Where Steve Jurvetson Sees Dead Companies

Is this the Twilight of the Incumbents?

(Midjourney prompt: “A young person wearing jeans and a hoodie and an old person wearing a conservative black business suit stare at each other across a rocket launchpad”)

“Any company that thinks they’re not a software company is not long for this world, because the agile way we’ve learned to build software is becoming the agile way we build everything. I sometimes feel like I have a sixth sense. I can see dead companies. They don’t know they’re dead, but they’re dead because they’re not responsive enough. And the companies that iterate more quickly will just run circles around them. They’re innovating every couple of years on something that you might take seven years to do.”

- Steve Jurvetson, 2022

“…it’s so difficult for legacy car companies to get software right… Car companies haven’t written software like this, ever. They’ve never written software. So we’re literally writing the software to operate the vehicle for the first time ever.”

- Ford CEO Jim Farley, 2023

A conversation I had with Steve Jurvetson a few years back taught me to listen attentively to him. We were at a small conference about The Future, and he and I got to talking during a break.

I mean “The Future” as shorthand, not condescension. It’s important to talk about how things are going to unfold, especially if you ground the discussion in theory and evidence. After Erik Brynjolfsson and I published The Second Machine Age a decade ago I started getting invited to discussions about how all kinds of things were going to unfold. I mentally lump all these chinwags together as being about The Future. In this case we were talking about, IIRC, the future of work, workers, and the workforce in a time of very powerful technologies (We’re still talking about this subject, as we should be).

Jurvetson is a gigantic space geek. When we visited his VC firm, the lobby looked like the atrium of the Air and Space Museum. He launches rockets in his spare time, and has an unaffected enthusiasm for all things orbital that makes the first season of For All Mankind look dour and cynical in comparison.

Kooks in Space

He started telling me about how SpaceX, in which he was an investor, was soon going to start providing us Earthlings with fast, cheap satellite Internet service. As he was talking, I felt my social brain initiate its procedure for dealing with a kook:

1. Freeze your face in a polite smile

2. Do NOT ask any questions or otherwise encourage the kook

3. Scan the room for someone on whom you can offload the kook

4. Generate a list of at-least-remotely-plausible excuses for needing to be elsewhere

5. Monitor the environment to match opportunity with excuse

If you get known as someone who thinks about The Future you have to activate that procedure a lot, but I didn’t think I’d ever need to with Jurvetson. There he was, though, talking like a space kook. He had the gleeful “here’s a story most people just don’t know!” mien common among alien autopsy specialists.

Humanity began launching communication satellites with Telstar in 1962. By the time of my chat with Jurvetson, bouncing signals up into space and back was a large global industry full of incumbent companies that understood all that bouncing very well, got paid a lot for it, and had (I thought) ample incentive to keep getting better at it. The idea that it was possible to leapfrog all of those incumbents and start bathing us in fast, cheap space bandwidth was kooky to me. If that were anything close to possible, wouldn’t we already be doing it?

I stopped listening closely to Jurvetson after activating my mental kook procedure that day, but he didn’t seem to be telling me that SpaceX had made a breakthrough in physics or communication theory. It wasn’t the case that Elon Musk had screamed “Eureka!” one day, flung himself at a whiteboard, and scribbled the equation that blew past all previous extraterrestrial Weissman Scores. So then what was going to enable SpaceX to boldly go where no satcomm company had gone before?

Nothing, was the reflexively skeptical answer I came up with as I edged away from Jurvetson. When I saw how wrong I was, I edged back toward him. And decided to write an entire book about a critical innovation that’s led to Starlink and many other good things.

New and Improved

The innovation I’m talking about is an upgrade to the company itself. An upgrade, not a complete rethinking. Upgraded companies like SpaceX look normal in many ways. They have meetings and conference rooms and budgets and badges. They have division of labor, specialization, hierarchy, and professional managers. They don’t practice holacracy or internal Marxism (in large part because these approaches don’t work). If you walked around them, you wouldn’t say to yourself “I’ve never seen anything like that before!!” over and over.

But you do say that when you look at their performance. SpaceX was founded in 2002. Since then it has

• Designed and built the world’s first commercially viable reusable orbital rockets, and still the only ones. As of March 12, 2024, the company had stuck 283 rocket landings and completed 249 relaunches. Reuse greatly reduces the cost of getting things into space, and has made SpaceX a near-monopolist. In 2022 it was responsible for 80% of all the satellites that left Earth’s surface, and more than 60% of the mass (according to Bryce).

• Also designed and built the Starlink constellation of low-Earth orbit Internet communication satellites. More than 5400 of them are currently overhead, providing service to more than 2.3 million customers around the world (of which 1.3 million are in the US where, in sharp contrast to most other ISPs, its users love it.) In the wake of the Russian invasion Starlink was the only option for quickly deploying rugged, portable, reliable, fast satellite terminals into Ukraine. There are now more than 40,000 of them there, and “absolutely all front lines are using them” according to Ukraine’s intelligence chief.¹

• ALSO designed and built the Dragon capsule, which is the only US-made vessel certified by NASA to carry astronauts into space. After the Space Shuttle was decommissioned in 2011 Americans had to use Russian Soyuz craft to get to the ISS until Dragon came along in 2020. Boeing got a contract to build an astronaut capsule at the same time SpaceX did (in 2014), but still has not completed a crewed test of its vessel.

To summarize: SpaceX entered a knowledge-intentive, capital-intensive industry full of experienced, well-capitalized, well-connected incumbents. And from a standing start, it’s been able to blow past these incumbents and dominate large global markets in less than a quarter century. It might even be profitable now!

I suggest that how were they able to do all this? is a question worth spending some time on.

Is it because of Elon Musk? It’s certainly partly because of Musk. He’s a rare talent, as his accomplishments at SpaceX and Tesla attest.² But what does he actually do to accomplish big things? Does he just drive people harder, micromanage at a higher level of resolution, think more clearly and deeply, take more risks, and work more hours than anyone else?

That does seem to be part of it. But not the whole part. And although Musk appears to be everywhere, he’s not. He’s not part of the story of the recent disruption of industries as diverse as consumer electronics, journalism, retail, advertising, filmed entertainment, and computer hardware and software.

Those all look like the space industry in that they’re populated by a set of incumbents who, during the 21st century so far, have mostly underperformed and lost ground (sometimes a lot of it) against upstart interlopers. I call these interlopers geeks, and my book The Geek Way gives my explanation of how the geeks have taken over so much economic territory so quickly.

That explanation is: a bunch of geeks, concentrated in but not exclusive to Northern California, have given the company an upgrade. I call this upgrade the geek way. I’m going to use this venue to talk about it; to explain what the geek way is and why it’s such an upgrade over the mainstream “best” business practices of the 20th century (which I’ll often call the Industrial Era).

As I was trying to come up with my explanation I interviewed a lot of my favorite alpha geeks, including Jurvetson. His prescience about SpaceX and Starlink told me that I had been a dope when I labeled him a kook, so I listened carefully and asked a lot of questions this time.

His quote at the top of this piece comes from our interview. I love it because it makes a strong and non-obvious claim, and one that I agree with: getting good at writing software the modern/agile way makes you better at doing other difficult, complicated things related to product development and project management — whether those things involve bits, atoms, or a combination of the two.

Incumbents’ Reasons Why Not

Farley's quote at the top of this piece is an honest assessment of where his company and its peers are in their ability to write software the modern/agile way. Let’s call such companies IEIs, for “Industrial Era incumbents.” If we assume there’s at least some truth to what both Jurvetson and Farley are saying, an important question comes up: How many IEIs are going to be able to upgrade themselves? Well…

Some IEIs might not recognize the need to change. They might think that the practices and philosophies that led them to success in the 20th century will continue to do so in the 21st.

Some will look at the sheer amount of work involved in adopting the geek way and decide it’s too much. Farley does a great job explaining how big the challenge is:

to save $500 per vehicle, we farmed out all the modules that control our vehicles to our suppliers… So Bosch would do the body control module, someone else would do the seat control module, someone else would do the engine control module. We have about 150 of these modules all throughout the car… The problem is that the software is written by 150 different companies, and they don’t talk to each other. So even though it says “Ford” on the front [of the car], I actually have to go to Bosch to get their permission to change [their] software. So even if I had a high-speed modem in the car and the ability [to do over the-air-updates] it’s actually their IP… We have 150 different programming languages, the structure of the software is different, it’s millions of lines of code and we can’t even understand it all.”

Given all this, it would be completely understandable for an IEI CEO to say “Screw it. I’m going to look elsewhere for ways to improve performance and share price during my likely time remaining at the helm here.” According to Farley, many of his peers in the auto industry are taking this route.

I won’t name names because they’re competitors, but it’s shocking to me how many people are sticking with very old… architectures and software from… a confederacy [of suppliers]. It will never work, and no matter how many software engineers they hire, [their] car’s not going to work.

I agree with Farley’s prediction. But I can also see how an IEI CEO could talk themself into disagreeing with it, and into believing that tweaking their technological and organizational status quo — instead of overhauling it — will get them where they want to go.

Are IEI boards pushing their CEOs to do the full overhaul? Probably not. The board of an Industrial Era stalwart like a legacy car company is composed largely of Industrial Era stalwarts. Across the boards of the Russell 1000, the average age is almost 62. I’ve talked with a lot of IEI boards, and I can vouch that their members are aware of the need for digital transformation. But how many of them share Jurvetson’s view that unless the companies they help govern overhaul the way they build software and everything else, they’re dead? Not many. Not enough, IMO.

Some will try and fail. Farley looks to have a Jurvetsonian view of things, and has taken big steps to fix Ford’s vehicles’ software mess:

In the second generation [electric vehicles} we’re going to… write all the software [ourselves]… I had to split the company into three pieces because I kept watching our [internal combustion engineers] try to figure out how to do over-the-air updates or change the software for the vehicle. They didn’t know; they’re not software people.”

Let’s wish him luck and hope things go better for Ford than they have for VW, which since 2019 has been trying to write lots of its cars’ software itself. Instead of a model for other IEIs to follow, it’s instead provided many customs to be honored in the breach. The software has been late, buggy, and often un-updateable over the air; there have been frequent rotations in the executive suite (most notably an out-rotation of VW head Herbert Diess); demand has been tepid; and reviews have been …mixed. Last October VW announced further delays on some car models, the “redevelopment” of an entire EV software platform (which would push its launch close to the end of the decade), and 2000 layoffs in its software group.

Deep organizational change is hard. So is changing the fundamental approach used to create any piece of technology as complicated as car-level software. Combining the two kinds of change is a climb-K2-in-winter-level challenge.

Some will fail to realize that the agile way the geeks have learned to build software is the way most everything should be built. In 2015, Musk said “In an average week, for the Model S, there are 20 engineering changes. So, we make 20 sort of small changes per week…. And we will continue making improvements. I mean, Tesla is really a continuously improving company… Our friends ask me, ‘Well, should I wait to buy a Model S? Will there be a better one in the future’ Well, yeah, of course there will be a better one in the future. There will always be a better one in the future.” There’s not a single 2015 Model S; instead there are hundreds(?) of different versions of the Model S produced that year. Many geek companies’ offerings have this feature; they keep getting better over time.

Some will fail to upgrade themselves in the other ways I describe in The Geek Way. The geeks haven’t confined their innovations to product development and project management, as important as those are. The geek way also extends to many other fundamental aspects of running a company:

1. Making good decisions at all levels

2. Avoiding sclerosis and excess bureaucracy; remaining nimble at scale

3. Seeing accurately into the future

4. Surfacing good ideas and letting the best ones win

5. Wrong-footing the competition

6. Avoiding cultures of silence, undiscussable topics, and the emperor’s new clothes

7. Ensuring that everyone’s work remains aligned with the goals and strategies of the organization.

8. Ensuring that the ringing words found in annual reports and motivational posters bear at least some relation to day-to-day reality³

9. And so on.

How many IEIs will go for the whole geek upgrade? How many will succeed? Of those that do, how long will it take them to accomplish their Jurvetsonian overhaul? How long will it take them to get to the point he describes, which is where leading geek companies are today (and they’re not standing still)?

I truly have no idea. But we’re about to find out.

1. Musk has at times used his control over SpaceX and Starlink to hamper the Ukranian military by “geofencing” Starlink so that its terminals didn’t work in some areas.

2. Much less so at whatever we’re supposed to call Twitter now.

3. Or, to use the late, great Chris Argyris’s formulation, making sure that the organization’s espoused theory lines up with its theory-in-use