The following is my own analysis of Planet Labs. This is not investment advice.

———————————————————————

Planet Labs images the whole world every day to make global change visible, accessible, and actionable. To understand how, let’s explore the two parts of their business: imaging and data analysis.

In 2013, Planet began building and launching small satellites into low-earth orbit. Their goal was to establish a satellite network that could photograph earth’s entire landmass each day.

By 2017, Planet achieved this goal, and had more satellites in orbit than any other company.

Today, Planet has two mature satellite networks in low-earth orbit, called PlanetScope and SkySat. Each performs different tasks:

Planet built the satellites in their PlanetScope constellation, and acquired their SkySat network from Terra Bella in 2017. Planet also acquired a satellite image archive dating back to 2009 from their 2015 acquisition of BlackBridge.

Now that Planet has a full network of earth scanning satellites in orbit, their focus is on improving their imaging capabilities with two new products, Pelican and Tanager.

Pelican is a direct improvement upon SkySat. It will be a fleet of 30 satellites, with image resolutions of 30cm/pixel and the ability to re-visit locations every 30 minutes.

Tanager is a new type of hyperspectral satellite that will be able to see over 400 spectral bands, which can uncover new information that other satellites cannot detect. One example is Tanager’s ability to see and measure methane as it is released into the atmosphere.

To really understand Planet’s imaging product strategy, consider how humans observe the world.

Our eyes have both low-resolution peripheral vision and the ability to zoom in on specific objects with high-resolution. We also have senses (hearing, smell, taste, and touch) which give us data that our eyes cannot capture.

PlanetScope is the low-resolution peripheral vision, SkySat (and Pelican) enable high-resolution visibility of specific spots, and Tanager will detect new kinds of data other satellites cannot.

The earth’s surface is far too large for humans to thoroughly scan each day, which makes Planet’s software such important part of their business. After all, what good is scanning the entire world each day if you can’t make sense of those changes?

Planet’s software tools combine, refine, package, and analyze raw imagery to make global change visible, accessible, and actionable. Below is a list of Planet’s key data analysis products.

Note: Planet has announced a new data product on April 11th, which may make this list irrelevant.

Planet Basemaps are a mosaic of images stitched together to form consistent views of the earth. Planet’s “best-scene-on-top” method filters out abnormalities like clouds or haze, using only the best available image for each piece of a Basemap.

Planet offers two different types of Basemaps:

Basemaps help customers analyze evolving landscapes without spending time on image pre-processing. The consistency and clarity of Planet Basemaps is also well-suited for machine learning.

While Basemaps do improve upon the quality of Planet’s raw images, they have longer processing times so they may not be ideal for tracking rapidly developing situations.

Just like Google displays estimated commute times, points of interest, and road markers on their maps, Planetary Variables combine satellite data to display measurements for things like soil water content, crop biomass, land surface temperature, forest carbon, and field boundaries.

By layering this quantifiable data on top of their images, Planet customers can closely monitor the variables that matter to them and take action faster than ever.

Planet’s acquisition of VanderSat in November 2021 was the key to launching their Planetary Variables product, and their acquisition of Salo Sciences in early 2023 led to the addition of a new Forest Carbon Planetary Variable.

Last year, Planet went a step further in their quest to make their images more accessible by teasing an AI collaboration with Microsoft called Queryable California.

Their demo showed how Planet’s data can leverage AI to interpret a text prompt, browse their image library for an answer, and layer the result on top of a map wrapped in a familiar search interface.

By abstracting away all the complexity of scanning a map, interpreting data, and calculating results, this demo marks a big leap forward in Planet’s quest to make global change accessible.

In August 2023, Planet acquired Sinergise, the developer of a cloud platform called Sentinel Hub.

Sentinel Hub leverages data from dozens of satellites, and is a widely used tool for browsing, analyzing, and integrating earth observation data.

Their most approachable tool is their Earth Observation browser, which anyone can use to analyze time series data, mark areas of interest, get spectral band measurements, view elevation profiles, and even create timelapse animations.

Advanced users can also create new visualizations with a drag-and-drop spectral band interface or a custom script.

Planet’s software tools each serve different customers, but they all share the same goal of making global change visible, accessible, and actionable.

If Planet’s satellites are designed to resemble our five senses, Planet’s software is designed to resemble our brain. By combining Planet’s historical images with AI, customers can easily recognize changes in the world and take action. In a similar way, our brain relies on our library of past experiences and lessons learned to make decisions and take action.

Just like how our five senses are only useful when connected to our brain, Planet’s imaging satellites are only useful when connected to software tools that can make sense of the data they collect.

Now that we have an understanding of Planet’s business, let’s explore some of the unique advantages Planet has, and some potential challenges the business could face.

Nobody cares about yesterday’s weather. If you don’t believe me, open your weather app and try to find it. Chances are you’ll see current conditions and future forecasts, but no information on the past.

At least it appears that way.

Dig a little deeper, and you’ll realize current conditions are not really “current”, they’re just the most recent historical weather readings. Likewise, future forecasts are not really from the future, they’re just predictions based on past weather patterns.

The lesson is that while nobody cares about yesterday’s weather, certain collections of past data can produce incredibly useful insights that help us make better decisions.

And for data companies, that’s where the money is.

Facebook, Google, The Weather Network, and Planet are all playing different versions of the same game. They recognize that pricing power does not come from collecting historical data, but from the useful insights, predictions, or outcomes for customers that their data can uncover.

Planet already has some experience generating predictive insights for customers, and I suspect they are only scratching the surface of the insights their data can generate for customers.

One example is Planet’s contract with the Brazilian Federal Police.

Each week, Planet scans Brazil’s jungles to search for new roads. Their satellites identify new roads as they are built, and the roads are leading indicators for future drug smuggling routes and illegal deforestation.

This means Brazil’s Federal Police can use Planet data to thwart drug smuggling or illegal deforestation before they happen. In the first year of the program, Brazil has eliminated 55% of their illegal deforestation and has collected over $2 billion from fines, seized goods, and frozen assets.

If you were Brazil’s president, how much would you pay for these insights? Would your answer change if you could only see deforestation and drug smuggling but could not stop it?

If Planet can continue to solve customer problems before they happen, they will be able to charge their average customer far more than $250k/year, and many new markets will appear.

Once a tool becomes an entrenched part of an industry’s workflow (Excel, Photoshop, and Salesforce are good examples), the challenge of disrupting it becomes monumental. Switching costs for advanced software tools are very high, and Planet is acting on their opportunity to establish their software as standards for the emerging geospatial analysis field.

The acquisition of Sentinel Hub establishes Planet as an open platform where independent analysts can explore earth observation data.

Planet has already embedded their Planetary Variables into Sentinel Hub, which both improves the platform’s features and entrenches their measurements and variables as standards for geospatial analysts.

In 2019, Sentinel Hub had over 34,000 registered accounts and 600 paid subscribers, and each new user that learns to trust and rely on Planet’s data makes Planet’s moat stronger.

For context, Planet co-founder Robbie Schingler’s LinkedIn bio says Planet “serves more than 30,000 users and 700 customers in more than 40 countries”. Planet now has 976 customers, not including the paid subscribers acquired via Sentinel Hub.

Combining Planet’s customers with Sentinel Hub’s userbase, it would not surprise me if 100,000 people are using Planet’s data in their geospatial analysis workflows today.

In 2017 Planet became the first company to image the entire world every day, and nearly a decade later they are still the only company performing daily image scans of earth’s landmass. This data advantage is hard to quantify, but it is more important today than ever before.

With thousands of images of every single point on earth over stretching back to 2017 (and a RapidEye archive going back to 2009), Planet can train AI models to detect and identify rare events on earth that other companies with smaller datasets may have never seen. They can also more accurately identify a larger set of trees, buildings, roads, ports, and any other object on earth.

What truly makes this data special is that no amount of money or effort a competitor can spend will let them go back in time to replicate Planet’s historical satellite data. Planet’s data is unique.

Reddit earns $60 million each year for providing Google with non-exclusive training rights using their data, what is the likelihood of similar deals for AI training rights in different domains?

If you assume Planet’s revenue will only come from customers that want to understand the world, it is easy to build a revenue projection based on their past financial results.

But what if Planet starts earning revenue directly from AI platforms too? $60 million is more than Planet’s entire quarterly revenue today.

One benefit of having two satellite networks in orbit (PlanetScope and SkySat) is that while PlanetScope is busy scanning the earth, SkySat can perform important imaging tasks for specific customers.

This means Planet can sell their PlanetScope data to an infinite number of customers with no marginal costs, just like a typical SaaS internet business. At the same time, their SkySat network can perform imaging tasks for high value customers that need hi-res photos, videos, and multiple re-visits of specific areas.

Other satellite imaging companies do not have the benefit of a background satellite network that can collect data with no marginal costs. Each of their satellites is “tasked” with a job for a particular customer, which means each additional customer comes with an added marginal cost.

Planet has the most accomplished team in the satellite imaging industry:

While experience at Twitter, Instagram, and YouTube may not strike you as relevant to the space industry, Planet’s business depends on effectively searching their vast libraries of images and bringing the most relevant information to customers.

In that sense, it’s exactly what YouTube and Instagram do too. They scan through their own vast libraries of photos and videos to show you the ones they think will be most relevant.

Planet has acquired 6 businesses to date, and has a strong balance sheet with $300 million of cash and no debt to support further acquisitions if necessary.

As a business built on proprietary data, Planet’s acquisitions not only expand their capabilities, they also eliminate competitive threats and consolidate unique data under a single company.

Future acquisitions could enhance Planet’s competitive position, and the combination of high interest rates and a new, unproven industry has put some satellite companies in uncertain financial conditions.

Planet is one of the few in a good position to buy struggling companies at reasonable prices.

One of Planet’s advantages lies in their “agile aerospace” approach to building and maintaining satellites. Instead of spending billions to build large satellites with long lifespans and launch cycles, Planet builds small, inexpensive satellites that last a few years, and constantly iterates to improve their designs as old satellites are replaced.

To illustrate the benefits of iterating quickly, Planet’s latest satellites are 3500x less expensive than NASA’s Landsat 8, which cost close to $1 billion in 2013. Planet’s satellites also have 10x higher spatial resolution.

The costs to produce an imaging satellite have plummeted in the last decade. The costs to launch a satellite into space have also fallen, and may fall further as fully re-usable rockets like SpaceX Starship are commercialized.

On the one hand, this means Planet’s ongoing costs for replenishing their satellite fleet gets lower each year. On the other hand, the cost barrier that would otherwise stop competitors from replicating Planet’s satellite fleet is also shrinking each year.

If Planet’s business proves to be profitable, lower satellite and launch costs will attract the attention of competitors. Those potential competitors may even own their launch infrastructure.

Of course, any future competitor will be behind by nearly a decade of imaging data by the time they start, and the imaging is not worth anything without software to analyze it and return useful insights.

That being said, there is a point of diminishing return beyond which more data leads to smaller marginal improvements. The question worth asking is whether a competitor could serve half of Planet’s customers with only a small library of satellite images and a new software platform.

Even with a small market share, a single competitor can have on an outsized effect on Planet’s pricing strategy.

As a thought experiment, how much lower would Apple set their App Store fees if Google’s Play Store was available on iOS devices?

In theory, Planet’s data can be useful to every company and government in the world as they can see changes happening everywhere.

However, just because Planet can see everything on earth, it does not mean Planet will be able to monetize everything on earth. The earth observation industry is still nascent, and other technologies may be able to generate better insights than satellites.

Planet has made a strong case that their unique method of collecting data via satellite is useful to some of their customers, as it would be nearly impossible with drones or other technology.

However, it is also important that we ask whether small satellites taking photos in low-earth orbit will continue to drive the most relevant insights about the world for decades to come. Planet’s business is not yet profitable, so their valuation depends on being able to serve customers for many years, even decades into the future.

Reddit has already set a precedent for tech companies with access to unique data selling licensing rights for AI platforms that want to train their models.

And while we know the revenue outcome of Reddit’s deal, we do not know whether exposing this information will eat into Reddit’s moat as a sustainable business.

Said differently, does selling access to training data hurt Reddit’s long-term viability as a business? If Planet chooses to go down a similar path, they will need to answer the same question.