The Exchanges

Q Stanford then, and it was really, really impressive to me how they work together and how they push things forward. And, and you, during the space race, even the cold war, it seems like the government and the universities were oftentimes even a decade or two ahead of the private sector. And today it oftentimes some, in some areas it's the opposite, especially the government's the opposite. What's your take?

A You know, I think you're right. It's hard to remember, but you know, the largest purchaser of electronics in the fifties and sixties was the U S government. There really was no commercial market at scale. In fact, the first large customers for integrated circuits from Fairchild, the semiconductor, which was one of the early chip companies were not only for the Apollo guidance systems, but for the minute man to, um, guidance system, which actually kept them in business because they could charge an exorbitant amount of money. But that shifted when we went to consumer products and PCs in the U.S. The amount of dollars being spent by PC companies and others just far exceeded what the military.

AI assessment note: “the largest purchaser of electronics in the fifties and sixties was the U S government”

Q of interesting talking to you about defense now, because in some ways, Palantir was like a fat startup, as was SpaceX, as was, as was, uh, you know, even Andrel in some ways. So a lot of these defense things, it's like, you can't quite do those things that lean startups get to. Is that how you think about it, too? Are the principles still apply, or is it different?

A Well, what's really funny is, um, you know, I, of all things, I use SpaceX as an example of how you actually can build, um, fancy word is an ambidextrous organization, a 20 dollar word that you use in universities to mean can you chew gum and walk at the same time, meaning can you execute, you know, that is do continual processes and incremental innovation at the same time you could do crazy things. In the same large organization. And, and of course, I think most people listening understand that SpaceX is launching out of three launch pads, two at Kennedy, one at Vandenberg, and that's operational excellence. Those Falcon nines are popping off. What about we might hit a 120 this year. Maybe we'll hit Elon's a 144, but, but in any case, they don't screw around. There's a, there's a book and, and that process needs to be fail safe. But when we look at Starship, you know, Elon's model is we're not blowing stuff up. We're not innovating. And if we're not blowing them up fast enough, we're not innovating fast enough. Different culture, different people, different risk profiles. But the big idea is everybody understands Falcon nine pays your salary, but starship will pay your pension and that innovation stuff that's at the edge will move to the core. And they're not separate organizations, even though they are, they're connected horizontally. The learning is going back and forth. The…

AI assessment note: “I use SpaceX as an example of how you actually can build an ambidextrous organization”

Q So the market just became much more important than the government on the free market side versus the government, and that's a big part of it, you think?

A And the rate of innovation, because the rate of competition in the, in the consumer space was much faster than the, the, the need of, uh, innovation in the military side, um, because the Soviet Union and the United States in the Cold War both had their same bureaucratic kind of processes, which had kind of equaled out, even though each country was trying to do an offset strategy to the, to the other. So the consumer space, uh, basically dominated and, uh, And by the way, that didn't matter until the last 10 years or so, because inside of, um, FFRDCs, federally funded research labs, and, and our prime contractors, and, and service weapons labs, we still had the most advanced technology For, to deter or win a war, whether they were drones or hypersonics or overhead sort of, you know, whatever those, you know, things in space were Battlestar Galactica's that required billions of dollars, et cetera. Um, but that changed about a decade or two ago, as you know, and, and a good number of your listeners. Now, all of a sudden, all those things that were owned primarily by the DOD and IC are now commercial products and half of them being made in China. And here's the, here's the rub, and, and you live this with Palantir. Unfortunately, our acquisition system is still stuck in the system where all that technology was owned by the primes or internally to the DOD system.

AI assessment note: “And the rate of innovation, because the rate of competition in the consumer space was much faster”

Q Video games is hard. What, what, what was the enterprise software company? What were you selling to enterprises?

A Um, basically we were kind of the first, you know, back in the nineties, uh, Enterprise software was client server. Um, and so we were one of the first companies to pick up on the internet and realize that the browser, um, it would be a much more efficient interface and more importantly, architecture. And so we built the three components And a lot of tool, a data mining tool and a campaign manager. And it was essentially the beginning of CRM though, uh, Tom Siebel did a much better job in, in positioning and marketing. We called it enterprise marketing automation or something EMA. And he came up with CRM and, but it was the same thing when we were, um, we were essentially the same type of company for awhile. Um, And, uh, that went public in September, 16th, 99, and I retired on September 1519 99.

AI assessment note: “a data mining tool and a campaign manager. And it was essentially the beginning of CRM”

Q Tell our listeners what MIPS was. John Hennessy obviously is an amazing man. He did a lot of great things as president of Stanford. What'd you guys build together? Tell us a little more about that.

A So MIPS was just a different way to build, uh, computer architectures on a microprocessor. It was called the reduce instruction set computers, which basically said, uh, you can have a minimal number of instructions, number one, but most importantly, uh, almost all of those instructions would execute in one cycle, which theoretically would make them faster than the existing kind of clues to architectures. And they were kludges back then because they were kind of developed in piece parts and layers and, you know, added features, et cetera. Um, and so we went head to head to Intel and, um, Andy Groves crushed us, but it was like, it was a great learning experience. So I did two semiconductor companies. Um, we did a supercomputer company. I still know what our one half is and vector scale our startup time and was teaching. Salespeople, supercomputer architecture. Um, let's see, what else did I do? I did an enterprise software company, which was my last one called Epiphany. Um, I did a video game company, which reminded me never do a company where your customers are 14 year old, 14 year old boys.

AI assessment note: “MIPS was just a different way to build, uh, computer architectures on a microprocessor.”

Q I think at a very high level of theory, it makes sense. The government is very good at breaking things. I think, you know, for a long time, we couldn't build stuff because there wasn't enough female construction workers because of the DEI rule. There was just all this comical kind of bad stuff they put in place with these things. Like, is there, is there a way around that?

A Yeah, so Joe, you just pointed out, I think, something interesting. So, and now let me give you a discreet answer to your question about national industrial policy. To compete in a world that we live in, not the world that we want, we do need national industrial policy. One. Two is we need to take this stupid cruft out of national industrial policy and, and figure out what the goal was, right? It's like, We could decide that the goal is jobs or that the goal is DEI or the goal is something else. That's not really national industrial policy. That's something else. So, so if we separate out the two, which is, do we need to get indigenous chip manufacturing in the US? Sure. Do we need to have a secure supply chain for, for an industrial base for manufacturing? Sure. What do we need to do that? And what incentives do we need to, to, to put in place? I think we did a pretty good job of thinking about. On a high order. Bit of, we need chips. Um, but I would probably agree with you. We've probably cluttered it up with things that probably shouldn't.

AI assessment note: “we need to take this stupid cruft out of national industrial policy”