Humans’ fascination with navigation and exploration seems somehow innate. I guess our mammalian brains are inherently looking for greener pastures, food or water sources. And those who control the maps control the world, right? An understanding of mapping and navigation has been the key to conquering new lands or opening up new trading routes. And now apparently, it’s the key to working out whether your roof gets enough sunlight for those new solar panels.
I am sure a lot of us remember the first time that we saw google maps and how exciting it was. We all gathered around our PCs or Macs and eagerly zoomed in looking for our letter boxes or if we were really lucky we got a view of our cars parked out on the street.
Technology has come a long way since those first days of google maps. The camera systems today are a different beast with orders of magnitude more detail than their prehistoric ancestors. Data capture is becoming increasingly fast and more cost effective. So much so that companies like Nearmap no longer have to offer a bespoke service but can fly multiple times a year capturing huge amounts for RGB data that can be captured ahead of demand. Currently this company flies roughly 90% of Australia and 80% of the US.
For those interested in tech changes occurring in the geospatial / GIS/ remote sensing space – I highly recommend Daniel ODonohue’ s MapScaping podcast (“Map” not man).
When investing in tech you never know when the next disruptor or advancement will come along. This podcast series has several great interviews with key industry insiders that have really helped me to broaden my understanding of this unique and fastmoving space.
Nearmap has not had a good run of late. As mentioned by others the market was not impressed by its updated guidance or level of churn and SP took a huge hit this week. Nearmap can be a tough stock to hold. It’s previous history of creative accounting, its recent short report attacks and legal entanglement with Eagleview have been hard to weather. As Strawman has mentioned it is tough to see the value upside with all these risks. I can see why many are selling.
I have decided to continue to hold for now. I am betting on Nearmap AI to provide a very helpful solutions based proposition for Nearmap’s customers. I believe this is where the value lies solving problems not just in data capture.
So this week I wanted to look further at the industry.
My goal this week has been to further understand:
1). Whether there is a potential risk of disruption to Nearmap’s data capture space. How likely is it that drones or more specifically satellites will usurp aerial mapping in the near future.
2). Do Nearmap’s AI algorithms themselves widen the MOAT of this geospatial player?
Advantages of plane capture vs satellite
I have frequently heard throw away lines about how pinpoint accurate satellite imagery is and how this will likely be the way we map the world in future. I never really thought much more about this and took these comments at face value. It is only after a little digging that surprisingly I don’t think this statement is accurate or realistic. Well at least not for the foreseeable future.
Joe Morrison from Umbra (a satellite start up) risked his professional reputation recently and wrote an article lamenting his frustration with the utter lack of standardisation in capture and data interpretation in this space. He received a lot of praise for his brave and honest article and it resonated deeply with his peers in this community.
After a little research I feel even better about plane mapping. Why?
1. Firstly hurling large, heavy, metallic objects into space is a hugely expensive endeavour.
2. Optics of satellites mean you can’t get RGB images when there is cloud cover (the majority of urban capture areas are equatorial and this area has particularly dense and regular cloud cover).
3. Defence Forces are the major customer of satellite data (I will explain this further below).
4. The current optical resolution of satellites is not as good as planes.
5. Satellites are too expensive and the cost of capture is not going to reduce any time in the near future.
The service that Nearmap provides requires details. Good resolution is essential. The smaller pixels also help to build more accurate AI models that can quickly provide data insights - how many pools are there in such and such area? useful insights for pool product suppliers or how many rooves in the area look like they need repair for an insurer post hail storm. The Hypercam 3 due to roll out in 2022 has 6 cm pixel resolution. You will see that rust on your roof with this level of resolution. This detail is essential for mapping urban areas.
Satellites on the other hand at their best provide about 30cm pixel data. There are a few new startups launching satellites that offer 70cm-1m+ pixel resolution. At this sort of level imagery is only really suitable for land or terrain mapping, it is not good enough quality for seeing the details of builds on a job site for instance.
Cost is the other huge difference. Satellite imagery is very expensive as mentioned and it is worth repeating. The barrier to entry for a satellite player to compete with Nearmap will be cost prohibitive for now……unless Elon Musk wants a new hobby on weekends.
Really there are only 2 players in the US that are capturing satellite data for sale Maxar and Airbus. So if you want to gain access to satellite data you will be waiting for months wading through legal contracts and paying 100s of millions of dollars to task a satellite to fly over the area you care about. So accessing satellite data is a super long, difficult and expensive process. Something that I did not realise.
Defence is the primary customer for satellite imagery. These customers have very deep pockets and can afford to keep Maxar and Airbus in business. There aren’t many commercial customers – Google and Apple obviously but with the lack of demand and competition in the industry there is no downward pricing pressure and I don’t see this area getting cheaper any time soon.
The satellite imagery space is in its infancy.
My very basic Neanderthal understanding is that the complexity of satellite capture is staggering:
1. You have the variability in cameras and capture type RGB vs non-visible light spectrum etc….
2. The math in the geospatial plotting is fraught (matching the movement of satellite through space while targeting a point on earth repeatedly and accurately is complex)
3. Analysis ready data (ARD) problems - so what is captured and sold varies and doesn’t have a standard. This means it is very hard if not impossible to switch data platforms or readily switch algorithms to provide answers to your imagery questions quickly.
I can see why this area is still so expensive. In reality due to the lack of standardisation, the cost and the variability in quality imagery it does not sound like Satellites will be a direct threat to more traditional ways of aerial mapping with planes anytime in the near future.
Drones are a companion operator in my mind. Drones are brilliant for low lying monitoring. LIDAR capture is also brilliant with drones. LIDAR can penetrate gaps in trees etc.. to produce incredibly detailed point clouds however they will not likely replace RGB capture in planes any time soon. The camera systems used by capture companies are incredibly heavy. Hence planes are required to carry the weight and for fuel efficiency to increase flight times and hence reduce data capture cost.
AI algorithm and MOAT
Dr. Michael Bewley discussed insights about Nearmap technology and Nearmap AI, in Mapscapes most recent podcast.
What is clear to me is that Nearmap has a huge MOAT of capture data as well and the detailed resolution to train incredibly specific AI algorithms. The semantic pixel training on the data sets allows Nearmap to classify and provide answers for customers incredibly quickly and accurately.
The power of the AI is huge as result of the huge data training sets that Nearmap has access to. Furthermore RGB data seems to provide incredibly effective detailed information to customers. I don’t see LIDAR replacing Nearmaps capture imagery any time soon.
Nearmaps edge over its competitors is coming from the following:
1. The camera tech – it allows for very fast high flying capture
2. The AI self-labelling system. The more data Nearmap captures feeds into these models making them more and more accurate, better and faster. -There will be endless use cases for these data sets and tech insights.
3. Change detection – because Nearmap has so much old data from previous flights this can be overlayed with more recent captures to see small changes which is very handy for customers like insurers.
4. Because Nearmap is a first mover and growing so quickly in the US it’s AI technology is shaping the development of a universal mapping language for AI models.
The accuracy of the AI algorithms is everything and will be a major MOAT for companies like Nearmap. Furthermore Nearmap’s management understands that its data needs to solve problems. Capture will only be a part of the business. Predictive ability from AI data sets will be the next big advancement.
Nearmap can provide answers for companies say insurers – take the latest aerial images – overlay flooding risk history and land elevation and voila – prediction models for which houses are likely to flood in the next big rain event. Nearmap is entering into the space of processing the data and providing answers for companies.
For now I am happy to hold. I see a bright future for Nearmap and I don’t see a tech disruptor managing to provide affordable competition to plane capture in the near future. Hopefully there will be greater growth than predicted in helping businesses find answers to problems – not just providing data sets. I am happy to see if this stock can turn around in the next 12 months.
Cheers
Nnyck777
