BlackSky ($SFTW, $BSKY) DD
Disruptive, vertically integrated geospatial data and analytics
SPAC merger news:
Osprey Technology Acquisition Corp. (SFTW) is reverse-merging with BlackSky, a “real-time geospatial intelligence, imagery, and data analytics company”. Quick numbers from the investor presentation:
Pro forma EV = $1.1 billion
SFTW trust = $318 million
PIPE = $180 million. Investors include Mithril Capital (Peter Thiel) Hedosophia, and Tiger Global.
Ownership = 63% existing shareholders, 21% SPAC shares, 4% founder shares, and 12% PIPE.
Thesis statement:
BlackSky is a disruptive, vertically integrated business model that removes the barriers-to-entry historically present in the utilization of satellite imagery, which in the past has severely limited utility and application in the private sector. BlackSky reduces the cost of image acquisition by 90% compared to existing competition and their Spectra SaaS works to democratize access to petabyte-scale imagery and AI/ML analytics. Longer-term, satellite imagery will be in use across all major industries including but not limited to agriculture, security, mining, and insurance, offering an absolutely massive TAM.
Disclosures:
My background: Research scientist using remote sensing (satellite observations) and machine learning. I love remote sensing and see it as the future of observational datasets for, but not limited to, natural resources, environmental change, natural processes, anthropogenic effects, and global security. I’ve been waiting for an RS company to jump into the SPAC market. I don’t have experience with many proprietary datasets, especially those discussed here, so most of my research is based on publicly available information.
My positions: warrants and shares.
Financial advice: I’m not an advisor and this is not financial advice. Do your own DD.
Ok, lets get into it…
What is satellite imagery?
When you think of satellite imagery, you probably picture something like Google Earth, which is just RGB images over the Earth’s surface. The reality is that imagery comes from both passive and active sensors. Passive sensors are much more common and are what companies like BlackSky, Maxar, and Planet typically use, although some incorporate a limited set of active sensors as well. Passive sensors don’t just measure RGB, though. Sensors from the Landsat and Sentinel missions, for examples, measure radiation along the electromagnetic spectrum from 400nm up to as high as 13,000nm. For reference, visible light is 400nm to 700nm.
Who uses this information? Well, not everyone yet, but it’s quickly gaining popularity across a variety of industries. This NYT article visualizes changes to surficial water supplies over a 30 year period in Bangladesh, the U.S., Tibet, and Brazil using 30m Landsat imagery. RS data is used to monitor and optimize agricultural operations, identify illegal fishing, quantify wildfire effects, map harmful algal blooms globally, etc. I purposefully chose large-extent examples there, because they tend to be more relatable to people unfamiliar with satellite datasets. But those are also kind of the good-will, “science will save humanity” examples.
So, what is BlackSky?
BlackSky is a geospatial intelligence company. What does that mean? As usual, the best definition comes from Wikipedia:
“Geospatial intelligence is intelligence about the human activity on earth derived from the exploitation and analysis of imagery and geospatial information that describes, assesses, and visually depicts physical features and geographically references activities on Earth.”
BlackSky was born as one of three primary business services within parent company Spaceflight Industries, with the other two services being (1) excess capacity reselling “rideshare” business, sold off in Feb 2020, and (2) LeoStella, a joint venture with Thales Alenia Space, that manufactures smallsats. For the SPAC merger, the name Spaceflight Industries is being dropped in favor of BlackSky, which will own a 50% stake in LeoStella.
BlackSky has a vertically integrated business model that starts with the manufacturing of the actual satellites themselves and finishes with an AI/ML-driven data analytics product. Lets work our way from start to finish.
Business Model Part 1 - Satellite development and launch
First, BlackSky builds smallsats, via their 50% stake in LeoStella, which they then launch into low earth orbit (LEO, most common orbital path for observational systems). The actual launches are performed through outside companies. For example, BlackSky launched two satellites on the SpaceX Falcon 9 rocket in August 2020. Currently, 5 satellites are in orbit and another 16 are in production with plans to be in orbit by the end of 2021, as well as another 4 spare satellites ready to deploy when necessary. The primary competition, Maxar, outsources satellite manufacturing to Raytheon while Airbus builds their satellites in-house.
What about costs? BlackSky satellites are MUCH less expensive to develop and launch than the largest competition. For example, Maxar’s previous program, WorldView-4 (single satellite), cost $850 million and was lost in 2019 due to technical failure. Their newest program is WorldView Legion, with an estimated cost of about $600 million to build and launch the 6 satellites ($100 million each) making up their constellation. The other competition, Airbus, is launching the first two satellites in their 4-satellite Pleiades Neo constellation in 2021. Cost to build and launch the four satellites is approximately $232 million, averaging $58 million per satellites, with cost split between Airbus and the French Space Agency CNES. In comparison, BlackSky was able to build and launch their Pathfinder-1 satellite at a total cost of just $10 million! What this means is that the monthly cost of imagery for a 25 sqkm site is just $12,000 through BlackSky (p. 10, investor presentation), whereaslegacy competition costs up to $120,000. This is extremely important for tech updates and innovation, as well as competitive pricing. BlackSky can deploy new satellites with improved sensor capabilities extremely quickly and for relatively low cost.
Business Model Part 2 - Satellite imagery
This is where the debate over the imagery itself comes in. The main competition, at least in terms of generating the satellite imagery itself, are Maxar (WorldView Legion) and Airbus (Pleiades Neo), both of which are moving towards satellites with 30-cm and 50-cm resolution, respectively. Meanwhile, BlackSky’s imagery from the previous generation (Gen-2) satellites are just 100-cm. Is this a problem? No. Because smallsats are so quickly produced, and because BlackSky owns the underlying manufacturer, successive generations are becoming more and more advanced. For example, the current Gen-3 satellites host 50-cm resolution as well as short-wave infrared readings, providing night-vision imagery and smoke/haze cut-through. So in the short-term, BlackSky has lower resolution imagery than Maxar, and matching Airbus, but can improve and adapt their sensors much more quickly, and for far FAR lower cost than the competition.
The best competition against BlackSky are other private companies working to scale up large-count nanosat, microsat, and smallsat constellations, such as Planet (closest competitor, imo), Satellogic, and ICEYE. Honestly, if any of these were to go public via SPAC I’d be as excited to jump in on them as I am on BlackSky. Lets go over some numbers for each.
Planet (great company, as per anecdotal evidence from colleagues) utilizes a constellation of 130+ shoebox satellites that are even cheaper to construct and deploy than the smallsats operated by BlackSky. Planet has a couple advantages over BlackSky, namely their existing satellite network, robust data lake, and low cost of deployment. But spatial resolution of their imagery is 6-10x coarser, ranging from 300-500 cm (3-5m), which results in a pretty significant divide in use cases between the two companies. The cheaper, lower resolution data from Planet may be more useful for general purpose applications, such as approximating agricultural production and monitoring large-vehicle traffic (shipping, aircraft). But for larger organizations more fiscally invested in the outcome of satellite imaging monitoring, such as agricultural companies in California, Wall Street hedge funds incorporating imagery into models, or insurance companies measuring impacts of natural disasters, it’s much more likely they would select the much finer-resolution data products. I think that if Planet were to public it would be a lower-risk but lower-reward choice compared to BlackSky.
Satellogic (based out of Buenos Aires) hosts a 10 nanosat constellation producing 70-100cm resolution spectral and 30 meter hyperspectral imagery. Series B round in 2017 was led by Tencent, so great vote of confidence there and a 2019 round of funding, also partly funded by Tencent, raised another $50 million. Satellogic is unique from the other companies in that they are targeting a government customer base, with the goal of selling exclusive access and control over their fleet to municipal, state, and natl governments (caveat: from Wikipedia with outdated reference). Expected customer charges are $2 million/satellite per year, for a revenue stream of $20 million/year for the existing fleet.
ICEYE (Finnish) has 10 existing microsats in orbit, each at a cost of approximately $3 million (estimated in 2018), with plans to launch another 8 by the end of 2022. ICEYE differentiates itself from other smallsat companies by using Synthetic Aperture Radar (SAR), an active sensor that uses longer wavelengths to see through clouds, haze, and foliage at fine resolution. Downsides to SAR are that it results in slower acquisition speed due to higher modeling costs (ICEYE has a turnaround time of 12 hours, compared to under 2 hours for BlackSky), greater energy consumption and thus decreased sensor lifespan, and lower image coherence for non-stationary objects. ICEYE raised $34 million in 2018, and existing customers are primarily research institutions.
Some other example competitors are Spire and HawkEye360, which we don’t have time to get into here. But an important point about the companies listed above is the amount of funding they have raised in the past, which is generally 1/10th the amount provided ($320 million) to BlackSky by the merger with SFTW. I believe this will provide a massive boost forward for BlackSky against the competition, providing funding for both further expansion of their constellation as well as greater build-out of their unique Spectra platform (see below). Their strongest competitor, Planet, raised $168 million last year, and so will likely be aggressively expanding as well. However, given the absolutely massive TAM, there is plenty of room for several satellite constellation companies in the market.
Business Model Part 3 - Cloud-based analytics via ML/AI
This is where I think my expertise and personal experience with satellite imagery comes in to play. Let me explain to you how, historically, working with satellite imagery went: you access a server, you pull individual images (usually in GeoTIFF format) to your local machine, and you process them. If you’re working on the scale of 1 sqkm, it’s not that big of a deal. But if you’re working over larger areas, such as cities, states, mountain ranges, seas, continents, etc., all of a sudden you’re dealing with terabytes, or more often, petabytes of data. Clearly, the old method of accessing individual images doesn’t work anymore. The data is too large to pull in the first place, plus the processing times would be obscene (think days, months, years) because (a) your local system doesn’t have enough cores, (b) common process/analysis methods aren’t optimized for massively parallel computing via HPC or cloud systems.
What changed in the last few years was the advent of cloud computing for geospatial data processing. In terms of research, Google Earth Engine (GEE) completely changed the game. They put together a massive data catalogue of almost all publicly accessible satellite imagery going back to the early 1980’s, hosting petabytes of data. GEE combined this with their own JS libraries, code editor, and APIs to allow users to access, manipulate, and analyze petabytes of data in seconds using Google’s cloud platform. It’s hard to overstate the importance of this, and it’s one of the reasons I’m so bullish on BlackSky.
See, BlackSky is not just a satellite imagery company. The reality is that they are also a data analytics company. Before now, if a company wanted to utilize satellite imagery, they would need a full team of scientists with geospatial, cloud, machine learning, and AI experts. What BlackSky, Maxar, and Airbus do is provide access to their datasets hosted on cloud systems. But what differentiates BlackSky from the competition is their SaaS product Spectra (2017 press release, 2020 press release). Spectra allows clients to access imagery on-demand with built-in AI/ML tools, streamline a mountain of external monitoring datasets, ability to access or purchase additional public or proprietary datasets, manually task the satellites yourself, etc. Case-in-point, job listings on the BlackSky website are concentrated on cloud, software, and ML/AI positions, plus some business director positions. Engineers are working over at LeoStella, leaving BlackSky with the ability to focus on growing sales and building out their SaaS platform. Importantly, the CEO Brian O’Toole is the former founder/CEO of OpenWhere, purchased by BlackSky (then Spaceflight Industries) in 2016, which developed a “web-scale software platform for customers to rapidly request, receive and interact with satellite imagery via the internet or even a mobile phone” [via PitchBook]. So BlackSky, then led by founder (and serial entrepreneur) Jason Andrews, purchased OpenWhere and handed over the position of CEO to the now-former CEO of OpenWhere. Pretty strong statement that they are focusing not just on imagery itself but also on the accessibility of that data.
What does this mean? Say a mining company wants to incorporate satellite imagery into the security monitoring system of their open pit mines in Australia. Previously, this wouldn’t have been feasible - it would have required purchasing imagery from Maxar or Airbus, then having experts on staff to work with the imagery, integrate with external data sources, and build out automated alert/monitoring procedures. With Spectra, the mining company can simply sign up for a subscription plan and security personnel can utilize the data without any expertise or skills relating to data science or geo-analytics.
Given this, I believe BlackSky begs comparison with name-brand growth companies like Snowflake and Palantir. Through Snowflake, a company can utilize cloud storage/access services without a full team of cloud engineers/architects. Through Palantir, a company (or gov. org.) can utilize AI/ML methods without a full team of expert statisticians. Through BlackSky, a company can utilize massive proprietary and public satellite imagery through built-in AI/ML methods without a team composed of geospatial, cloud, software eng., and stat experts.
This is why slides 29-30 in the investor presentation compare BlackSky with both space tech companies (NPA/AST, SRAC/Momentus, Maxar, Airbus, etc.) as well data analytics companies (elastic, Datadog, Palantir, etc.). In terms of existing and projected revenue, BlackSky beats out the other incoming SPAC companies, though their 2020A revenue of $22 million is effectively insignificant relative to their $1.1 billion valuation, so I wouldn’t put too much weight on that comparison. Additionally, BlackSky is targeting completely different markets than Momentus and AST.
Company Management
Brian O’Toole, CEO - GeoEye, GE Aerospace, OpenWhere
Peter Wagner, CTO - Spaceflight Industries, Dir. of DoD Op. Resp. Space Office, Stanford & UWY educ.
Nick Merski, COO - U.S Air Force
Patrick O’Neil, Chief Data Sci - Board member for Center for Mathematics and AI; PhD mathematics GMU
Nick Tabbal, VP - JOT Analytics, comScore
Scott Herman, Chief Solns Arch - GeoEye, VisualCV
Economics
2020A Revenue= $22 million
2025E Revenue = $546 million (tenuous, as per the SPAC way)
Margin on imagery = 90% (accounts for 64% of revenue)
Margin on Saas product = 60% (accounts for 30% of revenue)
Contract pipeline = $1.7 billion, though contracts can be fungible so this number is far from guaranteed
Existing partnerships (mostly government): National Reconnaissance Office, National Geospatial Intelligence Agency, IARPA, U.S. Air Force, U.S. Army (taken from inv. pres.)
BlackSky EV/CY’22E = 9.7x. Compared against SRAC/Momentus at 27x, Snowflake at 54x, Palantir at 44x, MongoDB at 27x, Elastic at 17x.
Misc. thoughts
Obviously, BlackSky is a target for the ARKX fund whenever it launches. In my opinion, that’s almost guaranteed. However, I also think that ARKW (holding Palantir) should be considered a target fund as well, given the data focus of BlackSky. This isn’t to say that Ark purchases are a sole reason to buy, but if/when those purchases occur that would be a significant catalyst for share price to move above the current $14-15 range. Additionally, Ark purchases give buyers more confidence in companies like this (those with high valuations but little to no existing revenue) and so would provide a general floor price to the shares. That, combined with existing investments from Peter Thiel (Mithril Capital), Hedosophia, and Tiger Global, make for a pretty solid foundation of successful investors. So if you’re a swing trader and don’t like to hold past merger, buying prior to any Ark investments could be profitable. Personally, I see this as one of the only strong candidates among the low-/no-revenue SPACs to hold long-term. That being said, it is still a more speculative investment than SKLZ (my baby) or other SPACs with more established revenue. If run-up is significant going into merger, I’ll probably convert a portion of warrants to shares to reduce volatility a bit.
Conclusion
BlackSky offers disruptive tech, in terms of (a) hardware - satellites, (b) data - imagery, and (c) analytics - Spectra SaaS. All of this is built into a vertically integrated business that yields costs which are just a fraction (~1/10th) of their competitors (Maxar and Airbus), but maintain a much more versatile, innovative, and adaptable constellation of satellites. Satellites are built in-house via 50% ownership of LeoStella, imagery is competitive in quality with that from Maxar and Airbus but with much greater ability to improve with progressing technology due to significantly lower cost & time for construction of new satellites, and the Spectra SaaS allows companies to utilize advanced processing and AI/ML geospatial methods on-demand without the need for additional employees. Existing nanosat and microsat private competition are out there, but funds raised from the merger with SFTW will provide BlackSky with an almost unparalleled opportunity to achieve first-mover advantage.