Invest Like the Best with Patrick O'Shaughnessy
· Patrick O'Shaughnessy, Scott Nolan
[00:00:00] S0: In your operating time at SpaceX, what did you learn about making great things that work quickly? [00:00:00] S1: I think it was all the classic things. There's the Elon algorithm at this point. I don't think it was explicit back then. It was just hey, we have to get this rocket launched. Let's make sure it works. Let's not over-optimize or have analysis paralysis. Just define what the goal is, come up with a good solution. This looks like a good solution. That's a good plan. Let's run with it. Let's not deliberate for months and months and months over things. Let's just decide and move forward. And if we're wrong, we can always go back, because there's extensive testing along the way. So it's really use good engineering principles. Think from first principles, move fast, get it to 90, 95%, not 99%. Get operational and you can make it better later. But if we just never launch, if we never get operational, this isn't going to work.
· Patrick O'Shaughnessy
[00:00:00] S0: One of the things that's most interesting to me today is this whole learning by doing thing, which has now been carefully studied, that it's in the iterations that you gather lots of your learnings, and there's a literal predictable curve to these things. And we've outsourced so much of this, especially in manufacturing to overseas. And so it's not that we do innovation and someone else does manufacturing. It's more that the manufacturing creates a lot of the innovation. And I'm curious how important you think that is in the next decade or two here in the US. Obviously, you're doing something that got shipped over to Russia and now we're trying to bring it back. Curious for you to say a little bit more about the broader effort to bring more of that stuff back onto our own shores.
· Scott Nolan
[00:00:00] S1: Yeah, I think there's the on-shoring piece, but maybe even simpler, it's just vertical integration. So even domestically, you have companies in the nuclear space, the aerospace industry, like defense – many different sectors where subcontracting is the norm. And you're going to subcontract a subsystem to somebody who subcontracts a component. And then that component has different inputs and they subcontract that all the way down. I think in aerospace at the SpaceX days, it was 30 layers of subcontractors in this one system somehow, I think it was the case from the space shuttle, some crazy number of subcontractors. And then in the nuclear space recently, there was one analyst call where a company was bragging about having something like 900 subcontractors, and so many that they needed regional organizers of the subcontractors. And on one hand, it's like okay, clearly what you're doing is really complicated then and there's deep barriers to entry. But every one of those interfaces that crosses another company is typically a fixed interface that's not going to move very quickly. And so you have to treat it as fixed. And so everyone's designing their individual piece against preconceived interface requirements, and you end up with a really calcified architecture of your system. This is just at the system design level, and you can't optimize across layers for the overall goal. Now, if you bring all that in house and don't have a lot of subcontractors, you can actually, as an engineering team, optimize with every iteration, trade off interface requirements. Hey, this thing that you asked of me over in the electrical team, for me, the mechanical team it's going to be really, really hard. Do you mind giving me a little bit of breathing room on that? I'm going to make it up somewhere else. And so you can have those conversations much more easily when two people are sitting side by side at a desk than when you have two separate companies across the country. And so at a minimum, I think you need to pull the engineering together and vertically integrate and just sign up for doing more of the engineering yourself and not outsource it. And then part two is the manufacturing.
· Scott Nolan, Patrick O'Shaughnessy
That's just on the design side, if you're actually going to make the thing and you're trying to design for manufacturing, well then maybe you want your manufacturing actually co-located with engineering. And so that's what all of the great hardware companies are now doing, is you at least have your first of a kind manufacturing co-located with engineering, under the same roof. You basically talk it out of – here's why this thing that you just designed is hard for me to make. If you can make it this way, I don't need a 6-axis CNC. I can use this with a laser cutter and make this part. We're going to get 10X throughput at one tenth the cost. And so I think you need to, at a minimum, put your first of a kind manufacturing co-located with engineering. And then better yet, small scale manufacturing, mid scale manufacturing. Maybe at large scale, you can push it somewhere else and go to a lower cost center, as opposed to like Southern California where so many of these companies are. But I think it's mandatory. You've got to bring it all together. [00:00:00] S0: What have you learned about the role that energy plays in a civilization or a society? It's funny, we didn't really talk about energy for a long time. I think the per capita growth of energy was flat for a couple of decades prior to this recent surge because of AI and data centers and everything. What is your framework for thinking about the role of energy in general? And then we'll get into the specifics of what you're doing.
· Scott Nolan
[00:00:00] S1: There's two of my favorite charts around this topic. One is GDP per capita versus energy consumption per capita. The R-squared on it is certainly over 0.8. You cluster every country on earth. You plot GDP per capita, you plot energy consumption per capita, and there's a very obvious line through them. There's outliers, but it's so predictive. It is the thing. Energy use and production is the ultimate proxy for human prosperity, for economic activity, for all these things. This has been, I think, pretty well understood for a while, but completely under discussed. And then you look at the US – this is the other chart I always look at – we grew our grid for a long time, and then starting around the 90s, essentially no growth until today. China was growing a long time, 2010 we were equal neck and neck. I think this year they now will be triple us on total energy production. And so the US just needs to do something if it wants to continue to be relevant economically. You need more energy production if you want to grow your economy. And I think just purely outsourcing stuff overseas just gives up capability, especially as it gets to new manufacturing techniques, how to scale manufacturing rapidly, all these things that are so important.
· Patrick O'Shaughnessy
[00:00:00] S0: How do you think the causality runs? One idea could be if China surpasses us on a per capita basis, and let's say they get to double. Is there some rule of the way things work that will just mean they'll find ways to use that energy in some way that raises GDP per capita and drives all this progress? Or is the causality the other direction as GDP rises or we find new innovations, like energy catches up to rise to meet that demand or something? It's curious to imagine if I just airdropped, I don't know, three times the US per capita capacity of energy on some random country, would that country necessarily become prosperous?