Wonder what the pricing in on that location data is for financial sector customers, and if it's cheaper than a regular purchase of satellite data. It'd be interesting to compare the mobile location data to the satellite data and find the deltas— that would make it a lot easier to build efficient models w/ existing satellite if it's cheaper.
This example of counting cars has been around the geo industry forever, and satellite imagery is pretty cost prohibitive to get on a regular basis. I'd be surprised if hedge funds haven't already moved on to purchasing location-based data (ie mobile phone data) to get even more accurate counts of people in box stores.
While margins in on-demand are usually pretty slim, companies can get pretty creative about opening up revenue streams: ad-serving, corp partnerships, enterprise pricing, licensing out tech).
None of these are a silver bullet, but they open up new growth that isn't tied to the standard hockey stick growth curve people use to get funding.
Satellite imaging is the same way, too. It's pretty hard to be an independent commercial satellite imaging company—most have been scooped up by MDA (Digital Globe) or PLanet Labs (Skybox, Rapideye)
I’ve seen them making unprotected left turns down in the financial district in the evenings—it’s a lot more complex in mid day with full traffic. Looks like the cars on the video are making riskier turns than I would make!