of this process? Because I don’t think people have probably thought too much about is this aluminum, is it steel, is it titanium, is it something else? What’s the 101 on the actual raw materials that are important in this process? Because out of nowhere, after a decade of silence, the commodity world has come alive. There’s issues shipping, there’s issues sourcing, there’s issues in pricing, there’s inflation. This becomes a really important thing really quickly. So give us a little tutorial on what are the important metals that go into all of these shops as raw material and anything that you think we should know about the nature of that today?

[00:23:30] Chris: Basically there are four main alloys that all space defense semiconductor satellite companies use. Aluminum, 6061, 7070, steel variants, so 306, 316, 30X, titanium, and then Inconel variants. There’s a ton of aluminum on satellites. There’s slightly less aluminum on rockets. And then on rockets, you start to get into steels and harder metals like titanium and inconel because the closer you get to the engine, the hotter it is, so you need material that can withstand heat. And then it’s the same thing for the defense side. So if you look at a fighter jet, there’s a bunch of structural aluminum, there’s a bunch of structural titanium, because it’s incredibly lightweight, and then the engine is incredibly hard, heat withstanding materials like inconel. Those are the input materials to the parts. So let’s talk about that. I mean, let’s talk about the parts that are on the machines that we run in our factory, because that’s a little bit scarier. In a sense, I think the aluminum price over the last year, it’s come down a little bit now, but I think it doubled. That was a supply chain shock from the inputs, but then the mills themselves in America had a labor shortage. There was just lack of supply, so the price went up. The parts that went on those satellites during that nine month period, the machine shop can’t absorb them, so it gets passed on, so the satellites are now 30% more expensive. And if you look at where those materials get sourced from, we have a pretty good supply of aluminum in the United States. 90% of the titanium in the world comes from Russia and the Ukraine. A bunch of aluminum and steel comes from Europe as well. And actually, if you look back to the Cold War, our spy planes were made out of mostly titanium. Skunkworks had to sneak titanium out of Russia to be able to make spy planes.

This is why I think a lot of the hand waving around sanctions is ridiculous, because if you’re in an adversarial position and you say, “Hey, we’re doing all these sanctions,” and then there’s 50 exclusions because you’re kidding yourself about the fact that this guy’s the only one with a titanium. It’s just ridiculous. That is a real problem both on lack of production, but also availability of supply to American companies because a lot of that is offshore, which makes me crazy, because the State Department years ago should have been going to Latin America and Africa and getting supply of all this stuff and partnering with these countries and raising them up. Whereas China through Belt and Road has secured a lot of this global supply because they’ve got Russia locked up with the whole energy pipeline thing. They’ve got Africa locked up. So it’s a real huge challenge. For rare earth materials, which are more things that go into batteries, chips, that sort of stuff, they’re not on the parts that we’re producing, but our machines obviously have a ton of chips in them. And then every single satellite or rocket has a bunch of chips or circuit boards in them. So that’s a huge problem, and that’s way more strategic because obviously 70% of the world’s chips come from Taiwan, most of which is TSMC. And then the other thing is the rare earth minerals like lithium or cobalt are largely Latin American, but the Latin American mines are much, much less developed. That’s a huge challenge as well…

…[00:27:54] Patrick: Chris, if we zoom all the way back to the unit here of the mom and pop machine shop, what are the key set of jobs being done by one of those given shops? You mentioned the nature of them. It’s the 60 year old soon to retire making 10 or $12 million revenue per shop or something like this. What are the key components that are inside each of those shops that you want to lift out as core functions or jobs to be done and then start innovating on inside of a Hadrian factory?

[00:28:20] Chris: There’s three big chunks. One is the digital side of manufacturing, which is taking a customer PDF print and doing a bunch of creative geometry work to get that into machine code that tells the machine how to cut the part. That’s one big chunk, and that’s very software automation heavy. The second chunk is running the machine itself, which again is less of a robotics problem. It’s more of a software engineering problem. So what a master machinist does on the control is a lot of manipulating code on the fly as they respond to slight differences in the cutting tools, slight differences in the raw material. That’s a big operations software problem. And then the third layer is general logistics. So you’ve got unpredictable cycle times of each operation in the factory, you’ve got huge variances in how long something takes to inspect from one part to the other. And then you’ve got a lot of customer requirements that are incredibly variable for each purchase order that comes through. And as an example, this is a very simple example, but can create a lot of operational noise if you don’t get it really right at the top end of the funnel is laser marking a part. So producing all these space components and then at some point an engineer’s going to want to test it, often there’s a call out on the print that says, “Hey, engrave this part with a serial number, the purchase order number, or the print revision that the aerospace engineer said, ‘This is my part number.'”

You might think that’s easy, but there are about 10 aerospace known specifications of the depth of the laser engraving, how big or small it has to be, and what the function of that is. In a regular machine shop, you might have a guy running a laser machine that’s staring at a PDF print that might remember the specification. So a lot of it is that documentation of that engineering knowledge and then systemizing it so that the whole thing flows smoothly, yet haven’t got a bunch of random art going on. Even load balancing that is an insane challenge because you might have all these machines set up for making the part, inspecting the part, cleaning the part. But if you get one of those throughput messages wrong, where you’ve got, say, 10 jobs running through a facility at once, but they all happen to hit the quality inspection station at the exact same time, all of a sudden you’ve got a bottleneck and all the jobs are late. Before you tell the customer, “Hey, we can get this in two weeks,” having that load balancing like a data center with foreknowledge of where the capacity bottlenecks might be in three weeks so you can make good judgment calls on what you’re promising versus what you’re delivering is a huge data science and operational excellence challenge…

…[00:32:09] Patrick: Josh, how do you prosecute diligence for something like this as an investor. Hearing all that, I’m going to ask more follow up questions in the minute on the unit of the machine and the areas of innovation and the machinists, et cetera. But when you’re facing something like this early on and it is factory one or factory one’s still a glint in your eye, how do you do diligence on someone’s ability or a team’s ability to execute something like this?

[00:32:33] Josh: I’m probably going to get myself in trouble with this. First you make an investment in another company that fails.

[00:32:38] Patrick: Good start.

[00:32:39] Josh: And that’s what we did. We and Founders Fund actually were co investors in a company that did not work. Part of that was narrow focus, part of that was team and structure. There was something proverbially different with Chris that their light shone brighter. As you can hear him talk, not only understanding the macro, but if you have a customer like Brian, Andrew, or Palmer, they want to anodize titanium and aluminum and parts that not only have electric chromatic coatings that strengthen and provide performance, but look really cool. They’re super high demanding, yeah. So Chris’s understanding of the macro to the micro is something that was inspiring, but we were still making a bet without any existence proof of why we were basically going to make a very similar investment as we had before, but this time was different, those dangerous words. And it truly came down to his vision of understanding industry structure, his vision of seeing the technological pieces that could be put together. Notably, he told us why we lost money in our last investment, which was super valuable. So he diligenced our failure to diligence properly our prior investment. And part of that was you don’t want to automate everything.

He’s like, “You don’t want a 100% automation. You need humans in the loop in some of these aspects.” Maybe you want 80/20 or 70/30, but you need people that are there able to very quickly look at the geometry of a part or design, make a human decision, let the computer do it. A lot of it really came down to Chris understanding the macro, the micro of individual parts, the flow, where bottlenecks were. And then I think this is really important, you really have two different cultures. You have a machining culture, which is very blue collar in many cases. It is people working with their hands and really deep narrow specialists. And then you have this coding software culture, which is almost the antithesis of that. It would be good to actually hear from Chris, how do you think about those two people speaking very different languages, sometimes growing up and going to very different schools, actually being teammates and working with each other, because that answer that we got from him was super confidence inspiring.

[00:34:29] Chris: The difficult thing, going back to previous failures in his space, was both in private equity and software engineering trying to automate manufacturing, the previous approaches have been very egotistical in the sense of, “We don’t need any industry knowledge.” Either I’m a guy with a spreadsheet and I know how to do an IRR calculation. Operations doesn’t generate profit, finance does. So the attitude towards machinists or manufacturing people is very downwards looking. I saw the same arrogance in Silicon Valley, which was, “Let’s not try and work with the best in the industry to automate this in the right way.” It’s, “Let’s grab 30 PhDs and don’t hire a machinist until employee number 28 and just try and figure it out ourselves,” which for me is just this very coastal elite looks down on flyover state dynamic. What I recognized was in machining, all of the problems have been solved by people, the knowledge is in a bunch of people’s brains. It’s not like we’re inventing a new algorithm for machining. If we did nothing, but just find all the right answers and get them into software and process, we would win immediately. To do that you have to create a culture where people feel comfortable working with a software engineer and machinist and an operations person all in one conversation and setting the standard that just because you’ve got a maths degree from Yale and this guy didn’t graduate high school, setting that culture so they work collaboratively and there’s no finger pointing or whatever and everyone’s pulling in the same direction is really, really important. This was one of the most important things that I worked on, and it’s a combination of making sure, even really simple things like no matter of whether you’re a machinist or a software engineer, your equity that you get in Hadrian is the same based on your rank. The pay is the same. All this other stuff is super, super, super important.

Really finding the people from industry that want to share their knowledge and want to train people, is an incredibly rare thing. So we’re incredibly lucky to have the 20 or 30 people in industry that actually want to share their knowledge and understand that we’re all pulling in the same direction. And that is a really unique thing. To give you an example of how scary this is, even at the most innovative space companies, to train someone on how to inspect a part is usually this thing of like, “Hey, we’ve got all these amazing people that want to work in manufacturing but I’m on X number of dollars per hour, and I don’t want to share my knowledge because my job is at risk.” Even something as simple as training new entry to the workforce is incredibly hard because of this protectionism. People ask me what the secret sauce is, and I think investors think we invented this new technology and that’s the core of the company. The core of the company is 50 people soon to be 80 and 100 pulling in the same direction. Understanding that what we’re building is a culture of, here’s a problem let’s solve it and no matter where the solution is coming from, implement it and work together. That’s the core of what we’re building which long term is going to be a huge, huge advantage. Because if we get Hadrian right, there’s no reason why we can’t take the same team and go solve tube bending or raw material or whatever it happens to be.

[00:37:14] Patrick: You described this the first time we talked as the PhD arrogance trap, which I really liked as a phrase. Thinking you can just solve every single problem immediately with technology. Interesting to hear about the inner relationship between the two teams or the two modalities. When it comes to the individual machine and the machinist working together inside of a Hadrian factory, again maybe starting to squint a little bit and look out 2, 3, 4, 5 years, what do you think the innovation zones are on the machine side specifically? In what ways will a Hadrian machine be better five years from now than it is today? Because it sounds like there hasn’t really been much innovation on the machines themselves in these mom and pop shops?

[00:37:51] Chris: Actually, I think that’s slightly incorrect. But I would say that we are not really innovating on the machines themselves. And that’s part of the trick here is we are buying everything mostly off the shelf and then doing really tight software integrations to override the core software that lives on these machines to make them run better. But we’re not doing mechatronics and upgrading the machines themselves. Building your own machines while trying to scale a factory is like two impossible tasks. What we’re really doing is going, “Hey, these machines have APIs that control everything about them.” No one’s ever used an API for this machine ever before, and that’s really where the technology curve is honestly. Even down to simple things like, you’re meant to be able to run a machine overnight without it stopping itself. There’s actually 20 or 30 reasons why a machine would stop itself running. A Tool breaks, something goes wrong in the controller, it’s like a literal software bug. A lot of our automation is actually building the robustness into these vendor machines so that they self-correct overnight so we can get the throughput and the efficiency.

One of the reasons why you have a second shift at a machine shop, which is incredibly inefficient, is because someone’s hanging around waiting for the machine to error out and they know how to clear the error and get it going again. Which sounds insane, but that’s honestly 70% to 80% of the problem. It’s hilarious having people from industry where we come back in the morning and the machines run itself overnight and there’s 10 good parts sitting there. And people are like, “Wow, this is amazing.” I’m like, “What do you mean? These machines are designed to run overnight?” And they’re like, “No, well, it almost never happens in reality.” The reason is, because over the last 5 to 10 years, the amount of software that’s in these machines has grown exponentially, but no customer of the machines has ever been able to take advantage of it because, what machinists knows how to write software? What machine shop can afford to pour a software engineer into the problem? Or even if they had a software engineer, have them spent three months of R&D on figuring all this stuff out versus just firefighting operations because they’re trying to deliver for a customer. So that’s more what’s going on than us innovating on the hardware side.

[00:39:44] Patrick: And the innovation, the units of innovation themselves driven by software, is better-cheaper-faster the right way to think about what you hope to accomplish by starting to tune the dials using software?

[00:39:55] Chris: Definitely on the front end of the factory in the digital manufacturing CAD and CAM programming space, 100%. Because you just want to turn a 20 hour process into a 2 hour process. It’s possible. It should be done. We’re chipping away at the marble and will get there. For the factory, I actually think that simplification and robustness are the two most important things, because in manufacturing, complexity and lack of robustness are what drives costs. You’re actually better off having a system that works every single time that’s simple. That gives you two things. One is, there’s less errors so there’s not a bunch of people firefighting. And because it’s simple, you can train many, many more people into that system. Getting rid of a lot of the complexity of making everything truly error proof is a lot of the innovation there, which seems counterintuitive. But in the real world, you want as little errors as humanly possible versus trying to dial up the efficiency on something so high that it breaks one in even every 10 times and all of a sudden you’ve got three or four people standing around figuring out how to solve the problem. That’s really, really what’s important there.

Now, what you get from that is speed. So speed is not necessarily like, cut the part faster. It’s at every handover point, don’t have to go back in the step, go back in the step or have this station hanging around waiting for information because you’ve got errors. So the whole factory speed is optimized by having each of these individual pieces incredibly robust. For the customer layer, they get speed. What’s great about speed is everyone wants it, so we also get pricing power. As we hit the robustness layer, we have margin efficiency growth because people are hitting things every single time cleanly versus running around scrambling like, where’s this bit of paper, where’s this tool? Now on the customer layer because we are reliable and fast, we have enormous pricing power. It’s this interesting dynamic about manufacturing where, if you just focus on robustness and cleanness of the process, you kind of generate margin improvement automatically and therefore you get pricing power because you’re fast and you reliable.

2. Quotes from Seth Klarman Interview – The Transcript and Seth Klarman

2. The impact of rising rates: 

“That is going to test financial institutions who’s been writing derivatives they shouldn’t write, who’s been stepping out to take greater risks in their portfolio because if you can’t make it in bonds, people try to make it somewhere else.”

3. Watch out for anchoring

“After you buy something you paid for, it doesn’t matter. People cling to the idea that at least they should get their money back; maybe there is bad news, and you should sell before it goes lower; maybe put it into something else where you get your money back, but people prefer to make it back where they lost it. People anchor numbers in their heads, and they hold on to them. They have a way of remembering what happened relatively recently. If you recently had a pandemic, you over-worry about the next pandemic even though they don’t happen that often. I was certainly guilty of that after 9/11 myself. It seemed obvious that we’d get hit again, and then we didn’t for a long time.

4. Best business book: 

“We should not expect people to be rational all the time. Daniel Kahneman does a beautiful job in Thinking Fast and Slow. It is in many ways the best business book, the best investing book ever written even though it’s not ostensibly about business or investing because it tells us about ourselves”…

...6. On finding edge: 

“There are lots of ways to develop edge as an investor. One of the ways is deep fundamental knowledge. I have total respect for people who dig incredibly deep in an area where they’re doctors and medical researchers. They study biotechs and that’s formidable. No one should underestimate that power, but that’s not the only kind of inefficiency, as the inefficiency might be informational. Two things happen in markets; right markets are inefficient partly because of human nature, as I mentioned; greed and fear. People get greedy and panic; in some cases, the panic is legitimate. “Oh crap, I leveraged my portfolio, and I’m getting a margin call.” or “I have short-term clients, and they can redeem, and I’m getting redeemed, and I have to sell whether I like it or not.” There are other constraints on investors that also create inefficiencies.

Once in a while, we get a call from someone with one asset in their private equity fund who want to raise the next fund. They want to book a gain on that asset. And so, call it the last asset phenomenon. People literally will sell that more urgently, and maybe they’ll favor getting it done over the exact price they get because they want to raise their next fund and move on. They want to book a game and get paid. We live in an imperfect world, and their clients should probably not love that, but maybe their clients would love it. The manager has a lot of things to balance, so that’s just one little example. When a bond gets downgraded, there’s always an immediate rush to the exits by the investment-grade holders. A bond gets downgraded to junk, say when the bond goes literally from BBB to BB. Many bonds have to get sold; some are probably sold in advance. It’s good to know what a company does, its operations, and its worth. It’s also interesting to know that there’s a very large seller, and the bonds are 20 points lower. With essentially no change in any information, just the rating of a 26 year old at moody’s. So those are the kinds of things that can trigger our interest then we do fundamental work”…

…10. On making mistakes:

“Today, there’s not so much mean reversion. Things may not be mean-reverting because of technological disruption, so I think investors have had to raise their game massively in the last several decades, and I’m not done raising it. I probably haven’t raised it as high as it needs to be. It is a great time to be knowledgeable about technology; it was a great time if you could figure out what Amazon was up to. For a value investor, it looked hopelessly risky but for a tech investor, maybe with the right insight into the value of platforms and the value of winner take all business models, that would have been a good thing to have that I didn’t have. I pat myself on the back and say, okay, Seth, you were a schmuck twenty years ago and ten years ago for not figuring it out, but you were smart to figure it out five years ago. That’s all an investor can do; be intellectually honest, be self-critical we’re justified, and keep trying to get better every day. Like Warren Buffett, the best investors study read admit mistakes um always looking to get smarter and wiser because what else can you do as a person.”

3. Capital-Efficient Growth (with Zoom CEO Eric Yuan & Veeva CEO Peter Gassner) – Benjamin Gilbert, David Rosenthal, Eric Yuan, and Peter Gassner

David: Amazing. Eric, could you share your fundraising journey with us?

Eric: Sure. I started the company in 2011. First thing I did, I opened up a Wells Fargo bank account. It’s very easy for me to raise capital that’s why I opened up a bank account. Unfortunately, it took me several months. No VCs wanted to invest in me. Unfortunately, I do not know my brother […] Emergence Capital. Otherwise, life would be much easier. Finally, we targeted some of our friends. It reached $3 million seed funding. That’s how we started.

Here comes […]. I tried to target VC again, again, nobody wanted to invest in us either. We targeted friends and got another $6 million. That’s how we started. It’s very hard.

Ben: Nobody wanted to talk to you at that point because most people assumed video conferencing was either a settled frontier or a race to the bottom. Am I thinking about that right?

Eric: Absolutely right. That’s the thing. Everyone mentioned, Eric, you are crazy. The world has known you to have another video conference solution. Another VC friend even is a great friend, he told me that, Eric, I have a check for you as long as you do something else. I couldn’t say I did not listen. I was very stubborn. Also, he shared to me a story. Once I was told by a big VC, I do not want to mention the name, for sure, you guys do not like them.

He told me that, Eric, I do not think your […] works. Look at Skype, look at Google Hangout, look at Webex, they’re dominating, right? I debated with him a little bit. I failed. I cannot convince him.

On the way back, I told him myself, I’m going to change my Windows screensaver. Back then I was using a Windows machine. I changed the Windows screensaver—you are wrong. For several years.

Ben: Just to make sure I have my facts straight, I believe you raised a $30 million dollar round led by Emergence and then another $100 million dollar round after that. Similar to Peter, you did not dip into any of that $130 million to build the business. Is that correct?

Eric: For me, actually, I offered $30 million from Emergence Capital. I think we are on the right track. To be honest, actually, we don’t even need to raise a Series D because at the time, with that $30 million, I think the company was completely different again.

David: One thing we wanted to ask is a difference between your two companies. Peter, obviously, once you got to cash flow profitability, which was immediately, basically you never raised another round. Eric, you did make the decision to raise some more capital even after you were generating cash. Peter, you were on Eric’s board when that process happened? Why did you make that decision?

Peter: For Veeva, I didn’t raise more just because I thought I didn’t need it. It’s just that simple. As far as for Eric, when you’re on the board, that’s really Eric’s decision.

Eric: As I mentioned earlier, I offered to raise $30 million from Emergence Capital. At that time, seriously, they had no plan whatsoever to raise another round of capital. The reason why we still wouldn’t move forward to have a Series D is because I thought the economy would go down quite dramatically.

David: This was 2017?

Eric: Sixteen, ’17 timeframe. I was completely wrong…

…Ben: As we were preparing for this interview, our first thought was, if we just had one of you up here and we were interviewing you about capital efficiency, it’d be easy to chalk it up to business model and cash flow cycle. Multimillion-dollar contracts upfront in the case of Veeva, or in Zoom, customers flocking with their credit cards for a self-serve experience. These are two completely different models.

I think one of the things that it illustrated to David and I is capital efficiency is a mindset and culture thing more than a business model thing. I’m curious to hear both of your reactions to that, but also, what are the things that enabled you uniquely, more so than 99% of startups to be so capital efficient?

Peter: I can take that one. I guess I’ve seen a little bit of Zoom and a little bit of Veeva. I would say, probably, it starts with a mindset. Just run a profitable lemonade stand. From my point of view, for me, there’s safety in that. Cash generating business is always going to be valuable to somebody. At some point, a business that’s not cash generating is going to be valuable to nobody. There’s security in the long term. It starts with the mindset. I think Eric shared that.

Then you have to have product excellence, too. That’s something I think Eric and I share. We’re both product people. I think also, we both worked really hard. We work really hard now, especially Eric. Probably in the first five years, I worked really hard. You didn’t see me working really hard, but I saw you working really hard. We worked really hard, we worked really focused. Anything that wasn’t related to the product or the customer was just BS, then just don’t do it.

The first five years, I was not at a conference like this, for example. I was just maniacally focused, and then the market really helps too. That’s something you just have to get lucky on. It was the right timing for Veeva, it was the right timing for Zoom. Maybe if you started Zoom five years earlier or five years later, it would have been hard.

Product excellence, real focus, mindset, and then you have to have some luck in your market. I’m sure there are some things that I could have tried to do or Eric could have tried to do. We might have picked a bad market and then it just wouldn’t work.

We’re outliers and so is Eric. You have to pick something that most people think is going to fail to be an outlier. Otherwise, by definition, you’re picking something that most people think is going to work. A lot of people are picking it, therefore, you’re not an outlier.

Just like Eric, all VCs have any kind of note except for Emergence turned us down. Ours was really simple. Vertical specific software, that’s a small market and it doesn’t work. That’s what they would say. I was encouraged by that because I thought, well, it has an opportunity to be really good because it’s something non-obvious.

David: One thing that I want to double click on that we were talking about beforehand. Yes, you need to be non-obvious, to have a chance of a great outlier outcome, but you also need to be correct. What you both did was not, hey, I’m going to pick some random idea that other people think is crazy.

I know Veeva, as one of your core values, clear and correct target markets that you have written on the wall. What did each of you do ahead of time that led you to really genuinely believe, yes, the world thinks this is crazy, but I really think this is going to work?

Peter: I’ll go first, this is really easy. I talked to three or four potential customers for our first product. They all said, we don’t need that. That’s not interesting. It’s not a good thing to do. But I wasn’t listening to that. I was listening, are they emotionally attached to where they’re getting their product now?

Are they emotionally attached to those people? Do I feel like they’re getting value out of that thing? I could tell in their responses that they weren’t attached and they weren’t getting value. All four customers said it was a bad idea. They’re all customers now, though.

Ben: Let me understand the Peter formula to build a business. Ask a customer if they want your product, they say no. You dig deeper and say, what are you using now? And they say, oh, yeah, because I have a solution for this. But they just don’t love it, so you build for them anyway on the bet that you can be better than their current.

Peter: Yeah, you have to listen to what they feel, not what they say. They would say, yes, we’re very happy with the solution. But then you dig, oh, tell me more. Why is that? What is it that you get out of it? It’s like, uhm, uh, and that’s when you know.

David: That sounds like the video conferencing market circa about 2015, 2016.

Eric: For me, it’s very straightforward. Of course, I was an original founding team member of Webex. Two years before I started the company, I knew that Webex really sucks.

David: Did you try to tell Cisco that?

Eric: I told my team. I do not dare to tell others. Anyway, Skype is also not reliable. Google has done no work. Every day, I spent a lot of time talking to every customer. I know if I can build a better solution, I think at least I can survive.

I never thought that everybody was going to standardize on the Zoom platform. At least I know for sure, if customers do not like something, if you can do something better, you have a chance.

Ben: Eric, did you think from the outset that you were trying to build Zoom as a big company, or did you just think that you wanted to build a profitable company to survive and then you would sort of see where it went from there?

Eric: I think two things. First of all, at that time, my passion was very straightforward because Webex is more like my baby. I feel like I worked so hard for so many years, I let a customer down. I really wanted to fix that problem, but Cisco doesn’t want me to start over. I had no choice but to leave to build Zoom. This is the number one reason.

After I started a company, I realized, wow, it’s so hard to raise capital. By the way, the money that the VC gives to you, don’t think that’s the money. That’s trust. Every dollar matters. That’s why every day I was thinking about how to survive, how to survive, how to survive. Even today, seriously. I still think about, I wake up at night, how to survive?…

…David: Can you also tell us the story of lending your first big customer, which I believe is probably the deal that really made the business?

Peter: There was a set. There was the first guy who just peeked at his IT team and then worked up to the next size deal and the next size deal. It was always a step function. The first multimillion-dollar annual deals were a big customer of Pfizer. It was just hand-to-hand combat. There was a partner at the time. Actually, salesforce.com at the time said, I’ll send a note that Veeva will never win this deal. I replied back, I said, we will win this deal.

Ben: They sent it to you during the Bake Off?

Peter: Yeah, because they didn’t want to even come into the meeting with us. They were like, oh, we’re going to go with this other system integrator or something like that. I sent an email back and said, we will win this deal. Why? Because we have better people that will work harder. We’re Pfizer’s only shot at greatness and I think they want to shoot for greatness.

I remember there was this big meeting with Pfizer. There was a guy in there in charge of it. We had a certain amount of people in the meeting and the guy stood up for Pfizer. He said, we have more people in this meeting room than you have in your company. Why should we buy anything from you? I just said the same thing. We’re your only shot. We’re going to make something great and we have the best people. It seems simple to me. Then we got lucky.

I remember after winning it, thinking, oh my God, now what? Now, how are we going to make them successful? The whole company got a bonus when that customer was live and happy, which didn’t have a formulaic metric. It was based on interviews.

Ben: Did you use the invoice from that customer to then go fund product development?

Peter: Yeah. I thought, oh, we’ve just raised a $3 million round of capital. It didn’t cost us any dilution. The check came in. That’s exactly what happened…

…David: Eric, for you. I’m curious, maybe you can talk to us both in the beginning days and then also now at Zoom, how do you think about pricing and account strategy?

Eric: Our case is a little bit different. Ideally, when you start a SaaS company, either focus on vertical market or focus on departments. That’s probably the best business model. Unfortunately, we started from building a horizontal collaboration solution. It’s really hard because a lot of other competitors are already there.

David: Including free competitors.

Eric: Exactly, and a lot of free solutions. Our strategy is more like opening up a new restaurant business. You have better service, a better price, and better food. That’s pretty much it, even today.

I want to make sure our products are better than our competitors. I make sure when it comes to pricing, also better. I also make sure to offer better service. You look at any time, our product is always, always a better price across the board for any product compared to any competitors.

Ben: Life is about trade-offs. If you’re telling a customer, oh, we’re better, faster, and cheaper, what has to give? Is it something organizationally?

Eric: Efficiency. Let’s say customers, they are probably going to spend a lot of money on marketing. What can we do to leverage the network effects? If they hire 100 sales reps, what can we do to have 50 sales reps who can deliver the same value? That’s why it’s very important to have internal efficiency.

David: Which is so funny. That efficiency translates to capital efficiency, which translates to operational margins, which translates to cash flow, which is the whole point.

Eric: Totally. Yeah, it gives you more flexibility.

Peter: I would say the key also is just product excellence. That comes from the core set of engineers you hired, I think. You were especially very focused in the early days, right?

Eric: Totally.

Peter: You were not thinking about something else. You were thinking about video conferencing. I would say that’s why I got to know Eric. I got to know Eric, I thought, that’s a pretty focused guy and that his product is good. And then I tried out his product. I’m like, oh, this is really good. I want to join his board. I think that product excellence can make you more efficient, your sales cycles more efficient. Everything is better. Your product was twice as good as Webex, right?

Eric: No, 10 times better.

Peter: Ten times better? I guess my point is, if your product was only 20% better, it wouldn’t have been enough. It wouldn’t have mattered.

Eric: You’re so right. That’s why I always like the restaurant analogy. You’re buying a brand new restaurant. If the food doesn’t work, even for free, you don’t know if I’m still going to buy it anymore.

Again, back to Peter’s point. It’s extremely important. Everything starts from one thing, product excellence as a foundation. You can optimize a lot of things. If a product does not work, forget everything else. Just double down, triple down on the product. That’s the number one thing. Peter’s right.

4. 20 rules for investing in Vietnam – Michael Fritzell

Vietnam is following the East Asian playbook of manufacturing export-led growth – just like Japan, South Korea, Taiwan and China before it.

After the Vietnam war ended in 1975, formerly capitalist South Vietnam was taken over by the Communist Party of Vietnam and the country was unified.

The first measure taken by the communists was to nationalise and centralise the entire economy. Around 800,000 Vietnamese fled the country after the war, including Andy’s family.

It only took three years before war broke out again – this time against Cambodia’s Khmer Rouge, led by dictator Pol Pot. That war continued until the late 1980s. So Vietnam was almost in a constant state of war for almost half a century.

By the late 1980s, the country was in disarray. And it was becoming clear that the planned economy was not functioning properly.

The Communist Party introduced a new reform program called Doi Moi to create a “socialist-oriented market economy”. One of the first Doi Moi policies was to permit foreign investment to modernise the economy.

Today, Vietnam is buzzing with activity. The country has more free trade agreements than any other country in Southeast Asia. It’s become the default destination for companies wanting to diversify their manufacturing supply chains out of China. Vietnam is a perfect choice for manufacturing – in close proximity to key component suppliers in Asia and along the key trade route between Asia and the West.

Vietnam’s success is most evident in the country’s exports, which have risen the fastest of any country in Southeast Asia.

This export growth is also showing up in the country’s urbanisation, with young Vietnamese moving to factories to improve their livelihoods. Vietnam’s urbanisation rate is still only 38%, compared to China’s 70% and Japan’s 92%.

Vietnam’s potential is massive. Its GDP per capita is only US$2,800/year, compared to Thailand’s US$7,200 and China’s US$10,500. Manufacturing wages remain competitive, even against countries with worse infrastructure such as the Philippines and Indonesia.

Out of a total population of 97 million, Vietnam now has a middle class of 30 million people. And it’s rising rapidly. Many of those individuals are starting to buy properties, cars, home appliances, electronics and more…

…In addition, Vietnam’s demographics are excellent, with two-thirds of the population below 35 years of age. Vietnam’s working-age population is going to grow for another 15-20 years.

The country is also highly educated. Vietnam’s PISA scores are higher than the equivalent scores in the United States, the United Kingdom and even South Korea, even though its GDP per capita is minuscule in comparison.

5. Tobi Lutke – Embrace the Unexpected – Patrick O’Shaughnessy and Tobi Lutke

[00:02:44] Patrick: Tobi, it is almost exactly two years to the day since we last did this. It was early May in 2020, there was still a ton of uncertainty related to COVID. I guess there still is some extent today, and the world in Shopify and lots of things have changed a tremendous amount. I know certain things haven’t changed too. I’ve been really excited to do an updated version of our conversation and we’ll bounce all over the place, but before we hit go here, we’re having this fascinating conversation around the concept of infrastructure, generally speaking. I think it started with this idea that we might be about to come on stream to a lot of good, useful, new, history books written by people who are really there to see this stuff get built in the digital world. I’d love you to sum up that idea of what your interest is in infrastructure and the way that history is written. Even things like payback on infrastructure and the ways in which we might underestimate it. I think this is a great tone setter for what we’re going to be talking about today.

[00:03:38] Tobi: I’m thrilled to be back. Thanks for having me and those were quite some two years and a lot has happened. I think people are just underestimating the value to society of infrastructure by some incredible factor, because you see these kind of things like the interstate system. How do you imagine this thing would’ve looked if these things wouldn’t have been built? I’m not an atoms person, I’m more like a bytes person. I find that infrastructure, especially with software has this incredibly unreasonable leverage and unreasonable payback period and often we have these conversations about what’s the state of planet earth. What are things truly like? Are things getting better? Are things getting worse? There’s a lot of people sharing excellent opinions on these things.There’s a website. I hope I say this right. I think what happened in 1971, it might be a different year, but something around that time, there’s a collection of charts where once the right year comes around, a lot of numbers sort of disconnect from their previous correlations. I have no idea what happened in that year, but as a student of history and especially of digital history, increasingly I’m thinking about a very, very tangible thing that happened is that just simply most of the value creation in the world has slipped out of the things that is represented in GDPs, where a whole bunch of people built the upper net around this time then we got modern operating systems.

We’ve built a lot of silicon based computers in the nineties, but none of this was reflected anywhere. Dot com happened and everyone tried on the idea like that this tech could be very big and then found some of the ground truth to be wanting, but really sort of early mid 2000s, web 2.0 I think we call it or, at least coinciding with the emergence of that term, I think was the moment where the world of technology said, Hey, we actually know exactly how to provide value for everyone. We know exactly how to deliver services and goods and things over the internet.And by the way, there’s a lot of tweaks on the intuitions that people develop in the physical world. Physical world is very rivalrous. If you build a bridge in one place, you probably don’t build a bridge somewhere else. At some point in the world of atoms, things become zero sum, limited amount of attention at the very least and then resources as well. The digital word is different. Basically you have Turing machines, you load something on a silicon chip into memory, and then you apply electricity and you get this thing. Infrastructure and internet. I mean, I like to believe Shopify is infrastructure, but there’s public domain libraries. Just pick one, you know, SQLite. It’s like a library, probably none of your listeners have heard about, but you have probably like something to the tune of a hundred SQLite databases on your phone right now.

It’s just file format of the world basically and increasingly runs more and more and more parts on servers as well. It’s just this brilliant open public domain piece that was written by a team and great leadership, incredible conviction, but it’s not software, it’s infrastructure. And now people are using it every day for different things. And no one has to decide if we use SQLite, that means someone else can also have SQLite because all of us just add electricity. What that stores then is like an unbelievable compounding value.Again, in a lot of the ways we look at the world through GDP and other things, it’s impossible to capture the value that’s created here. Everytime someone updates something on GitHub, theoretically, it can be copied infinite amounts of times. These are not new ideas I’m sharing here, obviously. In a way, we’ve talked about this zero marginal cost of software and of course it powers a lot of value in a lot of software companies. I’m starting to believe that we haven’t fully set this idea to its logical conclusion. How much of a change will this cause over the next while?…

…[00:09:44] Patrick: Yeah. It’s amazing how much prevailing market conditions and prices can impact people’s mood. We’ll talk about that a little bit later, sticking with infrastructure though. I wonder if you’ve developed any principles or principle thinking around what makes for better infrastructure or valuable infrastructure to build. And I asked this question from a place of Shopify zone history. When we last talked a lot of the things that if you go to Shopify’s website and see what you can do as a merchant didn’t even exist two years ago. So you’ve obviously had to make choices. We’re going to build this. We’re going to not build this. What do you think about in terms of just base level principles that help you with decision making, for what kind of infrastructure to build that will have the most leverage in the world?

[00:10:23] Tobi: There are some guiding principles in Shopify product that really help us make these decisions. For instance, there’s a very basic sentence, which actually does a lot of work within the company. “Shopify wants to make the important easy and everything else possible.” Probably everyone who listens to this has bought from Shopify stores. You might have not known that it was a Shopify store because they look very, very different. This is powered by a template language I wrote forever ago called Liquid. Basically the merchants can open a text editor and just make their website look however they want, or buy a theme from someone. That’s infrastructure in a way, because here’s something I learned about infrastructure, which might sound very abstract, but maybe it’s useful. If you imagine an hourglass. An hourglass has sort of a narrow waist at some point, maybe a comic book version of an hour glass is like two triangles, inverted pointing at each other. Great infrastructure can be done when you can define what this sort of narrow waist is between the triangles. For instance, let’s use Stripe because it makes this point I think quite well. There’s one triangle on top, which is the internet, and all the engineers, and all the developers. They have a set of desires. They want to accomplish tasks, which involve movement of money. And then there’s a bottom triangle, which is like a world of COBOL code in banks. There’s a lot going on. And a lot of things you need to know, but if you manage to create a thin waist in this case, in the form of an API, now you have an agreement in the middle. This almost acts as a protocol. Here’s the fantastic thing. Once this protocol exists, it actually allows the two triangles to be replaced over time. In the case of something like Shopify, Liquid is again this templating language. People can write it. If you wrote some in 2005, the first time the Shopify went into Beta, it will still work.

Shopify is the Ship of Theseus. Nothing about Shopify is the same. The Liquid part has been rewritten many, many times, everything changed about the triangle below. Everything changed about the triangle above. Most people don’t actually even write Liquid. They actually just use drag and drop editor, which we built on top, which then writes the Liquid for you. The amazing thing is, again, once the protocol has been defined, once the demarcation line has been created, once the narrow risk is defined, then really incredible things can happen because as long as the thing keeps working, that’s in the middle, you can evolve all the pieces. And I think that’s a really, really, really powerful idea for product creation. People encounter this. If you’ve ever queried a database again, you use sequel and that’s just a thin waist system. It’s an agreed upon system, which gets you the data and as long as you keep it simple, if you send something to Microsoft SQL Server or SQLite, you’ll get the answer assuming they have the data. So that idea unlocks, I think, the right approach to internet infrastructure creation, because once these protocols have been defined, teams can go and saying, okay, these sort of made this work with duct tape and regular expressions in terms of Liquid, but let’s build this up properly, scale it out, make it so that people can use this from now on forever.

[00:13:16] Patrick: So someone once explained it to me as the equivalent of an outlet in your wall, that’s become standard that anything you plug into it like electricity flows through it very reliably and in a way that’s a standard or a protocol or something that is sitting right next to us all day every day, that without it, who knows what would’ve been invented. I’m also struck by the examples being the choke points, if you will, the most basic natural things that humans have been doing forever, like Stripe people in paying stuff, Twilio, communicating, Shopify, selling, buying. How much do you think just that is the guide for good infrastructure just looking for the longest lasting perennial human use cases and then starting from there? Maybe they’ve all been mined. I’m curious how much room you think there is left to go talking, paying, some of these things I’ve listed are like the major human motions. But I think my sense from you is that we’re still pretty early in digital infrastructure building. So how do you think about that?

[00:14:10] Tobi: Some parts are and some aren’t. It’s sometimes very, very surprising, which ones aren’t. Other things that are very, very long lasting is ownership. People like owning things. We like to acquire assets. We like to have title to them. This is not just the utilitarian value. This is also for starters and for all sorts of reasons that are uniquely human and we didn’t have good infrastructure for this. We probably still have not great infrastructure for this. It’s just barely becoming possible to own things on the internet. I think there’s lots of white space.I do fully agree though that one of the best things you can spend some time thinking about is what are things that people have been doing for a very long time. If I’ve been doing something for a very long time, like making something on the internet that taps into this emotion or into this sense for community or whatever that is you identified. I think you can analyze almost every major success story in the digital space right now and you really see a digital version of something that people have already been doing, which tells you how early it is. They’re pre the emergence of new things. Maybe the video game world is sort of there, but I think we are spending our time on computers, on the internet, very, very different right now than people will in 20 years from now. So there’s plenty of opportunity to be part of being pioneers.

[00:15:21] Patrick: So when you think about this applied specifically to Shopify and let’s just call it like a funnel of ideas for marginal infrastructure that could get built, or I guess, improvements to existing pieces of infrastructure. How does that funnel work? How are ideas fed into the top of it? What are the layers of decision making that ultimately lead to something getting green lit? What is the way that that product funnel works, given the amount of white space that might exist?

[00:15:47] Tobi: We were talking about last time, the sort of difference over the last two years. I think that we’ve gotten a lot better at this and spent a lot of time thinking about this because frankly here’s an experience I’ve had. When the COVID pandemic and the stay at home orders happened and we all did that two years ago. It was very clear that this is going to be a very, very, very white knuckle affair for everyone. There was untold stories there still, like, I mean, the world almost ran out of service in a very significant way, but probably most people don’t quite understand how close of a call that was. If COVID would’ve happened like two years before, I’m not sure we could have pulled off, not we as in Shopify, but the internet. The Cloud hosting providers, they’re like very close to food rationing. A lot happened during this time. I pulled the entire list of things that everyone was working on and basically recalibrated everything from like, does this help right now? I’m a very vocal proponent of long term thinking. People should make decisions based on the decision they assume the company 10 years from now wishes they would’ve done, but sometimes you got to just look at what’s there and be very, very practical. So I went through. In the end, I think I stopped about 60% of what we were working on. None of the things we were working on was because people made incorrect choices. Sometimes just maybe not quite applying the larger frame of reference.

For instance, there’s a lot of projects to customize Shopify to be better for brochures and so on. I understand the pitch of like that’s so and so big market and if you just get 1%, this is not my favorite form of communication, but I recognize that it happens. So a lot of the projects have been going on we’re trying to drag Shopify into adjacencies. I’m a very firm believer that you have to pick your place and then try to be ideal for that. And actually maybe to a certain point actually discourage people to pull your product into areas it’s not meant for, because Shopify should be the best piece of software everyone uses who’s in our space. Because like cheap, and fast, and delightful, and is an integration point, and simplifies the business, and magically anticipates the next step, and has something, a product, good service for you that can just help you do your thing. Shopify wants to be the mushroom to Mario or the fire flower to Mario, or just give you powers that are awesome. Moving it in all these adjacencies increases the TAM, but it stratifies it into concentric circles. For some people it’s going to be ideal in this way, but for many people it will be just never quite there. And I think that can actually have some really negative effects for feedback and all these kind things on companies.

Anyway, from this, we learn we need to have a really good mechanism by which we get the best of what we have. Shopify is very bottoms up. People can write proposals for every opportunity they see that goes into a system called GSD, which stands for get shit done. Then there’s these phases there’s proposal phase, prototype phase, build phase, and a releasing phase, and this system allows everyone in the company to see everything that’s going on. This entire plan once a year I write product themes for a company, things that we cause to make true over the year. And then they sort of decompose into different projects. Then as this proposal is submitted for transition to the build phase or to prototype phase, and then we can have great conversations about, is this a not yet? Is this a hell yes? Where does this go in a priority stack? And I think building this out has been incredibly clarifying and very, very good for the company. So a lot of the work I think over the last two years has been to get companies just really, really, really aligned on their missions. Companies can get very, very distracted in a lot of ways when they allow themselves to do things that aren’t the mission. This is especially true in a world of product. Again, if you follow a moving into adjacencies, I don’t think you will have a world class product in your adjacencies. You’re not out competing someone’s main mission with your side quest…

…[00:24:22] Patrick: People are probably less familiar with that example you ended on, Shopify fulfillment network. I would love just to take that as a microcosm of these ideas and maybe explain literally what it is to people. But I’m especially interested in its evolution. Why, obviously you were incredibly good at purely digital infrastructure. And one of the things that’s interesting that’s happened in COVID is forced the digital and the physical to smash together out of necessity, as you pointed out, thank God for the internet during COVID, and pushed everyone into this intersection unless often atoms or bits only. Maybe start by saying, what is Shopify network today? And then really, I’d love to hear on how it evolved and how it began, because I think it would be a great way to get into your company in your head about this kind of decision making and where to go next.

[00:25:09] Tobi: I’m on WhatsApp threads with probably 100 merchants. And from all backgrounds, I just talk to people and then I upgrade us into a chat. And then we talk about what works, and what doesn’t. And very quickly, this usually becomes talk about the business rather than the software, because the software hopefully works really well. But that’s actually even more helpful because it just gives you a sense for where do things get really complicated? Our observation with Shopify has always been that the journey is uphill. It’s not easy. Shopify never claims it is. Entrepreneurship is fundamentally a little bit unreasonable. There’s wonderful quotes, not by me, where people point out that you end up spending 100 hours a week working for yourself so you don’t have to work 40 hours for someone else. Often this doesn’t make sense, but again, for some people it’s super important. And frankly, for our economies, it’s really important that people do this because most people in the world are employed by small and medium businesses. There’s about five and a half million people employed by the millions of merchants on Shopify. And that’s very, very meaningful. We talk with them. What we found is it’s an uphill journey, which is okay. Everyone’s willing to do this because it’s very gritty people who embark. But if it becomes a technical climb, it filters out a lot of people.

A lot of people just opt out of the journey, basically just forgo future growth at a point where things become very, very obscure. This actually started really early. Once upon a time, for instance, actually one example was just getting a payment gateway. I know this sounds crazy that the internet was ever like this. But when Shopify started and saw a lot of parts of the internet, it was very hard to get a payment gateway. That’s trivial now because it’s built in, you just get one. So we build up the infrastructure, us and our partners to just underwrite people. And then this particular technical climb disappears. It becomes just a slope, which again, everyone will continue on. You actually have more entrepreneurship because some obstacle like this was overcome. Think about the importance of tooling infrastructure and also UX here. There are significantly more people employed today because of good UX and not getting people to be stuck and integrating more. I think this is really overlooked part of the effects of this type of friction. This is really how Shopify thinks about what we do next. People have lots of problems accessing capital from banks. Banks have in charter that the point of why they get these privileges, especially retail banks, which they have, is so that they lend money to small businesses, because that’s, again, a huge return on investment for society if that happens. However, banks do not want to do this anymore. You have to give up. And some point realistically, that’s how it should work. But in reality, they want to lend money to companies that have huge revenue, it’s lower risk. It makes sense, but that means they disappeared from playing an important infrastructure role in society. So then we have Shopify capital, because people are willing to be underwritten and for advances, and again, their business can grow significantly only even there’s capital available to grow business. We are going through all the obstacles.

The one that just is a slam dunk thing is it depends on your product somewhat, but at some point, you really have to have a plan for how to get to at least two day, ideally, overnight delivery for products you have. In the past, it was an experience unlike anything else entrepreneurs have done to this point. When they decided to go into a new channel, like sell on Facebook, on Meta or Instagram, that was a click of an app which they added. And when they did that, that’s how people are used to growing their business. Getting logistics set up is work with whatever factory and contact manufacturer you have, figure out freight across the Atlantic and Pacific. You then have to find warehouses, it is a completely different world, which involves a lot of different people to talk to and complexities. It just felt very obviously in scope for a long time, that at some point we have to solve this. In fact, I started talking with the board of directors and they wisely told me that this was too early, over 10 years ago, wanting to go into this direction. I think this is important to say. We are doing this not because we want to be in the logistics space, we rather actually don’t want to be going into the logistics space. Although it is wonderful and fascinating, and there’s lots we can actually bring given our unique experience about processes and digitalization, technology, and digital infrastructure and whatnot. But integrating end-to-end is one of the goals we have. We would like to get to the point where running a sizable retail business could, if you choose, be treated as passive income. We want to automate as many parts of it as possible so that you and your team can focus on product creation, which is the most valuable thing you can be doing. Doing undifferentiated work, figuring out where you have packages, to me that is the digital system just should really know where packages are. Otherwise, what the hell is going on? That’s not differentiated work.

Now, we found that the more entrepreneurs end up spending time on their product, the better the products get, and this is one of the wonderful things about the direct to consumer world that emerged in the last few years that there’s much more alignment between the people making the products and the people getting them. And they’re happy to send feedback. And there’s no reductionist channel and merchandising team in the middle that optimizes your products for being easy to stack or just a higher profit margin so you can compete against other products around it in the eye high shelf space in the supermarkets. Those are all influences on products that don’t lead to better products. And I think this is actually at the root of a lot of the criticism about disposable consumerism that I think is being leveled. It’s not because people love stuff. It’s because people hate the stuff they get. We are starting some of the processes and helping getting people to have this direct relationship, which just leads to actual Allbirds, like wonderful products like this, which are clearly just built with feedback from the people who wear them and want to recommend them. I think that works better for everyone and it’s what we want to see more of.

[00:31:03] Patrick: With something like this in particular, thinking back to your point about, you got to be careful about which adjacencies you get dragged into. Obviously, logistics is firmly in the vertical of core muscle movements or something, whatever you want to call it for a merchant that’s selling online. They have to get their stuff to places. What lessons have you learned entering into a much more atoms-driven world in terms of what good product means? What is a good fulfillment network? I wouldn’t know how to answer that question. Obviously, there’s the 800 pound gorilla and Amazon that proves you can build incredible logistics networks over time. I mean, it’s just a very different kind of calculus than a great new piece of software, which I don’t think anyone would say Amazon builds great software. They seem to build great infrastructure. What have you learned about that? Is it radically different than what makes you good at software? Or is it a different set of skills required than what makes you good at software to be excellent at fulfillment and logistics?

[00:31:59] Tobi: Yeah, I think so. We tend to talk a lot about intuition because intuition is also one of those underestimated things. Intuition is actually all of your life knowledge channeled quickly. I always recommend people to actually actively build their intuition for kinds of problems they want to solve in their career. There’s this uncanny thing. People were just incredible, effective, and so on. They can look at an architectural drawing and instantly tell you if it’s good or not. And then they need to think maybe 10 minutes to figure out what the problem is. But something pinged their brain about maybe call it weak signal detection like, “There’s something wrong here.”And I think this is the way intuition can be really helpful, but you have to understand that it’s task-based. Intuition built in world of bytes is not good intuition in your world of atoms. Actually, you almost want to get away from having the people who have that kind of intuition make choices. And the other thing, sometimes the bytes people end up being the most useful people in the meetings because, of course, everyone with industry experience will understand how things are. And a lot of engineers have a really good ability to think from first principles and just figure out that’s what it is, but what ought it to be? How could this all work together? And then you don’t just pivot to that. You figure out from now on, every step we do, everything we implement, how can we make it so that we can get closer to the ideal eventually? That’s a humility that’s really, really important. What does good look like? I mean, good looks like if we can put on a website that this thing will be with you tomorrow and then it does, that’s good.

At some point, this crunches together to SLAs. It becomes quantifiable in this way. And you’re right. Another thing you can do is also look at what Amazon build. And that’s also very, very good. Shopify’s relationship with Amazon, the media is trying to make this very zero sum. We treat them as a very worthy rival. Sometimes you ask or say what you can learn from them? And sometimes you ask what you can do better from them. And I hope they treat us the same as well. But again, and in those circumstances, I’ll be thinking about how to capture pieces of pies from our competitors, actually ever. Positive sum thinking is so valuable because it’s amazing how often people are trying to compete for pieces of pies rather than just grow markets. Everything about the Shopify journey has convinced me it really doesn’t pay to really have market analysis. Well known venture capitalists passed on Shopify in 2008, partly because there was only 40,000 online stores and that was not a big enough market for the investment. And I’m still disappointed with that because I realized, especially venture capitalists should not make this particular category mistake. If it’s common there, it’s clearly common everywhere.

[00:34:36] Patrick: I love this idea that if you bring this person into the atoms conversation, their intuition may just be wrong. In what ways is it most commonly wrong?

[00:34:45] Tobi: I mean, change management for software is deploy. Change management of people is a project that’s going to take you a while. The cost to switch is significantly higher. There really is a long itinerary of things that are wrong. It’s useful, but it’s useful as an input, not useful as a, “Let’s do that thing.” This goes beyond engineers, of course. Even UX has been really interesting because for instance, we’re designing UX for robotics. You scan an item, it goes onto a Chuck is what the robots are called, and the Chuck does the heavy lifting of moving it around. Just let the associates do the things that they uniquely can do well, and let the robots do the stuff that they don’t actually like doing. That’s the way we build our robotics, but this requires a very interesting human interaction design that ought to not wind up annoying after a while. And I think that’s really important. And designing interfaces that people are using every minute is different from software that people sign up once and then process some orders in every day. People that just have to recalibrate. I think that’s also makes our work really fun.

6. The Market Has No Memory. Should We? – Frederik Gieschen

In The importance of forgetting, Lauren Gravitz highlights research into people suffering from “severely deficient autobiographical memory (SDAM)” – people who are “unable to vividly recall specific events in their lives.” Interestingly, the researchers found that people with SDAM did well when presented with tasks that required abstract thinking. They were not constrained by a lifetime of episodic memory.

On the other end of the spectrum, people with “highly superior autobiographical memory (HSAM)” have an exceptional memory of minutiae, such as the clothing they were on any given day. However, “these individuals tend not to be particularly accomplished and seem to have an increased tendency for obsessiveness,” perhaps because they are unable to “extract themselves from specific instances.” The strength of their memories became a mental cage trapping them in the past.

“Why do we have memory at all? As humans, we entertain this fantasy that it’s important to have autobiographical details,” Oliver Hardt, a cognitive psychologist studying the neurobiology of memory at McGill University in Montreal, Canada, says. “And that’s probably completely wrong. Memory, first and foremost, is there to serve an adaptive purpose. It endows us with knowledge about the world, and then updates that knowledge.”

Forgetting enables us as individuals, and as a species, to move forwards.” Lauren Gravitz, The importance of forgetting

7. Neanderthal gene probably caused up to a million Covid deaths – Joe Pinkstone

A single Neanderthal gene found in one in six Britons is likely to blame for up to a million Covid deaths, according to an Oxford academic.

The LZTFL1 gene is a Neanderthal gene found on chromosome three and has been previously shown to double a person’s risk of severe disease and death.

But before now there had never been an estimated figure for how many lives were lost to this single piece of genetic code.

Roughly 15 per cent of Europeans have the Neanderthal form of the gene, compared to about 60 per cent of South Asians.

Dr James Davies of the University of Oxford, a genomic expert and ICU doctor who worked on the Covid wards during the pandemic, discovered the innocuous gene’s lethal role last year after creating a brand new cutting-edge way of looking at DNA in exceptional detail.

The method allowed him to identify LZTFL1 as the culpable gene increasing mortality, whereas previous methods had failed to narrow it down beyond 28 different genes.

Speaking at the Cheltenham Science Festival, Dr Davies said: “We used the technique and it identified a virtually understudied gene called LZTFL1 and at the time that this had not been linked to infection at all.

“It’s a single letter difference out of three billion. This tiny section of DNA doubles your risk of dying from Covid.

“It’s position 45,818,159 on chromosome three, and it’s a single change. If you’ve got a G at that site, it’s low risk. And if you have an A at that site it is high risk.”

His team believe that the Neanderthal gene changes how a cell behaves when the SARS-CoV-2 virus binds to the ACE2 receptor on a human cell.

In most people, this leads to the cell then changing shape and becoming less specialised and less prone to infection, stymying the progression of the infection.

“What this high risk variant does is it creates a new signal that tells that gene to stay on for slightly too long in response to infection,” Prof Davies said.

“And so they stay in this state where they’re highly specialised, and they’re prone to infection for longer.”

The number of deaths globally from this nefarious genetic variant “is in the hundreds of thousands to a million,” he told the audience.

Dr Davies and his colleague from Oxford Brookes University, Dr Simon Underdown, a biological anthropologist, also revealed that the Neanderthal gene first infiltrated humans 60,000 years ago after one romantic liaison and interspecies tryst between a human and a neanderthal. A solitary coupling event across species lines saw the deadly Covid gene jump from our now-extinct cousin species into us.

“If this dinner date between the human and the Neanderthal had gone wrong, we would have had a much better time in Covid, we would have had hundreds of thousands less deaths,” said Prof Davies.

“The reason that we know that is that it’s inherited as this block with 28 single letter changes, and you can track that all the way back and it has to be a single event. It’s just so unlikely that you get all 28 changes at the same time and in the same block.”

---

Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. Of all the companies mentioned, we currently have a vested interest in Salesforce, Shopify, TSMC, Veeva Systems, and Zoom Video Communications. Holdings are subject to change at any time.