LIVE: Jeff Bezos speaks at the VivaTech conference in Paris

Well, let me start by saying that was a rough moment. Nobody likes that. It was a gut punch for the whole team. But what we've learned since then is we got really lucky. Some of the long lead items on the launch infrastructure were preserved. For example, the propellant tank farm, all the liquid hydrogen, liquefied natural gas, liquid oxygen, all those tanks, those are very long lead items and they were fortunately undamaged. The booster, we had a booster in the integration facility which is right there at the launchpad and various pieces of shrapnel missed the booster. There was a lot of good luck in that incident. In the big picture, God knows how to appropriately price his goods and space travel is hard and it's worth it. By the end of that incident, within 24 hours completely spontaneously, the Blue Origin team, and this is a team of just incredible people, they were making themselves t-shirts that say it's worth it.

And the other thing is that we're busily rebuilding it. We got lucky in another way. There was a construction crew just down the road, so we brought in 400 pieces of heavy equipment. Brought in construction workers that were working 24/7, so the pad has now been cleared of all the debris. It's amazing how quickly that's happened. And just yesterday we started the reconstruction. I think you've said it best: we're going to fly this year. We'll fly before the end of the year.

Well, what we're trying to do is build all the infrastructure so there can be a completely dynamic space economy. Today the cost of admission to space is still very high. If you look back at what I witnessed in the internet space over the last two and a half decades, you saw an environment where because all this infrastructure existed, global networks and so on, very small companies could build very large enterprises. Two kids in a dorm room could build a giant company. And we want space to be like that. We want space to be this dynamic, entrepreneurial place where two kids in a dorm room can build an incredible space company. That is going to happen. The job of a company like Blue Origin is to help build the road to space, that heavy infrastructure, so that many other companies can do incredible things. There are so many resources in space. You're seeing that now with LEO constellations and the potential for orbital compute. We have the moon as a gift, it's right there. We're working on all these things. We're building a lunar lander. We're working on in-situ resource projects, learning how to build solar cells out of lunar regolith. There are many other initiatives we're undertaking, but all of them are focused on building this infrastructure so there can be dynamism in the space economy.

It gives the whole team a chance to show everyone who we are and what we're made of.

The team at Blue is the most missionary team I've ever worked for because we've explored every planet. We've been to the deepest parts of the solar system and this is the good one, right? This is the planet you want to be on. It's beautiful. And as we build this infrastructure, you can move polluting industry into space and you can turn every place on Earth like the parks in this beautiful city, and we could support three times the population on this planet. That's the long-term goal. That's what we're trying to do. That's why it's worth it.

Well, this is a great question. The evidence right now is very obvious. If you look at demand for launch, demand for launch is insatiable right now. We have a tremendous backlog already on our books for launch. Every space launch company has tremendous backlog. We are supply constrained, we are not demand constrained. It's being driven by communications with LEO constellations, by national security missions, and in the future by things like orbital compute and lunar resources and NASA programs to return to the moon this time to stay. There's just a tremendous amount of demand for launch.

100%. I think people underestimated it tremendously.

And it's a circle too. If launch is very expensive, then the satellites have to have really long lifetimes and they have to be very exquisite. But if the launch starts to get cheaper, then the satellites get cheaper. And if the satellites get cheaper, there's more demand for launch. If there's more demand for launch, then you get more practice with launch. Launch gets cheaper. So there's a virtuous circle there that once you get it spinning in the right direction, that's where we want to go.

So you get out there and you get ready. By getting reusable rockets, you lower the costs because these boosters want to be reused. They're expensive, and when we land them we can reuse them many times. And these constellations want to be gigantic. Especially communications, people are insatiable on how much bandwidth they want to use. When I had a 300 baud modem, I wanted to use all of it. When I had a gigabit in my home, I want to use all of it. And so these constellations want to be very, very large.

Yeah, that's awesome.

And this is a very large vehicle. This vehicle will take 45 metric tons to LEO. The variant we're working on right now, the 9x4 variant, will take over 70 metric tons to LEO. This is a super heavy class booster and to watch it land like that on an ocean going platform is awe inspiring. The team is rightly very proud of themselves for it.

No, the ocean is highly uncooperative.

We're now building a BE4 engine, our heavy LNG engine, every four days coming off the line now. Rate manufacturing is critical to this. Space travel is a solved problem for six decades. We're not trying to invent space travel, we're trying to make it cost effective. That's the key. That's why you need reusability as Dave is saying and why you need rate manufacturing. It really is about being world class manufacturers, and Dave and the team are crushing that. They're doing an amazing job.

Yeah, I think people are underestimating even in the academic world how much building physical things is going to be valued moving forward. We spent the last three decades with not enough computer science majors, but I think the pendulum is going to swing the other way. I was so inspired seeing the kids at Columbia and people that know how to build real physical things and understand the complexities of how to manufacture them. They're going to be in very high demand.

Rocket engines operate right at the limit of physics. The internal combustion temperatures inside a main combustion chamber of a rocket are 5,000 to 6,000 degrees Fahrenheit, beyond the melting points of any materials. The complexities of pressurizing those propellants in high performance turbopumps that have to be very lightweight, of cooling those materials with regenerative channels so they don't melt — this is very challenging. Designing the engine is hard, and manufacturing it is hard, and manufacturing at rate is hard. We have a little clip I could show you. Our lunar lander has an engine called the BE7, a relatively small engine with 10,000 pounds of thrust. The reliability of this engine is paramount. We just did the longest engine test in history, a 41 minute engine test. This engine ran continuously for 41 minutes, beating the record for the Space Shuttle main engine which they tested 30 years ago for 36 minutes. So that's a multi-decade record that the Blue Origin team just beat with the BE7 engine. These engines are very complex. We have the BE4 engine, which uses liquefied natural gas and liquid oxygen with an ox stage combustion cycle. Our BE3U engine is our second stage engine, liquid hydrogen and liquid oxygen, the same basic architecture used by the Apollo program. A hydrocarbon booster stage and liquid hydrogen upper stages. Liquid hydrogen is a complex fuel to use, but it's much higher performing. The problem is its density is very low, so you don't want to use it on a booster stage, but on a second stage it gives tremendous advantages. The Apollo Saturn V did this and so do we. The lunar landing with liquid hydrogen has a further advantage: when we have permanent lunar settlements and we go to the moon to stay, the materials available on the moon — water ice in permanently shadowed craters near the poles — can be converted with electrolysis into liquid oxygen and liquid hydrogen. So by using liquid hydrogen as our lunar landing fuel, one day in the not too distant future, you'll be able to use in-situ materials on the surface of the moon to refuel your lander.

We've been fixated on not skipping any steps. I'm very excited to see NASA and most of the world recognize that we should go to the moon first. We'll go to Mars and do all the other things, but the moon is the first best step. There are many reasons for that. It's a gift. It's so near Earth. We can get there in three and a half days and return in three and a half days. You can go anytime you want without waiting for planetary alignment. You can only go to Mars every two years or so. The moon's gravity well is so much lower than Earth's that when you get materials from the moon, you can lift them off with 28 times less energy per kilogram than needed to lift something off Earth. If you are producing liquid oxygen on the moon, lifting it into space is very easy compared to lifting it off Earth. As we explore the solar system and build colonies on Mars, the moon is an important first step. When you skip steps, it actually doesn't make you faster.

By the way, that Viper rover is going to go find lunar water ice in those permanently shadowed craters. That's a very exciting mission. We've seen it from orbit with various methods, but now we'll be able to go look at it up close.

You're right that the timing is good. I would also tell you this is so early. As a species, as far as space is concerned, we're just still warming up. We have not even begun. This is the earliest. The idea that we've been to the moon before, which is a great accomplishment, is much different than what we're doing now. The permanence of it is key. When we went before, we pulled it forward in time. We did it before we were ready. It was pulled forward because of geopolitics and the race with the Soviets. Now is the right time to really get into it and go to stay.

To do it, and that's just not sustainable. Give me 3% of GDP, I'll give you fusion.

Moon first.

Mars and everywhere else, too. We'll build large O'Neill style colonies in space. We'll use asteroids, near-Earth objects, and the moon to build compute in space and solar cells in space. A lot of our compute will be done in space; it'll make more sense. Ultimately, we'll even manufacture the chips that the compute runs on and the answers can be beamed back to us on this planet. Dave said it before, but our long-term vision, our dream, is that all the polluting industry can be done off Earth. If space travel gets reliable enough and inexpensive enough, and we can get materials from asteroids, near-Earth objects, and the moon, then this garden planet can be returned to its pre-industrial revolution state. This is the only way in which the world is worse today than it was 500 years ago. Everything is better today: global illiteracy, infant mortality, global poverty. Everything keeps getting better. The one exception is the natural world. We can actually have both.

Yes. There's a great quote from Jim Lovell. When he went to the moon and looked back on Earth, he said, 'I realized you go to heaven when you're born, not when you die.'

And there will be many of these constellations. Just the first phase of Terrowave is 5,000 plus satellites in LEO and another hundred satellites in MEO. You start thinking about the scale of these constellations; they want to be five, 10, 20,000 for each shell and will have different use cases. You have Starlink and Amazon's LEO, focused mostly on consumers. We're focused on giving enterprises and governments really high bandwidth with high reliability. There will be 10 more ideas for this. This goes back to why launch is so constrained now: each one of these constellations can quickly fill up many hundreds of launches. We've got to race to lower the cost even further and get more rockets in the air. Europe will have their own constellations, other countries will have theirs. It's just a very valuable thing to do. The human appetite for comms is insatiable.

Prometheus is building a set of tools designed to empower engineers to really invent and build much faster. Today there's a dream-build cycle: you dream of something, and depending on how complicated it is, it may take a few years to 10 years before you're producing it at rate. If you take a step back, all civilizational wealth is driven by invention. 6,000 years ago somebody invented the plow and we all got wealthier. Then the steam engine, we all got wealthier. If we can accelerate that dream-build cycle, it will create real productivity and prosperity. That's the idea. It can't be done with traditional large language models. They have a place, but they aren't trained with the right data to do detailed engineering. The analogy is: I could read a thousand books on how to be a great gymnast, but I would still be a terrible gymnast. It needs a different kind of training data. Designing real physical objects is very complicated, you can't do it just by reading. So we're building a model that is very good at doing engineering. That model becomes the basis of tools to dramatically accelerate that dream-build loop. The goal: if today you came to me and said, 'Jeff, I need a new jet engine with 10% more thrust,' even if I had built 50 jet engines already, I'd tell you it's a 10-year program. Can we at Prometheus build tools to make that five years, then three, then two, then one year? If we can accelerate that loop, it changes everything. It helps Blue Origin too. Dave needs these tools. It's not going to replace people; it's a tool for engineers. There's a lot of concern that AI will make humans redundant. I totally disagree. AI will create a labor shortage because it will make it possible for people to identify more problems. We have an endless set of things to invent. We are limited not by our imaginations but by what we can actually do. Every person in this audience has had an idea for a new business or product that stayed in their head and went nowhere because it was too hard. If we accelerate the dream-build loop, all those ideas become possible. You can already see this with vibe coding. I was classically trained as a computer scientist, but three years ago I was terrible. Now I can write an iOS app in an afternoon. That's one thing LLMs do well because it's symbolic. Imagine tools that bring that to the physical world: contemplate something and it comes off my 3D printer. I'm chomping at the bit. It's so exciting.

Thank you, Dave. There's a saying: when you're under 40, never hire your friends. When you're over 40, only hire your friends. I am a very lucky man with a tremendous amount of gratitude that I get to work with Dave.

It's going very well, especially since Dave arrived. Decisiveness is about speed, and speed matters in business. Amazon is a giant company, but we still make decisions very quickly. Decisions get slow in large companies when you treat all decisions as if they're the same size. One-size-fits-all thinking never fits. There are giant decisions that are consequential and irreversible; those should be made slowly with great care. There are other decisions that even if consequential can be reversed; those should be made by single individuals with good judgment. Speed of decision making is key. In aerospace, traditional aerospace often suffers from slow decision making speed. But building things with hazardous operations and life safety critical missions doesn't mean every decision is like that. If you treat every decision as life safety critical, you move very slowly. We want to move humanity forward into the next age. We're in the middle of a bunch of golden ages: biotech, space, AI. Every young person should be so excited to be where they are now because it's never been a better time to be an entrepreneur or start a company.

It's going to be very exciting to watch it unfold.

I think it's so important. Mike knows some of these stories about my grandfather. My grandfather, when I knew him, was a rancher in South Texas. I spent all my summers on his ranch from age 4 to 16 and eventually was able to help him. We did everything on our own. Like a lot of people in rural areas, you can't call to repair something, you fix it yourself. We repaired bulldozers, took out giant transmission gears, fixed them. He even made his own veterinary needles to stitch up cattle. He'd get a piece of wire, heat it with a blowtorch, pound it flat, drill a little hole, and sharpen it. Some of the cattle even survived. This idea, this attitude that any problem is solvable, is a good starting point. Some problems may take a long time, but if you start with the opposite point of view — that the problem is not solvable — that will be a self-fulfilling prophecy. And with that, our time is up. But thank you.

Thanks Mike.