Closing Bell - Manifest Space: AI Powered Predictions with Planet Labs CEO Will Marshall 10/5/23

Episode Date: October 5, 2023

Planet Labs, boasting the largest constellation of earth imaging satellites in orbit, generates thousands of images for governments and companies documenting changes over time. Can that data, powered by AI, be used to predict future events? Co-founder & CEO Will Marshall joins Morgan Brennan to discuss predictive analytics, satellites under development & the outlook for the earth imaging market.

Planet Labs CEO and co-founder Will Marshall first joined this podcast shortly after Russia's invasion of Ukraine last year to discuss the Earth Imaging Company's role in capturing and publicly releasing satellite images of the Russian military buildup on the border and subsequent attack. Planet, which boasts the largest constellation of Earth imaging satellites in orbit, continues to monitor the globe for governments and companies alike. The space company is generating so much data, Marshall says generative AI is a natural fit. The more exciting thing right now with AI is actually applying it on our historical archive, because we have over 2,000 images for every point on the Earth's landmass documenting change over time. So we can train AI on top of that to then do analysis of what is new on a day-to-day basis. It means Planet can offer space-based data

that enables more analysis and even prevention of things like natural disasters, for example, wildfires. Nonetheless, shares of Planet Labs have had a tough run, down more than 40% this year, more than 55% over the last 12 months, as the company has cut full-year revenue guidance twice. On this episode, we dive into AI-powered predictive analytics, the next-gen satellites under development, and whether investors grasp where this Earth imaging

market is headed. I'm Morgan Brennan, and this is Manifest Space. You're staying busy at Planet. There's a lot for us to talk about. But first, let's start with maybe the fact that, you know, you've had, you've inked a number of contracts, a number of new deals, you extended an agreement with NASA recently, I guess, I guess, break it down for me. Yeah, I mean, we have three main go to market segments. One is commercial, one is civil government, and one is defense and intelligence. And so on the those, yeah, we on the first, we've done a whole bunch of

deals in the agricultural space. For example, we work with companies like Bayer and BASF on improving crop yields and precision farming, it's called. It's a lot to do with digital transformation of those sectors. With civil government, you mentioned NASA, yeah, we just extended our partnership with NASA as a fantastic arrangement that has enabled scientists across their organization and then across universities across the United States to get access to our data. And it has led to incredible innovation that is enabling and underpinning applications. So it's not just

scientific research about what's going on on the planet, but it's also underpins things like those agriculture, disaster response and other use cases for other people. And then on the defence and intelligence side, we are working with a number of organisations around the planet to really, it's about finding new threats, it's about monitoring security. For example, we've been working in Ukraine, and there it's been across various aspects of that. But in general, transparency in all three of these domains is helping to bring about accountability. I remember the last time you and I spoke for this podcast, actually, you know, Ukraine had just happened. Russia just invaded Ukraine and it really sort of thrust Planet into the international spotlight in the sense that

you're putting a lot of material, a lot of images out saying, hey, listen, this invasion is coming. Here's how it's playing out in basically real time. And for better or worse, you sort of become the go-to data for what's going on, you know, good and bad across the globe and sort of calling out lies, for example, with the Russia-Ukraine situation. Has that just, I guess, have you seen your prominence and demand for the products you have increased because of that? To some extent, I mean, certainly this is shedding light on the power of these tools. I mean, we're trying to do whatever we can to help, for sure. And it's helping in three main ways.

It's helping advance peace and security through operational help. It's helping the humanitarian operations of many NGOs that are doing crisis response, bringing medical supplies and so on. We've worked with the United Nations to do a damage assessment of all the buildings across Ukraine. What will it cost to rebuild the country? We did a lot of work through NASA Harvest with assessing crop yields. Then it's really important because 40% of the grain

for the poorest countries in the world comes from Ukraine. And so their production of crops is not just important for Ukrainians, it is also important for the rest of the world. Assessing that, assessing the impact on other countries that are dependent on those grains is really important. So we've been trying to do some of that work and release that. And to your point about accountability, we have in many cases shown accountability. You know, we have many cases shown accountability in the case of the grain

found when ships have been stealing grain against sanctions, taking it from the Russian controlled piece of Ukraine to Syria or Turkey against those embargoes. And then that's helping to shed light for everyone to see what's going on. I mean, I think the broader sense here is that this technology is shifting a new level of transparency to these sorts of events. And that brings accountability. And I think that's important because it advances peace and security because it helps open societies. It brings accountability. And I think it coheres society because we provide that imagery also where appropriate to news media, which is what you often see online when you see information about what's going on in the war.

And that helps cohere society. It reduces the gap between what the politicians know and what everyone knows so that we can all. So it's part of the if you like the hearts and minds of us all staying behind Ukraine in this situation. I have to ask one more question on this and then I'm going to I'm going to move on. And that is, you know, what what are you seeing based on your data and on the imagery you're collecting off your satellites where Ukraine is concerned right now? There's been a lot of focus on the counteroffensive, you know, in recent months

and just sort of the state of the conflict. Yeah, well, I mean, I'm not an expert on the conflict, but, you know, our imagery is used by myriad organizations. And, you know, it's obviously a complicated affair and it's going to take time. It's slow progress. And, but, you know But what we're focused on is whatever we can do to help bring that transparency, bring that accountability to the situation

and really working with our partners at news media, think tanks and governments that are trying to help. Okay. So I guess you mentioned that three different buckets. Where are you seeing the most demand and the most growth right now then? I would say probably the biggest demand is in civil government. We have done a whole bunch of collaborative partnerships. We work with the state of New Mexico on regulation enforcement, Humboldt County on marijuana growth enforcement, all sorts of things like that. But recently, what's been driving it most of all has been disaster response. So this is, you know, we've been trying as best we can to help in situations where there's disasters.

Take, for example, the fire in Maui. We were trying to be as quickly as possible assessing what buildings were damaged across Lahaina, trying to get that data into the hands of the first responders, like the Red Cross and others that were on the ground trying to help. They can then determine where to put relief supplies and so on. That's that's quickest and trying to help save lives. We've been working with three Canadian provinces on fire response there because they've been having terrible wildfires as well. But then once we get into that piece, we then quickly graduate into a discussion of how can we do preventative work in this sort of territory? So not just response to a fire, but how can we get ahead of it?

Our data enables us to scan and find all the actually we can measure biomass through our different spectral bands on our satellites. And this enables us to tell where's the stock of wood growth that could be the source of a mega fire? Where might you want to put a fire lane that might cut that off and reduce the probability of a big fire? And so we work with governments like these provincial governments in Canada to try and mitigate and reduce the potential. And, you know, the world is spending hundreds of billions, the US alone is spending a few hundred billion a year on disaster response, the EU similarly. And if you can get ahead of some of that, you can obviously save billions of dollars if you could reduce. In the

case of Lahaina, let me just get a bit more specific. We could see that there were invasive species in that area, that the soil moisture was very dry. We can actually tell the amount of moisture in the soil and we can see it was very dry. We could see that there were brush too close to power lines. That's the kind of thing you can get ahead of and clear so that you reduce the danger of wildfires. So we are trying to not just help with support afterwards, but preventative work as well. And that's of huge value to civil governments around the planet. It raises the question, what's the role that AI plays in all of this, especially when you start going from

defensive to preventative? Yeah, absolutely. I mean, actually, AI is doing more and more in predictive analytics, to your point. Soon you'll be able to just ask it where will where's the likely uh sources of megafires or where's the likely sources of drought as i mentioned we have this data on soil moisture across the whole planet every day we can measure the soil moisture and if it's too dry you might worry about fires and bushfires. If it's too wet, if it's saturated, you know that the next time it rains, it's going to flood. And already we can start getting a little bit predictive. But AI is going to take that to the next level in terms of showing where and when the most likely areas are. But I think the more exciting thing right now with AI is actually applying it on our historical archive, because we have over 2000 images for every point on the

Earth's landmass, documenting change over time. So we can train AI on top of that to then do analysis of what is new on a day to day basis. And let me get a little bit more concrete. In history, for years, we've been using machine learning to do what we call object identification. So in any image, here's a building, here's a road, here's a ship, here's a plane, count those objects. How many are there? How many over time, what's the changes. But each such model for houses or planes or ships or roads took months to develop, figuring out how to work in uniform ways across the planet. And we did indeed, after some time in those sort of efforts, do a global map of all the world's roads, all the world's buildings,

which I think was the first time that had ever been done. And we use that operationally to help customers. Take, for example, our customers in Brazil. We work with federal police there and we monitor the entire Amazon every week and do automatic searches for new roads. If there's any new roads, roads are normally the early signs of deforestation because the clear cutting, normally first there's a road encroachment into the forest and then they start cutting down trees. So we know that roads are an early sign of deforestation. So we do this weekly review of all new roads with this machine learning and then help them to determine where to go and stop illegal deforestation. And that's having a real impact

on helping us keep the Amazon. But that's what we call classical machine learning. Where we're going now with large language models, with generative AI is a bit of a different place. And it is shocked even me, just the progress that has happened in the last few months. So really, this has enabled us to from training models that take months to train to being able to train once and then search for anything within minutes across large territories. So the first use case of this was the balloon, the Chinese spy balloon that was floating across the United States. And some people in the U.S. government said, hey, do you know how to find that? And, well, we didn't have a model trained on finding balloons.

We worked with our partners at Synthetic, who've been using some of these large language models on our data. And we quickly together were able to find the balloon, and trace it back in time through our archive to where, through time it had traveled all the way back to where it was launched in Hainan province in China and that is incredible power. Now the next question is where are all the balloon flights across all the world through time and can you alert me if there's any new balloon flights? That power is huge and suddenly in the hands of an individual can do a search across a whole continent for objects.

And so we're getting much closer to a vision that I painted out in a TED talk five years ago called Queryable Earth. Whilst Google have figured out how to index the Internet and make it searchable, we're figuring out how to index the Earth and make it searchable. And the power of that is tremendous. I thought that was many years off and now it seems months off. Wow. I have many questions for you. The first one though,

I'm going to pick up on what you said and that's synthetic data. And I know you work with startups like Rendered AI. What is synthetic data? Synthetic is a company. Oh, do you mean synthetic data? Because sometimes we generate synthetic data? Synthetic is a company. Oh, do you mean synthetic data? Because sometimes we generate synthetic data, like on our Carbon Mapper mission, for people to test before the satellite gets launched. Or do you mean synthetic, the company that's doing AI work?

Oh, I meant the former. Okay, yeah. So in the case of our work on Carbon Mapper, this is our hyperspectral set of satellites. We call ground, some modeling methods so that we can give that data. And we have a set, a class of about 20 early users before the satellite is even launched to test out that synthetic data to get ahead and start training their AI and algorithms on top of that data. So that by the time we launch the satellite, they're ready to go. Okay. And so when we, I guess I'm just trying to understand when we talk about like synthetic data,

we're talking about like, we're talking about, you know, use cases that haven't, that don't exist or like data that doesn't actually exist or that's been like artificially created to help. Okay. Yeah, exactly. And the thing is that it isn't really a commercial hyperspectral market yet. We're generating something new here.

The hope, the main mission of this, hence the name Carbon Mapper, is to track carbon emissions around the planet, methane and CO2 emissions. It also has a wide variety of other applications, biodiversity tracking, really huge, defense and intelligence, because it can see from the paint, what kind of vehicle, where did it come from, because it's so accurate at detecting objects, that it can do things like that. So there's a lot of applications, but we have to explore that. There hasn't been systems like this in the commercial realm before. And so, yeah, we're priming the pump with synthetic data. How does this speak to where the planet fleet is going? The constellation of satellites is going. I mean, you have the largest constellation orbiting the Earth right now for

Earth imaging capabilities. You just mentioned Tanager. You've got Pelican under development as well. How does all of this continue to evolve? Yeah, I mean, basically, our job at Planet is to continue to have the best data in orbit. We want to continue to innovate. So that means getting better on the three ways you can improve satellite data,

the three kinds of resolution. Spatial resolution, that's the one can improve satellite data, the three kinds of resolution. Spatial resolution, that's the one you typically think about the size of the pixels, you want to get better and better spatial resolution, more accurate and accurate imagery. Secondly, temporal resolution, how quickly do we revisit places and how quickly can we get the imagery back? Because for example in disaster response we'd love that image instead of a few hours. If we could get it in 30 minutes, that would be better, right? And so that's the direction that we're going with Pelican. And then the final thing is spectral resolution, that is the number of color bands, and that's the direction we're going with Tanager.

Our goal is to continue to improve in all three of those axes. But I would hasten to add that whilst that's a core backend, what's perhaps more exciting is how we're going up the stack to enable people to make it easier to use, to get access and to get value from that data. So a lot of our work in recent time is more on the software stack that sits on top of the satellite imagery to make it easier to extract value. So it's this idea of software as a service, but space-based software as a service or space-based data as software as a service.

Absolutely. So Planet is a satellite company, but it's evolved over time from just a satellite company to really a satellite, a data company. Our business is selling data streams on via a platform. And this is why I use the analogy to Bloomberg, I think it really to planet is is analogous to Bloomberg, Bloomberg's serves up data feeds that help people make smarter decisions, right? Mainly in the financial sector, stack tracking stocks, and so on. We're doing similarly, except our data is earth imagery and analytics that sits on top of that. But similarly, it's provided by a platform, you can mix and match it, and it serves and helps you make decisions for smarter resource allocation. I think the only difference is that we sit on a proprietary data set, whereas Bloomberg primarily aggregates public source stock information from exchanges and so on.

We collect it using a fleet of satellites. So it's very hard, very hard to replicate. But just like Bloomberg, those data feeds go deep into people's decision making across those government and commercial users that we have. And that enables them to make smarter decisions. And it frankly makes it very sticky for a planet as a product. Would you ever go beyond just imaging and Earth observation? I mean, there's other capabilities and other satellites, other startups that are in this space, but not necessarily direct competitors with you as well. How do you think about the marketplace evolving and becoming a one-stop shop, essentially?

Well, we did call it Planet for a reason, to not have limits to our ambitions, not even just Earth. But in all seriousness, we really are mainly focused on Earth imaging. I mean, look, I think there's a massive market opportunity. The world is undergoing a digital transformation, all those industries like agriculture, as they figure out how to become more efficient using digital technologies, we're key to that. The world needs to move towards a sustainable planet, which is a multi-trillion dollar transition that planet can help, not just with disaster response, but monitoring ecosystems, valuing carbon that enables us to do carbon markets, that all needs data. And then finally, on peace and security, challenges around the world of incredible import

and exacerbated by the sustainability challenges and the climate challenges where you have more refugee crisis, more climate-driven droughts and so on. Planets needed more than ever before. And I think that's a massive opportunity for us to go after. It's worthy of a standalone company. We don't need to go into lots of different directions to do that. We're already sitting on a massive data archive.

We mentioned at our learning school, we've built a huge pipeline of opportunities, over 70 deals, more than a million dollars sitting in our pipeline. We've got to stay focused on executing against that and not getting distracted by going to Mars or something. Gotcha. How big is the market? I mean, you said it's a huge market do you sort of have a number in your head um yeah i mean we've we've talked about the fact that we think it's about a hundred billion uh the planet can go after as a market opportunity it's a little bit hard to estimate of course because it doesn't exist it's a bit like asking google how big's the search market before google existed right um there There isn't that market today,

we are making it. But if you think about, let me just take one example. We have to move to a sustainable economy, we are our economy, our system of society is battling against the planetary boundaries, but not just on climate, but also on biodiversity loss and other challenges. And we need to fix that. And the best way to fix that is to judo move capitalism to value carbon and nature into its economic system. Governments are going to force this by saying, hey, every company, you're going to need to balance not just your books in terms of finances, you're going to need to balance your carbon books. What's your input and output? And people are already doing this at some level. Microsoft is trying to offset its carbon. Many other companies are trying to do the same. But soon it's not going to become a voluntary

thing. It's going to become a regulatory enforced thing. The EU is already moving in that direction in many ways. Even the SEC in the US is pushing for the same. Companies are going to be forced to do that. Now, the next step. So a company says, well, I'm going to get to carbon neutral at the end of the year. Well, the first thing they're going to need to do is how much carbon are they using?

Where is it? How do I track it so that they know how to reduce it? We are now launching later this year a very exciting thing called a forest carbon planetary variable. Basically, not just we have forest data that we use to help 64 countries stop deforestation, all the tropical belt countries, but now we're going to start measuring the amount of carbon in each tree around the Earth on a quarterly basis, enabling us to underpin carbon markets. I saw a Financial Times article the other day that estimated by some economists that it's about a hundred trillion dollar transition, about three trillion dollars a year for about 30 years to transition to a sustainable economy. And everyone's going to need

to measure their carbon. And Planet is going to start with the forest carbon, but move on to other aspects of carbon, blue carbon, the ocean carbon, and seagrass, and kelp, and so on. Brown carbon, the carbon in the soil, we're already measuring the water in the soil, we want to measure the carbon in the soil. It's also measuring emissions using the Tanager spacecraft, the carbon mapper program that I mentioned earlier. Between them, we should be able to, for every hectare on the Earth, measure the carbon inputs and outputs. And that's going to be what enables people to put the carbon on their balance sheet. I checked the other day and Ernst & Young, the accounting firm, is about a $50 billion business, revenue business annually. And they're doing accounting for people's finances.

Well, there's a whole business opportunity in accounting for people's carbon balances and helping people to measure that. And Planet is in an incredible position to underpin that kind of sector. You think investors fully appreciate or realize where you're going with this? Nope. Next question. No, absolutely not. I mean, look, I think they understand that it's cool data,

that it's new, that it has value. I mean, when they see that stuff that's helping Ukraine, when we see the disaster response in the news and our imagery is there, there's a value proposition. But fundamentally, the scale of it, I think, is not understood. And this is why I keep on talking about those tailwinds, that digital transformation, sustainability, transformation, because that's the scale of the opportunity for planet to help and uh yeah i mean

you know we're in a transition but i think the the investors in the market hasn't fully understood that the power of these tools i think also you know we are seeing now we talk about ai ai is a big accelerant of that um it accelerates we've've got this huge stack of data. What a lot of investors, I don't think, fully understand is that there's a lot of power in algorithms and that can help improve what you can find in the value extraction from data. But honestly, data is where it's all at. The economists quit that data is the new oil because just like oil, it underpins lots of sectors. Data is going to underins lots of sectors. Data is going to underpin lots of sectors. And it's the data that's really unique. You'll see some of the

large tech companies open sourcing their AI algorithms. You will never see them open source their data. And there's a really good reason for that. That's their differentiated capability. That's how you train your algorithms to do new things that no one else can. Planet is sitting on an archive of a massive amount of Earth imaging data with huge power to help the world. And so AI accelerates our access to that. But it's really the data that's the bigger, powerful force than the algorithms now, because the algorithms are somewhat being commodified. Okay, so you've laid out the picture, you've laid out the vision, you've laid out where the company's going. I do want to ask, though, because to your point about, you know, sort of whether Wall Street has, you know, fully grasped this or not, I mean, you did trim your guidance for the

year, and I realize you're still growing top line double digit percentages. Why is it macroeconomic uncertainty is the fact that, you know, it's this idea of keeping up with demand in the pipeline. I guess. How do you see the near term versus this longer term picture? It's a great, great question. Look, I think fundamentally it's about execution. We have faced some headwinds coming to the year. A few big deals that changed and moved out. And then there is some economic headwinds on the commercial sector of our business where deals take a little bit longer. But honestly, the opportunity still remains as it always has.

The scale of the opportunity is huge. I mentioned the 70 deals over a million dollars in our pipeline. We've got, I think, five over eight figure deals, over $10 million in our pipeline. We've got a huge pipeline. But it's our job to execute against that, right? So it's our job to, and that's about more efficient sales process it's about focusing down it's and and we we did some restructuring early in the year partially because we need to stay really honed and focused on the growth areas that are going to work look it's not easy and it's a

tricky economic environment but planet has a huge opportunity to go after. And I think the way I think about it is in the long term. So Planet is helping those big tailwinds. If you think about what happened at the last economic downturn, the companies that came out of that strong are the biggest companies in the world today. What do you think are going to be the attributes of companies that come out of this economic situation strong. I think there's going to be two things. One is they're going to help us with the sustainability transition, because the world has to undergo that. The second is that they're relevant to AI, because AI is driving the next economy. And I believe Planet is well situated in both of those

regards, right, to tap into that opportunity. So that's the exciting thing that I'm focused on. And thankfully, we have a fantastic team. We have plenty of capital in the bank after we went public to get to profitability. But we've just got to stay focused on executing effectively and efficiently towards our growth goals. That does it for this episode of Manifest Space. Make sure you never miss a launch by following us wherever you get your podcasts

and by watching our coverage on Closing Bell Overtime. I'm Morgan Brennan.