Got the same email and that was quite impressive to see. The...

Got the same email and that was quite impressive to see.

The revenue is quite hard to estimate on a per GB basis due to how much the plans vary depending on who you are. Some customers like NYC DoT pay $1 million/year for Nearmap and get 64GB per month, some pay much less for 2 GB. Not to mention that plenty of subscribers don't use their full plan each month so there is clearly a difference between billings and total possible data that can be consumed.

On another note it might interest some people here to understand the costs to load/store that that. One of the best things that Simon Crowther did was move Nearmap to AWS. AWS is a no brainer for cost, realiability and scalability. Nearmap has over 2.5 petabytes of imagery as of July 2018. Since not all of these images would be accessible through the web browser let's assume 20% of this is stored in AWS S3 and 80% of it (raw footage) is stored on AWS Glacier.

Back of the envelope costs:

Data Storage:
Glacier = $8000/month for 2PB (Not included get requests as it's not clear they'd actually need to retrieve this often)
S3 (Simple storage) = $12400/month for 500 TB
S3 Get Requests (think each tile load of the map) = 999,999,999 (Max I could do) = $400 so let's double that. $800 / month. (This might be cheaper as they probably use Cloudfront but not quite sure)
S3 put requests (think each time you put data into S3 via data captures and post processing) = $3000/month

EC2 (Standard Compute Power):
20 x Linux m4.10xlarge (these are pretty powerful beasts so I don't know what they need but let's say this) $29,000/month
25TB data out - $2300/month

MapReduce (Big Data Computing):
Say one MapR M7 - $2560/month

Other Services:
Not going to bother breaking down things like SNS, Sageworker, Route53. Let's say $2000/month.

That comes to a nice shopping list of $58,000/month excluding Google Map API costs which would increase that figure a bit. All in all, not bad for serving 1.7 billion map tiles, processing of captures, DNS requests, S3/Glacier storage. Lots of assumptions made and given how they made a case study for Amazon and how large they are they'd likely get a 20% discount as well.

Here is the video of Simon's case study: youtube.com/watch?v=oYGPi_CvJY4