Last week's virtual dialogue between Elon Musk and JPMorgan Chase CEO Jamie Dimon served as a grand roadshow for SpaceX's potential public listing, though the opening interaction with his mother, Maye Musk, set a tone of ease and rapport.
The event was held on the 51st floor of JPMorgan Chase's New York headquarters, with Musk participating via video link.
The host shifted focus to Maye Musk in the audience, who stood up and remarked, "I want to say, when you were three, I told people I had a genius son, and everyone would roll their eyes."
Musk responded, "Of course. Frankly, that's understandable."
Maye continued, "But I just knew. Later, when you said you wanted to build rockets, it was my turn to roll my eyes."
"And then you actually did it," she added, brimming with pride.
Because the host was Jamie Dimon, the questioning was notably direct and to the point.
Dimon was not content to let Musk speak freely; he repeatedly steered the conversation back to practical matters: Why list now? How much of the undersea cable network could V3 replace? How will a space-based data center transmit data back to Earth? What is the actual scale of the U.S. memory chip production gap?
Musk's answers were delivered in his characteristic style, ranging from propellant costs to the mass distribution of the solar system, and from Starship reusability to the energy ceiling of human civilization.
The core logic underpinning all this can be summarized in one sentence: the company is entering a new growth phase requiring massive capital expenditure—100,000 satellites, space-based AI data centers, the Terafab chip factory. The reason for listing now is that capital needs to be deployed where it matters.
In closing, Musk stated he is still constantly learning and making mistakes, "but perhaps future AI will evaluate me and say: 'By human standards, this guy was okay.'"
Initiating the Discussion
Jamie Dimon: Let's start with the first question: Why is now the time for a SpaceX IPO? You had a choice; you didn't have to go public. Why now?
Elon Musk: Over the years, many people have asked me why I don't take SpaceX public.
For nearly a decade, people have been suggesting I should bring SpaceX to the public markets.
We've been cash flow positive for a long time. From my recollection, around 2014 or 2015, we became self-funding.
In fact, our past so-called private funding rounds weren't strictly for raising capital but to provide liquidity for investors and employees, as we grant stock to everyone in the company.
Moreover, in most of these 'funding events,' SpaceX actually bought back shares.
So what's different now?
Several things.
First, we are entering a very significant growth phase. More precisely, a phase requiring substantial capital expenditure.
We might launch 100,000 satellites into orbit, perhaps even more. That's just for communications satellites.
These will be third-generation and beyond, not the first or second-generation currently in orbit.
The capability of the third-generation satellites, depending on how you calculate it, is roughly 10 to 20 times that of the second generation.
Furthermore, SpaceX's own chip design team has taped out three custom chips for this system. These chips far surpass the current state-of-the-art.
What does this mean?
It means the future Starlink system could provide roughly one hundred times the bandwidth of the current system. Simultaneously, latency will be cut in half because the satellite orbital altitude will also be approximately halved.
I believe it will become the communication method with the highest bandwidth and lowest latency.
And in the future, with the development of AI and robotics, the world will need vastly more bandwidth than it does now.
You can imagine, what is the 'bandwidth' of a human? A person's peak bandwidth might only be a few hundred bits per second.
But a computer's bandwidth can reach a trillion bits per second.
So, the appetite for bandwidth from AI and robotics will be enormous.
Additionally, we are working on AI data centers in space. That's another direction requiring massive capital investment.
But I believe this will become one of the primary ways for AI to expand.
Because building power plants on the ground is becoming increasingly difficult. Few people want a power plant in their backyard.
If we want to double the U.S. electricity consumption—the average U.S. load is about 500 gigawatts—that means building nearly the same amount of power capacity again. I don't think most communities would be particularly excited about that.
But if we go to space, we can far exceed the scale of power generation possible on Earth.
This might sound crazy, but you could increase the energy available to humanity by a millionfold while still using far less than a millionth of the sun's energy output.
The entire energy usage of human civilization today is less than a trillionth of the sun's total energy output. Thinking about that is actually quite humbling.
We are truly very small.
When you actually see the size of Earth relative to the sun, you realize we are just a tiny speck in the vast darkness, while the sun is immensely huge.
The sun constitutes 99.8% of the solar system's mass. Most of the remaining 0.2% is Jupiter.
My answer might have digressed a bit. You just asked me why list now, and I started talking about the sun's energy output.
This answer might be a bit long. If I were an AI, you might be reminding me now: "Please be concise."
However, these things are indeed important.
Because people will start thinking about what future energy production will actually look like.
My answer: absolutely solar. Or better yet, stellar energy—the energy of a star.
Here's a crazy fact: if you took all the matter in the solar system except the sun and burned it, the energy produced would, after rounding, still leave the sun as 100% of the solar system's energy source.
Because the sun is 99.8% of the solar system's total mass.
Even if you 'teleported' two more Jupiters from another solar system and burned those too, the sun's share would still round to 100%.
So, the key is the sun.
You could scale the economy in space to a million times the size of Earth's economy—using accessible energy as a proxy for economic output—while still using far less than a millionth of the sun's energy.
It really makes you feel how insignificant we are.
And that's just one star; there are many such stars in the universe.
So, to summarize in one sentence: we are entering a massive new growth phase, and this phase requires capital.
Another point is revenue predictability. I now have more confidence in revenue forecasts. Previously, revenue was somewhat volatile, but now I feel it has become more predictable.
Destination Considerations
Jamie Dimon: When people hear concepts like 'multi-planetary species' and space travel, they think it's one of the most exciting ideas in human history.
Can you explain the bridge you mentioned before? I've heard you talk about it, from Earth to the Moon, then to Mars.
Elon Musk: Yes.
Actually, you don't necessarily have to go via the Moon to get to Mars.
But I think we could potentially establish a self-sustaining city on the Moon faster than on Mars.
Also, if you want to scale far beyond what Earth can support, the Moon has a potential advantage: it has no atmosphere, and gravity is about one-sixth of Earth's.
This means you could use electromagnetic accelerators, railguns, or mass drivers.
Essentially, you could launch AI data centers from the Moon into deep space without rockets. You could literally shoot them out like from a railgun.
Furthermore, you could use materials on the Moon to manufacture solar panels and radiators there.
This could enable scaling to over 1,000 terawatts per year. That number is staggering.
I believe, launching from Earth, we might achieve about 1 terawatt of space-based AI computing power per year. But launching from the Moon, it might be possible to reach 1,000 terawatts or more.
Also, as I mentioned, we could establish a Moon base. And I think it would be incredibly cool to vacation on the Moon in the future.
That would be an epic vacation.
Of course, not everyone wants to go to the Moon, but I think many would. I believe it would be truly amazing.
Naturally, the premise is that you can go safely and return safely, safe in every aspect.
But I think this could be possible in the future.
Then, Mars is the next step.
Mars is an entire planet. Its gravity is closer to Earth's, and it has an atmosphere, albeit a thin one.
If you warm Mars up, eventually you could terraform it to an Earth-like state: with liquid oceans, life, where people could walk outside without a spacesuit.
Mars is currently a sort of 'fixer-upper planet,' but it has great potential.
Starship's Core Innovations
Jamie Dimon: Let me move to the third question. But I never really thought one day you'd enter the hotel industry.
Elon Musk: Of course. We should have a Moon hotel, right?
Jamie Dimon: Yes. Musk Hotels or something.
Elon Musk: Or someone else could do it.
Actually, you could view us somewhat like the Union Pacific Railroad in its day.
When they built the Union Pacific Railroad, many thought they were crazy. Because people asked: Why transport so much cargo and so many people to California? There's nobody there.
But now, California is the largest state in the U.S.
Jamie Dimon: For now.
Elon Musk: Right, and maybe not forever.
Jamie Dimon: Yes, it is somewhat self-diminishing.
Speaking of Starship, you've built it. It's an incredible spacecraft. I've read it has flown 12 times.
But obviously, technically, this is very difficult.
What are the key breakthroughs that people should know about but might not fully appreciate?
Elon Musk: We do good livestreams. So if anyone wants to understand Starship, I recommend watching the SpaceX website or any Starship launch livestream. That's a great way to learn about it.
But the fundamental breakthrough with Starship is that it will be the first fully reusable orbital-class rocket.
That sounds like it should be obvious.
Because in every other mode of transport—whether airplanes, cars, bicycles, horses, boats, any kind—we take reusability for granted.
If you had to throw away the airplane after every flight, air travel would become extremely expensive.
And that's essentially what rockets have been like in the past.
But from a technical standpoint, achieving full reusability for rockets is extremely difficult.
We've achieved partial reusability with Falcon 9, and Starship will complete it entirely.
Once fully reusable, the cost to reach orbit is reduced to just the cost of propellant.
Because all parts of the launch vehicle can be used repeatedly.
Starship uses liquid oxygen and liquid methane as propellant, which are among the cheapest propellants you can find.
Oxygen can be obtained directly from the air, methane can come from natural gas.
So, the propellant cost for Starship will be lower than the jet fuel cost for airplanes.
This means the future cost of sending cargo to space should be lower than the cost of transporting cargo across an ocean by plane.
Starlink's Next Phase
Jamie Dimon: Great. Now on to Starlink. It's also an incredible thing, global communications.
You told me before that V3 might replace some portion of undersea cables. By the way, that's a huge security risk for all of us, as several undersea cables have already been cut.
It happened in the Baltic Sea. So, what's next for Starlink? Including V3, and potentially V4 in the future?
Elon Musk: The V3 satellite will be a massive thing.
You can look it up online. As I said, its capability is roughly 10 to 20 times that of V2.
It's a very large satellite. In fact, it can only be launched by Starship. It's too big; no other rocket on Earth can launch it.
The Starship cargo bay diameter is about 30 feet. The V3 satellite is about 7 meters wide, roughly 22-23 feet. Very large, basically the size of a small bus.
There are many technical details. For example, we use much larger phased array antennas and have more new ground links.
Our satellites communicate with each other via lasers; these laser systems are developed and manufactured in-house. So V3 will have more laser links, and the laser technology is more advanced.
It will also use the W-band and E-band. There are many technical details, but simply put, its capability will be much greater.
It's like a very insane orbital radio station.
Jamie Dimon: Can Starship carry 12 or 15 at a time?
Elon Musk: It should be able to carry 50.
Because the target for Starship V3, when fully reusable, is to deliver 100 tons of payload to orbit.
By Starship V4, our target is over 200 tons per mission.
Challenges of Space Data Centers
Jamie Dimon: And eventually achieving launches every hour.
Next question, space data centers. You keep talking about this, others have mentioned it too. But obviously, it requires different technical capabilities.
Space is colder, with less vibration. But you also must transmit data back to Earth somehow. I recall you mentioned possibly using lasers. It has to work in all weather.
So, after researching for a while, how difficult do you think this really is?
Elon Musk: We don't think it's particularly difficult.
In fact, we think it's easier than communication satellites. Much easier than our communication satellites.
The Starlink V3 communication satellite is an extremely complex machine. In comparison, an AI data center would be much simpler.
Because it's essentially solar panels, radiators, plus some basic equipment for satellite operation, and laser links.
These laser links will connect to the Starlink communications constellation, which will then relay the data back to the ground.
And this connection won't be affected by weather.
Because once you first connect via laser to the Starlink communications constellation, the communication bands between Starlink and the ground can penetrate clouds. In fact, to some extent, they can even penetrate roofs.
So, you can always establish a link with these data centers.
Rationale for In-House Chip Production
Jamie Dimon: Let's talk about the U.S. There's a big topic we've discussed for a long time: U.S. reindustrialization, bringing advanced manufacturing back to the U.S.
Now you're talking about building the Terafab wafer fab, a chip factory in New York. You're already doing so much; what's prompting you to do this now?
Elon Musk: We're always asking: what is the limiting factor?
The limiting factor we see now is chip manufacturing capacity. This includes both logic chips and memory chips, as well as advanced packaging.
Notably, there is currently no large-scale, mass-production computer memory chip factory in the U.S. Not one. Zero.
Micron is building one in Idaho, but I believe it won't reach mass production until 2028.
There are also projects in New York, but from my recollection, those are around 2029, 2030.
And this capacity is only a small fraction of what's truly needed for memory chips.
In fact, even under the most optimistic assumptions from memory chip and logic chip manufacturers, capacity still falls short of meeting projected demand.
That's also why you see Micron's market cap reaching $1.2 trillion, or some similarly high number.
So, clearly, the demand for AI logic chips, memory chips, packaging, and essentially AI computers will far exceed the capacity that existing manufacturers can provide under even the most optimistic scenarios.
That's why we need to do Terafab.
AI Integration at SpaceX
Jamie Dimon: Let me ask one more question about AI strategy. You've also introduced Grok at SpaceX. How does it integrate into the platform you're building?
Elon Musk: For SpaceX's AI satellites, we do intend to allow people to put any GPU or TPU they want on them.
If they want to put Nvidia GPUs, they can. Google TPUs, they can. Amazon Trainium, or any other chip, they can too.
In the future, we will also offer our own chips. I think we also hope to offer our own software in the future, meaning AI software.
But the system will be designed so that anyone's AI, whether hardware or software, can run on these SpaceX AI satellites.
Final Thoughts on Culture and Leadership
Jamie Dimon: The next three questions are completely different. Let's start with patriotism.
You've served this country. I know you're a patriot, and we've talked about this. How do you view your role as an American patriot helping the U.S.?
Elon Musk: I love the United States very much. I think I always have.
SpaceX does a lot of work for what was formerly called the Department of Defense.
We have a division called Starshield that provides military communications. There are other things that are classified that we can't talk about.
But we do help the Department of Defense, and we help intelligence agencies within the government. We are a very critical part of that.
Jamie Dimon: On culture and talent pipeline.
Brett has been with you for 15, 20 years. I know you've built a very deep talent bench at SpaceX.
How do you keep it running? What's most important to you? How do you ensure the best talent stays?
Elon Musk: Gwynne was about the seventh person to join the company, back in 2002. So she's been here 24 years.
Overall, the tenure of the company's senior executives is very long.
I think Bret Johnsen has been CFO for 15 years as well.
I believe it's because people truly believe in the mission. They want to stay and continue building it.
We want to make humanity a spacefaring civilization; we want to take humans to Mars, to the Moon, eventually further; we want to turn science fiction into reality, to go places never explored before.
Make Star Trek real. That's one of my favorite TV shows.
Jamie Dimon: You've built multiple companies. Compared to 20 years ago, how have you changed?
Actually, maybe I should ask you: from when you first started to today, what has changed? What have you learned? What has changed about you, both as a leader and as a person?
Elon Musk: I think I might be a bit more relaxed than before.
I might have been more tense before. I'm much more easygoing now than in the past. Of course, I'm still not particularly easygoing, but definitely more so than before.
Another thing I've slowly realized over time. I've mentioned it before too.
When you hire someone into the company, or have someone work at the company, the person's individual capability and intelligence level are, of course, important.
But equally important is whether they have a good heart.
It's not just about what someone's IQ is, or how smart they are. The key is also whether they are a good person.
This is very important.
I think I have learned a lot, although I also feel I still have much to learn and will make many more mistakes.
But perhaps future AI will evaluate me and say: "By human standards, this guy was okay."
Comments