Curious Worldview

Will Marshall | CEO Of Planet - Creating A Queryable Earth

• Ryan Faulkner • Episode 224

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 57:22

Will Marshall is the CEO and Co-Founder of Planet.

Planet own and operate a fleet of (200+) satellites which image daily, the entire world.

Planet’s ultimate ambition is to achieve a queryable earth. 

The way you might ask Google what the population of Australia is, you’d be able to ask Planet any conceivable question you might have about the surface of the world. The way Google would refer to the Australian Bureau of Statistics for an answer to the countries population, Planet will refer your query to their data, millions and millions of indexed images of the planet’s surface to present you an answer.

The applications of this are huge.

Take economic intelligence as an example… all types of queries that could summon early indicators of movements that aren’t already priced in.

  • For instance, as an early prediction of retail sales you could ask; How many cars are in Walmart parking lots across America right now? Or even, over the past 3 months, what’s the daily average number been?
  • Which Chinese ports are seeing more or less traffic than they usually might since the 2026 Iran war began?

And then there’s uses for climate and the environment.

  • I could ask, at what rate is a specific glacier retreating? Measure this season’s melt against each other year to date.
  • Monitoring and acting upon overfishing in protected zones.
  • Or as I ask Will in the interview, could Planet’s data be more accurate at early predictions regarding where an Australian bushfire season might be worst hit

You can imagine the applications for agriculture but as well, naturally, Planet’s data is also crucial for defence.

  • Will comment’s on Planet’s data indicating very early the Russian buildup of activity closing in on a Ukraine border.
  • And I caught Will just day’s after the 2026 war with Iran, a conflict where Planet’s data is also in use.

Will Marshall an incredible entrepreneur, but as you’ll see in the interview, he also has extensive interests beyond just those of his business.

Marshall’s PHD advisor was Sir Roger Penrose. He worked at NASA. He was on the team that discovered large quantities of water ice on the moon. He co-invented a space debris collision avoidance method using ground-based lasers. Will has lived in communal housing for 20 years. He’s a Brit abroad in America and is now the CEO of a company not only ambitioning for all the queryable stuff mentioned above, but as well is now partnered with both Google and Nvidia to explore the potential for data centres in space.

It’s a enormous pleasure to welcome to Will Marshall to the podcast.

Podcast Starter Packs

Leave a review on Apple or Spotify (nothing does more to help grow the show)

SPEAKER_01

Will Marshall is the CEO and co-founder of Planet, which own and operate a fleet of satellites who image daily the entire world and have been doing so for nearly 10 years. Planet's ultimate ambition is to achieve a queryable Earth. The way you might ask Google what the population of Australia is, you'd be able to ask Planet any conceivable question you might have about the surface of the world. And the way that would work is something like this. The way Google would refer you to the Australian Bureau of Statistics for an answer to the country's population, Planet will refer your query to their data. Millions and millions of indexed images of the planet's surface, and if achieved to the pinpoint of accuracy Will says is already here, it would mean you'd be able to retrieve the most incredible information about the surface of our planet that would otherwise be impossible to know so fast and at such scale. And the applications of this are huge. Take, for instance, economic intelligence. All types of queries that could summon early indicators of movements which haven't already been priced in. For instance, as an early prediction of retail sales, you could ask how many cars are in Walmart parking lots across America right now, or say, have been over the past three months, what's the daily average number been? You could ask which Chinese ports are seeing more or less traffic than they usually might since the 2026 Iran war began. And then of course these uses for the climate and the environment. I could ask, at what rate is a specific glacier retreating? Measure this season's melt against each other year to date? Monitoring and acting upon overfishing in protected zones? Or, as I asked Will in the interview, could planet's data be more accurate at early predictions regarding where an Australian bushfire season might be worst hit? You can imagine the applications for agriculture, and then of course, naturally, planet's data is also crucial for defense. Will comments during the interview about how planet's data indicated very early the Russian buildup of activity on the Ukraine border. And I caught Will just days after the 2026 war with Iran began, a conflict where planet's data is also in use. But this company is endlessly fascinating to me. They've been on my radar since 2021. I tried working for them while I was living in Stockholm, and I've been a patient shareholder of Planet through the lean years, always confident the success they had in the last 12 months was always just around the corner. Will Marshall is an incredible entrepreneur, but as you'll see in the interview, he also has so many interests beyond those of his business. Will's PhD advisor was Sir Roger Penrose. He worked at NASA. He was on the team that discovered large quantities of water ice on the moon. He co-invented a space debris collision avoidance method using ground-based lasers. He's lived in communal housing for 20 years. He's a Brit abroad in America and is now the CEO of a company, not only ambitioning for all the queryable stuff mentioned above, but as well is now partnered with both Google and NVIDIA to explore the potential for data centers in space. The first 20 minutes of the interview is about Planet and the queryable Earth. And then the last 30 is Planet's contribution to data centers in space, how real estate in space even works, how Will thinks about AI and its risks, his serendipitous story meeting the founders of Google, and then plenty more in between. If you haven't heard of Planet before this episode, I really urge you to look into them. They are an increasingly consequential solution to problems the world over, which have historically been invisible. It's an enormous pleasure to welcome Will Marshall to the podcast. You've had uh quite a remarkable week. How much of it can you talk about?

SPEAKER_00

Um, well, I mean, yes, there's a lot going on in the world. I mean, honestly, when world events happen, we get dragged in whether or not we like it. Um, I mean, uh, because we're imaging the entire planet every day, uh, no matter where the event is, a flood in Congo or uh obviously the events in Iran, uh, we our imagery is there. It is helping people understand the situation, is providing transparency and accountability, uh, it is shedding light, it is triangulating truth uh for people all around the planet.

SPEAKER_01

Is this too much of a reduction for surmising what planet does? High millimeter, low aperture photographers with a big data moat.

SPEAKER_00

Yeah, I think it's not a bad one. Um it's funny, I met some uh people the other day that were photographers, and I said, Oh, I'm a photographer too. And they were, what sort of photographer are you? Well, you know, a space one, you know. Take lots of pictures, and they were impressed how many pictures we took per day, which is about four million per day. Uh obviously I was being a little bit joking. Um, but but that's right. And uh, you know, a picture's a thousand words, and we're taking that across the whole planet. And we have been doing that, documenting change across the planet to help people make those smarter decisions. Uh, using these 200 satellites, not only do we have this daily scan of the Earth, we have 3,000 images on average for every point on the Earth's land mass, tracking changes through time. So it's a bit like when you go to Google Maps and see the satellite layer that you can switch to. But that was three or five years old. So our images is from today. Also, we have a time axis, which means you can go back through time and do what we call digital forensics, see changes at time, see what's normal in order to know what's abnormal. Um, you know, so uh that is an incredible power, of course, in principle. And now we're using AI to make it more and more powerful in practice. Um, you know, uh rather than it just being the specialty uh user like NASA that obviously has a lot of expertise in processing satellite data. How do we get it to the individual farmer, uh, to that uh citizen journalist, to that scientist, to the uh the the insurance company, to so many other actors that could get benefit from this? Well, we're in an era of democratization with AI, and AI is of course only about the data sets and as good as the data sets that you feed it, and Planet has really a gift uh to bring to the world in that front uh and bring to AI uh to extract out that value.

SPEAKER_01

But the the latter question around the moat of the business, what makes Planet competitively distinct from the rest of the market? Is do you did you just get the first mover advantage of having the fixed amount of real estate to have the satellites in the right position?

SPEAKER_00

Well, um I mean it's uh uh no no one else, in short, has a daily scan of the Earth. No one's erected a line scan. Basically, before Planet satellites cost about a billion dollars doing uh any sort of mission, there was a couple of companies doing things a bit cheaper and maybe ten times cheaper. We've made it another hundred times cheaper, at least. It's more like a you know, our satellites cost a few hundred thousand to a couple of million per satellite rather than a billion-ish per satellite. When we were working at NASA, I'm not exaggerating, it's of order of a billion. The cheapest ones a couple of hundred million, the most expensive ones up to ten billion. But either way, a billion plus or minus, right? Um what we did is we took satellites that are the size of a double-deck bus and miniaturized them, leveraging the miniaturization of electronics that you know you have in your pocket now. Um what has gone into these phones? Well, miniaturized computers, miniaturized sensors, your camera system, miniaturized radios for communicating, miniaturized GPS for seeing your location to have those maps work, all that stuff miniaturized into this little case. So when you look at what a satellite does, and you look at what a smartphone does, there's a 90% overlap. Batteries, GPS, radios, uh uh, a camera, and uh, you know, accelerameters, rate drivers, um uh uh processor, obviously all these things stuffed into this little thing, and it costs$1,000. And me and my colleagues were at NASA going, wait a second, um, this costs a thousand dollars, our satellites cost a billion dollars. What gives? What is those extra six zeros doing for us? And it wasn't quite clear. And so we threw a few phones into space, tested them, and showed that s this uh uh uh uh electronics that we have in our pocket is perfectly viable in space. And then we did this uh um plan to say, well, what could we do with that? We said, well, one of the things that we could do that would help the world the most if you look at all the challenges around the planet, that uh from security to sustainability is have more rapid information to help people make smarter decisions on a day-to-day basis. Truth is power, uh it enables accountability, um, and that's a good thing. Um again, in security and sustainability. And and so back to your question, I've gone round uh a long way here, but um we were the first to really miniaturize satellites to put up enough to put a line scanner for the planet. If you needed a hundred satellites to do a line scanner of the Earth, if you needed it to spend a billion dollars per satellite, that's an expensive game plan. But if you're gonna spend a few hundred K per satellite, that suddenly becomes something you could get VC funding for. So we developed that tech, we launched that, we were the first ones to do a daily line scanner. And um, you know, right from the get-go, we said this is gonna be really uh uh more powerful as we add algorithms on top, because obviously individuals can't look at those four million images per day. Um and so we were riding the wave, luckily being Silicon Valley, we were very close to it of all the AI developments, computer vision, neural networks, and now, of course, the large language models that you're seeing with ChatGPT and Gemini and Claude, um, which are further enabling this from being just a specialized discipline for being something that anyone can tap. And hopefully, this enables us to get to that vision of enabling what we call queryable earth, where just like Google figured out how to index the internet and make it searchable, we are indexing the earth and making it searchable and queryable, right? So you can have a semantic conversation with the earth like you can with the text of the internet, which is what you're doing, you know, with Gemini or ChatGPT. You know, now you can imagine how that unleashes huge capabilities if it has information about the real world. Basically, LLMs today don't really know much about the real world, and we are bringing real world data to that that unleashes huge applications uh for AI.

SPEAKER_01

It is extraordinarily exciting, and obviously you have a raw infectious enthusiasm, which is always bleeding through whatever piece of meter it is that you're doing. I remember back in 2022, actually, watching The Investor Days, and that was the big call to action. Look at this insanely valuable data set we have. We just need the data scientists out there to come and organize it for us so we can have this queryable earth. Today in 2026, how close are you with your data set? If I just gone onto the platform, um, the dashboard, whatever it's called, and I could say, how is Australia's bushfire season shaping up to be? How close are you, from that queryable earth perspective, coming to giving a very, very good answer to that?

SPEAKER_00

Very, very close. I would say it's here. Um, it's not all unleashed, but there's exciting developments. We've been working on specialized applications that enable people to answer those questions where um for some time, maybe you know, eight years, we have that where if you if there's a lot of customers that ask a question like that, we will go and figure out a specific solution that will enable it. But now it's becoming generic so that you didn't have to do that. And that is happening this year. Queryable Earth is happening this year.

SPEAKER_01

What's the most exciting part of the Queerable Earth application? You particularly?

SPEAKER_00

Oh, well, I like think looking at um uh uh how we do conservation more smartly, um, because we got this precious uh biosphere. Um, you know, I come from a background as an astronomer, and you look out in the cosmos, and uh there's uh huge numbers of planets around huge numbers of stars and huge numbers of galaxies. Um the numbers are just breathtaking. But the number that we found life on so far is exactly one. It's just this planet. Now, I hope that we will find life off Earth. But right now, all we know is that it's either we're the only ones or we're very rare. Either way, the biodiversity on this earth is so incredibly precious. You know, uh my co-founder, Robbie, he helped an exoplanet mission that was look literally looking for planets outside of our solar system and found thousands of them, which was cool. And none of them are as cool as this planet. I mean, not even close, right? This planet is really, really special. It's exactly the right conditions for life. So we have a cosmic duty. Not just to, it's not just cool that we have all this biodiversity around us, this incredible diversity of life in our rainforests, in our coral reef systems. Ask anyone that has ever gone snorkeling or scuba diving around coral reefs. I mean, they're just incredible life systems. Not only is it as beautiful as it seems, it is cosmically important. We have, you know, so it's our duty to protect that stuff. So what I like thinking about is how everyone could protect their biodiversity in their corner of the planet using this kind of data that helps them make smarter decisions in real time, that pulls in the whole of human knowledge of what smart interventions are to help us take care of that, and um has that local data and data through time to know what to do in a tactical sense in their corner.

SPEAKER_01

Is there a limitation that that application of protecting biodiversity is no not really core to the business model of Planet?

SPEAKER_00

No, not at all. I mean, so we uh said at the gate that um planet is serving a wide variety of users. I mean, I mentioned at the beginning commercials, civil government, defense and intelligence, NGOs. Um we're this is a very horizontal capability. Um, but one of the biggest areas is sustainability because you need big understanding of what's going on the planet and micro understanding. And one of the great things about satellite data is you can zoom in and zoom out. You know, it's like, what's the global deforestation problem? What's the local deforestation problem? What can I do in my uh part to stop deforestation right now in the corner? Um, and everything in between. Um, transparency drives accountability. Uh I'll give you an example. We work with the Brazilian Federal Police. Um there we scan 8 million square kilometers of the Brazilian Amazon every day. It's about the size of Australia every day of forest. Obviously, you can't look at all the images. It AI automatically finds all of the uh deforestation and illegal narcotics and uh illegal mines. It alerts the federal police, they get uh these alerts, and then 10 times per day they send a mission out to stop that illegal issue. Okay? Real-time data. And then at the same time, let me jump to security. We do work with security agencies that are monitoring large areas of threats. Take the US Navy. We're helping them monitor the South China Sea, the Spratley Islands, all of the developments that the Chinese are doing along there. Um giving truth and transparency to what's going on. The US Navy share that with the Philippines, uh, with other um uh partners in the region, so there's a common operating picture, so they all can be on the same page about what's going on. Who's doing what? What who's doing what vessels, where and when, is you know, from illegal fishing to military activities. This provides uh transparency, helps security, because it it it reduces the probability of um i uh of accidents and misunderstandings.

SPEAKER_01

Is is there room for within planet to have like an activist arm of the organization where it was simply posting out here is in the Congo an illegal road and some type of illegal deforestation. Here it is, now it's out to the world. And every single time a hit like that comes out. It would be complicated for a licit company to do, but that's that's bringing that transparency to light.

SPEAKER_00

Totally. The only tweak I'd say from that is that we let our customers do that. We have uh customers in uh media, trusted media program, we have NGOs, we have those civil governments, um, we have uh all of those to bring line. Um our job is to bring the best capabilities that are horizontal to all of that, those use cases, to those users so they can do it. We can't be all things to all of those applications. So um uh uh I I love the idea, but I think that's the way to do it is is enable our clients to do that.

SPEAKER_01

So you've got, or rather, you've spent five, ten years observing this data, and your team are bringing a lot of information to light to you. What are the key observations you've made over the planet that are most distressing?

SPEAKER_00

Well, I mean, back to my biodiversity piece, uh, we have whiffled down the biodiversity of the planet. Um uh, you know, the humans are are constantly pushing uh deforestation, uh uh uh and and clearing land for agriculture, for cattle. Um we are overfishing. Um we see violations to marine protected areas all the time. Um we see the bleaching of coral reef systems uh uh due to climate change as well as um uh uh uh uh trawling and other uh things that cause problems for fish is fish stocks. We humans um are driving serious biodiversity loss on this planet, there's no question. And it's separate from the climate change question, but related. Uh and they exacerbate one another. But if we even if we stopped climate change for a second, we would still have biodiversity loss because we're wiping out those forests not uh uh uh uh because we're uh burning those trees, but because we're doing it for agriculture. So our food systems are driving the biodiversity loss more than climate change, which is being driven by fossil fuel consumption, right? Um but but they hurt each other, right? Um and nature's part of the, they also can help each other. Um nature's part of the way out of the climate change challenge. If we get more forest back or more seagrass, they pull down the CO2 from the atmosphere and they course correct naturally. And so far, nature has already been doing that. Most of the excess CO2 we pump out gets pulled down by the by nature rapidly. But if we whittle our way that nature, we can't get away with that game anymore. And we're seeing that in a number of places. So I look the the the the the biggest thing I see when I look back in uh uh data is that the the planet is just constantly changing. It's constantly changing on its own things. Rivers move, floodplains change. Uh every year in India, Bangladesh area, the floodplains change completely because the last floods changed the the nature of the river flow uh uh attributes. So it just changes. But also we change it, right? We have agriculture, we have deforestation, we have fishing, we have all these things, and they change it. And uh so we see the planets changing, and humans are used to a static map. That's what you see online, that's what you see uh in your hand when you go hiking or what have you. We're used to static maps, but that's not the nature of the planet. The planet is evolving. So we have this sort of misconception between what we were taught and what is in in practice happening.

SPEAKER_01

Are you concerned that even though you bring a lot of this data to light by whether it's news orgas organizations or universities or even states being able to analyze what's happening in this part of our rural part of our country that we can take advantage of, that the transparency alone isn't enough to drive change? And is it giving you a bit more of a education into what actually moves the needle in terms of making a change to the environment?

SPEAKER_00

Yes, um we remove an excuse, we give people tools, but it's ultimately a sociological uh endeavor, right? Uh behavior change um is sociological and we can't force people to change uh and make smart decisions. Um but I'm hopeful because look at that Brazil case, where suddenly as a result of all those interventions by the Brazilian federal police, we reduce deforestation rates 55% year on year, right? AI and this tool in the right hands can do incredible damage on the problems that we are inflicting ourselves on the pl on the planet, right? Um and also with security. Okay, you know, uh look at Ukraine and Russia and the war and the tragic events there. I I often wonder like if we could see the Buildup of troops on the border with Ukraine before the war. But actually now looking back, we could have seen signs of that many months before. Had we pointed that out, could we have enabled the diplomacy that may have ended that or done something else? Enabled more preparation or whatever, you know. Again, human history is littered with examples where war happens because of mistakes and misunderstandings. So look, I think this technology has the potential to give humans the agency to make smarter decisions more with more forethought, more data-driven decisions, get away the politics, be focused on what's happening in actual practice, not you know, uh uh uh uh um uh you know just conspiracy theories, and actually um gives them practical actions to do about it, right? Because when you have data like that, so look, I'm an optimist in the end, um, and I'm excited by the even though it's slow, we are seeing the uptake of this sort of data fast, and it can help us with sustainability, security, and even fighting misinformation online, giving truth to a world that needs it uh more and more.

SPEAKER_01

A detail that really stands out about you in doing research for this is that you received your PhD from Sir Roger Penrose.

SPEAKER_00

Well, from the university, but yeah, he was my advisor at the end. Yeah, yeah. Um I was extraordinarily lucky to have his him in as an advisor. And um, you know, he he did all this work in black holes. We were actually working together on theories at the interface between quantum mechanics and general relativity. Um, he had some really interesting theories about that and wave function collapse, why we get the world that we do, which is not a superposition of lots of things in the classical world, but yeah, at the micro level, we're seeing all those things. Very exciting. I learned so much. I also learned that there's geniuses in this world like Roger Penrose, and I'm not one of them. Um maybe I'm just about smart enough to get a PhD, but like that stuff is so really fascinating, but uh uh uh you know it's genius territory, and uh, the geniuses are a lot smarter than others. So I learn a lot, uh, but um ultimately also that that was not the game plan for me in my career.

SPEAKER_01

That's super interesting to say, though, that there really are geniuses and you wouldn't qualify yourself as one. How do you qualify? What do you notice in their work that makes them truly a genius?

SPEAKER_00

Well, these guys can run rings around. I mean, Roger would have the best physicists from around the UK come on these Saturday meetings at his house, and then he'd run rings around them on the whiteboard. Um, and you know, talk about string theory and hopping to gravitational collapse and uh uh uh his his ideas. Um uh and you know, very few people can be fluid in all of those things, keep all of that in their head, just be able to uh and you can just uh tell that he could come up, he was racing uh at a pace. It was incredible. I was so lucky to learn learn so much from him um about uh the how the universe works. And it it you know, it got me exc very excited. Um I've always always loved theoretical physics and space and and and those are my twin passions intellectually.

SPEAKER_01

Do you lament a little bit that given your passion for astronomy in space, that an entire generation going forward, presumably every generation, is never going to be able to look into a clean, unporous sky. It's going to be it's going to have the light pollution of satellites.

SPEAKER_00

Yeah, um that's a good uh question. And I I do I am a bit sad, you know. Um when I first saw a string of Starlings, I was like, what? Oh, whoa. You know, that's it was a moment because I'd seen satellites, of course, but it suddenly got so much more at the big scale. You know, our satellites are so small you can't even see them, actually, with the naked eye. Um uh and and so they they don't fit into that, but the bigger ones you can, and yes, um that can interrupt um uh astronomers. I th I think we've got to do all we can to darken them so that they don't cause those streaks across the um astronomers uh window, or uh uh uh uh for the kid who's out in the woods and trying to uh get in nature and doesn't want to be, you know, crowded out to that. I I look, I um I think it's something that we should really try to avoid. Um and and there are things that can be done uh to make it less less bad on the satellite and how we and where we fly them. Um and so it's incumbent on the all the space actors to think about uh those implications as well.

SPEAKER_01

Speaking about those implications, like what is the geopolitics of space? What's the hierarchy between a state and a corporation? Who gets the best real estate? How do we even go about acquiring it? Is it just first come, first serve? How finite actually is the space here? Just help me understand all these sort of questions.

SPEAKER_00

Yeah. Yeah, so roughly it's a bit like the high seas. Um, anyone can go anywhere, uh a little bit first come, first serve. Um, geostationary orbit is a bit different because there's a very defined orbit which is geostationary, and their slots are uh uh organized through the ITU, uh the International Telecommunications Union in Geneva, um, and that's allocated by country and sort of first come, first serve. You have to demonstrate that you can put a satellite in there, otherwise it just gets wasted, sort of thing. And that's the regime has worked pretty well. Um in the rest of orbits, uh, it's kind of uh anyone can go there. You everyone's trying to avoid collisions, of course. Um uh uh, but that's relatively tractable now. There's a lot more space than you think. Um we do have a problem of space debris, which is that um these are mainly not the satellites, they are leftover rocket bodies or satellites that have exploded. Um we didn't really understand this problem in the space community in the 60s, and and so Russia and the US were putting up lots of stuff and didn't really care where it went, and it it created a long-term challenge. And then subsequently, we've even had countries blowing up satellites to prove that they could take out the eyes of the other country. And in doing this game, they have made a big mess. Um and so we really need to stop that behavior because unlike these other domains of land, sea, and air, if you blow something up on the land, it just stays there on the land. If you blow something on the sea, it falls to the bottom of the sea. If you blow something in the air, it falls to the bottom, uh just underneath where the planes were. If you blow something up in space, bizarrely, they all the bits, it just the satellites fragment, fly through each other, and continue going eight kilometers a second, but just now in thousands of pieces. Well, that is a big problem. Um, and so imagine if every time there's a car crash on the road, the two cars splinter into thousands of pieces and continue on the motorway going 100 kilometers an hour for a few decades. Well, that would be really annoying for all the other travelers, right? So now, even though we've got about 10,000 satellites, we've got about a hundred million pieces of space debris. So for every one satellite, there's 10,000 objects, smaller bits. It could be a fleck of paint, it could be a bolt, it could be a you know, uh exploded bit of a satellite. And all of these pieces um can could cause a collisional cascade where the pieces bump into each other, then that runs run away, and then that creates more debris, which then causes so it's like a nuclear cascade. Um, this is called the Kessler syndrome. And um actually, in certain orbits, we're already in that. So we have to care for that problem. And when we do that, we need to do either ground-based lasers, which can nudge the debris if two pieces of debris are coming to each other, they nudge them so they miss and then slowly everything comes down, or you need to go and grab the big bits of debris um and so that there reduces the future growth. Um, there's actually a uh a project in in Mount Stromno Observatory that I uh and some colleagues at NASA helped to instigate years ago now, uh, that is you doing setting up one of these lasers to try and stop to do, if you like, space traffic management of all these debris pieces. Uh but it's a it's a real challenge, and the sooner we nip it in the bud, a bit like climate change, the better, because it's on a it's on a curve getting worse and worse.

SPEAKER_01

And also there's there's more tonnage of mass being put into space every single year as well. So it's only getting it. 100%.

SPEAKER_00

Yeah, although we've got at figuring out how to deal with those challenges. So it's more to do with the historical stuff than the new stuff.

SPEAKER_01

If you could make a very rough analogy, how big is the is the space where we can operate our satellites from?

SPEAKER_00

Yeah, so um obviously it expand the entire Earth in an area, and then the there's just a question of how much volume because of the depth, if it if you like. Most low Earth orbit satellites, which is most satellites, are between 400 kilometers and a thousand kilometers. Okay, so that's you know, I don't know, the distance between Sydney and Melbourne or no, Sydney and Canberra or something, um, are vol distant thick uh uh shell around the earth. And the and the area of it isn't is sort of like the area of the earth. So it's a huge volume. Remember, all of human life and all life on earth, most of it exists within plus or minus a few hundred meters of the earth's surface. It's either just below the surface or just above the surface or on the surface, right? It's not far away. So we think of that as the critical zone of life around the earth. So maybe that's hundreds of meters. This is hundreds of kilometers. So it's a much bigger volume than all of the earth for practical purposes. The earth as a volume is a huge mass, but we don't have access to it. We only have access, really, to our plus or minus a few hundred meters, most life most of the time. And so um this is hundreds of kilometers thick. So the volume is, you know, you could think of as a thousand times bigger than the volume of all life on Earth, you know? Something like that. So it's a huge volume of uh of space. Uh uh that's why we call it space.

SPEAKER_01

That's why, despite the enormous amount of debris, we don't hear news stories all the time about satellites being knocked out of the sky and so forth.

SPEAKER_00

It's still a very, very big space.

SPEAKER_01

Talk to me about Project Suncatcher. First of all, what what is what is planet's role in building data centers in space?

SPEAKER_00

Well, firstly, we've been putting up-edged compute, like GPUs on our satellites for a while now, mainly NVIDIA GPUs. And the reason is because we've got all this data coming in and we want to process it fast. Um and uh and so we want to put the computers right there to process all that data. Instead of sending down all the pictures and then doing the analysis, you could just automatically look at all the images and find the ships you want to find and send down those locations, or find that deforestation and send down that location, and then you could enable more real-time interventions. Think about disaster response. After the LA fires, we um uh gave pictures pretty quickly, and a few hours later we got an uh a building by building damage assessment across the whole of LA to help the responders. But imagine if we could have got that in a few minutes. That would have been much better for the first responders than a few hours, right? So um that would be better. So we've been doing that for a while, but um um but Google approached us uh last year about um starting putting some TPUs in space for compute in space itself. And now this started after uh we did a study with them about actually eight or nine years ago, where we looked at all the costs of data centers and all the costs of putting such computers in space and said, is there ever a time it would make sense to put them in orbit? And when might that time be? And we did a study, looked at all the costs, and it we determined that it would be cheaper, just purely economic grounds, it would be cheaper to put computers in space once launch costs get to around$200 to$300 a kilogram. Well, um launch costs have been plummeting. Now that about um the cost for a Falcon 9 to SpaceX and uh the number of kilograms means that it's about$1,000 a kilogram for them to put uh uh uh objects into orbit. Well, at that rate, we're we're only uh a factor of three off where, and we've already come down about a factor of twenty. Um so we're only about a factor of three off the place where it would actually just be cheaper, purely cost-wise, because you've got several advantages. You can have 24-7 power of your solar panels. They get five times more power, a solar panel in space, than on the ground, because you get it 24-7, you don't have the absorption of the atmosphere, you don't have the funky angles, you can just keep it perfect if you like. And um you don't have to have the land, you don't have to have the buildings, you don't have to have the cooling, you have to have cooling of a kind, not water cooling, but you can use radiative cooling to outer space. We know how to do that. We've done that many times before with our devices in space. You just put out metal uh sheets that can cool off to the darkness of space. We know how to do all these things. It was just really a question of launch costs. And launch costs coming down, and just getting that point. So, what we're doing with Google is doing a first tech demo. This is a big project, right? But we're just testing the technology. Um, it's a moonshot, as they like to say now, just like Waymo was before. And by the way, their TPUs were years ago as well. Um, and that was you know crazy to do their own chips, but they do their own chips now. Um so now uh we're on the start of that. And as Sundar said, within 10 years, he expects, uh, and I'd agree, um, that most compute will be being built in space within 10 years. Um and just to give you a sense, just think of the magnitude of that. We're spending trillions of dollars over the next few years over the big companies on compute infrastructure. So all that trillions is mainly going to be spent in space within 10 years. That would make it the biggest part of the space economy ever. It would be bigger than all the other components combined. So, this is of course a very interesting time to be a space geek, you know, because finally that dream that we've been talking about for years, as I said, we started that study eight or nine years ago, is coming to fruition. It's an idea whose time has come. Uh, early days, but it's exciting.

SPEAKER_01

It's incredible, Will. I mean, as a sci-fi geek yourself, having grown up reading about all of these wonderful futuristic ideas, I mean, how does it feel to have gotten to a point in your life when you can literally be one of the few contributors to what would be a sci-fi reality?

SPEAKER_00

Well, it's it uh it's it's it's it's exciting and and uh exhilarating, um, but also it's a it's a wave, right? We're just at the right place at the right time at some level. Um, you know, you make your opportunities, uh uh, but you also um uh it's a lot of luck. And uh um, you know, I was very lucky to have advisors and and get to NASA and all these other things, as were s the the thousand or so colleagues here I have at Planet who um are enabling uh are part of that dream to become a reality.

SPEAKER_01

Speaking of luck, what what are few moments of serendipity that have really determined the course of your life?

SPEAKER_00

Oh, so many. Um my friends uh undergrad used to say I had an unreality field. I've had so much luck, it's just crazy. Um but I'll give you uh a couple of examples. Um so I was when I was at NASA, first came to work here in Moffat Field. Um I uh was hiking uh with my girlfriend at the time um from near where the NASA center is to the sea. And I hike, we hike it's a long way actually, 70, 80 kilometers or so. So it was a two-day hike. And um we um we missed the last bus back, and and um who turns up on the beach uh kiteboarding, but Sergey and Larry, Paige from Google, who gave us a ride home. And then that turned into a friendship uh that um uh that uh you know we still we we we took them to NASA to see these lunar missions, and then after when we uh did our uh uh even today we work with them very closely on AI stuff, the Gemini team, Suncatcher, obviously. Um and but that was pretty happenstance. I mean, I was literally hiking to the uh, you know, on the weekend. Um and I I've had many uh cases like that. Uh so uh that's uh just but yeah, you it's the craziness of life.

SPEAKER_01

And you you've lived in a communal house for something like two decades, and also you live in San Francisco. Do these do these uh situations that you've put yourself in, because you're a Brit, you're not a you're not a you're not an American, right? You had to move there. And so you forced a bit of serendipity in your life by being around these places to come in, uh come across a Sergei and Larry, but also the communal house. Do you ever think about the forcing function of serendipity that you might be putting into your life? Hence your friends saying you've got the unreality field, but it's actually just your agency putting yourself there.

SPEAKER_00

Well, definitely at some level, you create your opportunities and you put yourself out there and you take those risks and you go to the places that um, you know, um those people might hang out. Um and our community houses do relate to that because they're hubs of intellectual innovation, the latest tech. I mean, you know, really I I often feel like when I go away from the Bay Area, I come back and I'm like, everyone's like six or twelve months behind what's going on here. And sometimes it's even more than that. In our community houses, we are talking about things that often don't become in the public discussion discourse for five or ten years. The attention economy, the issues to do with AI, the challenges at the interface of space and AI and biotech and all these things, and um, where that might be going technologically, what could that can enable, how that could create challenges, and how do we service those challenges so that we as a society can take care of them? Those are the things that we yeah, um that that community creates itself and it attracts other people that want to talk about those things. And so, of course, uh it's a bit of a uh a self-fulfilling prophecy there.

SPEAKER_01

And I imagine it's not only discussions about the future of technology, but rather philosophy and political discussions and the state of our culture and so forth, they must also be brought up as topics. What are you hearing from these communal houses that are five, ten years ahead of us? Where are we going to be?

SPEAKER_00

Well, we've been hearing about for many years the challenges that come with AI. Um, years ago, uh we were talking about social media uh misinformation, the attention economy, echo chambers, how we're naturally with the advertising model going to drive lots of misinformation online. And now we see it, right? I mean, it's just all around it. There is a clear incentive for misinformation online, uh, according to our the logic of the system that we we've incentivized that. And that has hurt democracy, and we talked about that. Um and I still think that's a problem. But where the big one that's coming up now is, of course, the AI explosion that may come. I don't mean the job loss. I mean, that was uh a discussion some time ago, maybe 10 years ago, but now it's about where does recursive self-improvement go? If you could close the loop entirely on AI, what does that mean? It will likely mean, we don't know for sure, that an intelligence explosion. And an intelligent explosion that could have a timescale that's very short, way short. I mean, already now with the human in the loop programming things, um, it's uh, you know, a few every few months AI is changing quite dramatically in capability. But you hook it up to itself so it improves itself, that could suddenly change to two days or two hours or two minutes, right? We don't know, we don't have a map of what's on the other side of an intelligence explosion. We don't know whether it will explode or whether it will just flop. We don't know whether humans will be in control of it or whether they will not, or whether the AI will be controlling itself. We don't know if AI is controlling, uh, whether Humans, will it be in the interests of humans and the rest of life on the planet or not? We don't know. And anyone that tells you they know the answer to that question is wrong. That is a gargantuan gamble to be doing. And it's it's so I think that AI has incredible promises. We were talking here about AI and satellite data and all the promises. But we as a species have to confront the significant potential for um real risks of all kinds, troublingness, um uh you know, um fraud, simple things like that, to the to the existential ones like human control and sovereignty. And, you know, it might not be about control per se. I think control is a loaded word, but are we gonna have a symbiotic future with that AI in a way that's good for it and us, or not? And no one has shown me a strategy that looks like roughly we get the other side of that intelligence explosion with a clear game plan for ensuring that humans and the rest of life are good, roughly. Like no one has such a game plan. And that is a huge decision to be made. And I think at the very least, one could say, without any uh philosophical doubt, that that is a decision that should be made with a democratic involvement of a lot of people. It is not a decision that should be made by a couple of people. And I say that as a technologist, and the consensus on AI is very unusual. Like we don't have like all the leaders of all the space companies saying, oh, by the way, guys, if we do this, it could lead to the end of humanity. Right? We don't have all the uh nuclear fusion labs saying that, we don't have all of the biotech labs saying that, we don't have all of the any labs saying that, except for AI, where all the AI leaders are saying in unison, you can look up all the quotes from all the AI lab leaders. They all warn that there's existential challenges here. So it's not just a, you know, there's unusual consensus around this amongst anyone that thinks about it, um, really. And I'm not talking about being doomsdaying or uh it's all sunshine and rainbows. I'm just saying it's unknown. And all of them say that it's really unknown and there's real dangers there. And so that I think the gestalt here is that that is something that is way past its due date to be uh tackled. And so um uh I would say that's that's the thing humans need to get on with with great urgency. And we need uh red lines and we need red teaming um to figure it out. And we need we're gonna need AI to help us to do it because it's such a new problem.

SPEAKER_01

Given the unknown nature of it, you cannot predict a future of infinite possibilities based off our finite experience of the past. Like given how consequential it is, do you yourself make a judgment across the various AI labs and the people leading them and the way that they're behaving in public to the enormous scrutiny that they're under? It seems like it's just the safety rails have been put to the side, and it is simply a question of market forces. Who will win? Who has the biggest funding around? Who has the most clients, who's got the best enterprise deals, and that that market function is just too powerful for any other philosophical consideration to be taken into account, and it's not democratic, you know, it's it's it's it's a it's a handful of extraordinarily powerful, rich people in San Francisco.

SPEAKER_00

It's kind of wild. It is wild. I mean, well, look, they're in a race dynamic, right? Both between the companies um and between the countries, especially US and China. Um, and it's hard to escape a race dynamic. Uh, but we have done it before. Um, you know, the common uh challenge raised with any uh thing that questions a no-holds barred approach to AI is well, we can't trust the other side. We can't trust the other company that they were doing, or we can't trust the other country that they will uphold the same things. But I'll point out that we didn't trust, uh there wasn't great trust between the US and the Soviet Union, and yet we all agreed to reduce nuclear weapons stocks, uh, take them off the hair trigger, all sorts of things, which potentially are partly why we're here today and not uh it didn't obliterate ourselves. Um, several cases we came close, right? And had those rules not been in place, it might have been worse, right? We we obviously don't know, but that that uh is true. So humans have overcome this kind of challenge before. Not to say nukes and AI are the same, but definitely we know how to decide to curtail things and not to have to trust one another, but to verify one another. We did all this verification, by the way, the US spy agency, the NRO, was established in part to monitor the Soviet nuclear stockpiles, i.e., we invented an entire new agency with new technology, it's just in order to verify it. So it wasn't easy, it was extremely hard. So, yes, when people say, oh my gosh, we can't verify, you know, we can't trust the other company, we can't trust the other country, really they mean it's very hard to. And they're right, it's hard to. But hard is not impossible. And we have done hard things before, and when the um when the opposite possi when the uh current path is insanity, um, then you should try the hard but possible path.

SPEAKER_01

But given the absurd nature of history, this moment that calls for deep institutional robustness, the leader of your nation, well not your nation, but rather the the person who's in charge of America, where all the AI labs happen to be, is as anti-institution as anyone has been, probably in that country's history.

SPEAKER_00

Yeah, well, I mean, look, I don't want to get into the politics of this. I I I just want to say these are a part of it.

SPEAKER_01

They they're fundamentally.

SPEAKER_00

For sure. But I I what I'm saying is that um I don't think one has to be talking about partisanship to talk about the the fact that there's real um challenges here of this technology. There's upsides and there's there's challenges. And we need to be thoughtful about it. That's the same for every new technology that's ever come across humans. With before you get into any of the politics, that is true. Uh, but this is particularly unique because we could have an intelligence explosion, which is a one-way door to a future where we just don't have a map of how we ensure humans are safe on the other side. And to s to just blast through that without um thinking through what that strategy might be seems uh uh to be a huge gamble with life. And that's everyone's life, no matter what your political subscription, and by the way, not w no matter what your species, because there's a we're taking all of all of people on this planet with us on that journey.

SPEAKER_01

Let's round out AI, Will. Do you agree with your former supervisor, Sir Roger Penrose's skepticism about AI ever being able to achieve consciousness?

SPEAKER_00

No, I do not. I have quite a different. He's a genius. He is a genius, but that doesn't mean he's always right. Um, and I'm not saying I'm I'm right, obviously, uh we'll find out, but I'm reminded of the Star Trek episode um Um The Meaning of Man, I think it was, um, where um they questioned data on whether or not he's uh property of the Federation and therefore could be taken apart or not. And uh Picard comes to his defense and basically said, Well, there's three tests. Is it intelligent? Clearly, data is intelligent. Is it self-aware? Clearly, data is self-aware. Is it conscious? We don't know because we don't really have a good definition of consciousness ourselves, kind of thing. But at the very least, we could say, because it passes those other two tests clearly, we should at least afford it the ability to try to choose. We don't know, but it who it deserves that. I think at some point we'll find ourselves in that conundrum with AI. The idea, you know, we'll we we may prove that it doesn't have uh consciousness only to prove that we don't either. Um, or we might prove it does, uh um and then but either way, I think we'll we will end up being on the same basic um uh state uh as as as as AI. Um, you know, I'm I'm a bit of a functionalist in that sense. If it does all the things of a of a human and more, obviously it can do super calculations too, but if it can do all the generic things, which we're already starting to see in significant ways with our uh LLMs, um then at some point, if it walks like a duck and quacks like a duck, it is a duck, you know? Um and you know, we cannot discount the possibility that it will have consciousness at some point. Um so we're obviously not there now. It's very mechanical, it's semantic, it's just guessing the next words in the sentence, and it's doing a very good job. But ultimately, we might realize that we're doing the same thing. And so uh uh again, either you're faced with downgrading humans or upgrading AI. I don't think we're gonna be faced with we are clearly superior to AI in a generic way. I don't believe the arguments of Roger and others that do that. So, but but it's obviously no one knows the answer to that question. It's uh it's a debate.

SPEAKER_01

One of the great debates that at least we're lucky enough to be alive as uh contributors to them and observers in real time of actually what is the outcome going to be.

SPEAKER_00

I I want to mention times to be alive. I mean, I was recently saying there's we're we're gonna see perhaps three Copernican moments in this decade. The first is that we will give birth to superintelligence on this planet. That's amazing. Uh intelligence generally smarter than us, uh, that's obviously AI. The second is that we're gonna discover life off Earth. I think it's very good uh probability, even though I was talking about the cosmic significance of our life, because it's still very rare. I think it's a very good probability we'd find life on some of these, I think, 6,000 or 7,000 planets we've already found. Because now we've got these telescopes coming online that will be able to detect their atmospheric contributions. And I'm pretty sure one of them will go, What on earth is that? Uh just like if you were to look from there back at Earth, you would go, why is there 20% oxygen? That's crazy. There's only one way that that would be around, that's because of all life. So we the best explanation for some planet atmospheric contribution will be life. And the third is that we can decode conversations between animals. Uh, whales have been talking to each other for 70 million years, the chimps have been talking to each other for a few million years. What are they saying? Come on, let's just decode that. And it may be simple, especially for your dog. Uh, give me some food, you know, it's nice to see you again. Or it could be more complex, but man, in each case, those three things are going to decenter humans from being uh at the middle, like we did in the Copernican moment. Um, and so what an incredible time to be alive that those three things could happen, and we get to be observers to that. Hopefully, uh we bring some wisdom to those transitions, and and those transitions help us become wiser.

SPEAKER_01

Amen. Lovely, Will. Let's just finish out on one more. Uh, you publish these incredible photos around the world, and it is from a stylistic sort of artistic perspective. They're beautiful shots of nature. So I wonder in this library of fantastic aerial shots you hold, what's a part of the world that we wouldn't typically think of as beautiful, but when appreciated from above, it just becomes breathtaking.

SPEAKER_00

Hmm. Uh gosh. I mean, you know, I I have on my wall a place near your hood, um, uh which is, you know, up towards um, of course, the Great Barrier Reef, but it's not just the thing you think of, it's all these incredible at a slightly bigger scale, these sort of swirling um reef patterns. Um I find that just, and you know it's not quite clear where land ends and starts and where the water starts, and and uh it sort of all blends into each other. And you could be looking at something that's uh this big leaf, or you could be looking at something that's a thousand kilometers across. There's this sort of sort of symmetry across uh big spatial uh scales. So I find that beautiful uh to look at, and that's why I have it on my wall.

SPEAKER_01

Will, thank you so much for being so generous with your time. Really appreciate it.