Yes, someday TerrAvion might go unmanned, but as a former military drone unit commander, I both look forward to the day and think it is further off than people in the Valley think. We've had the technology for unmanned trains for 60 years and yet when all you care about is price per ton/passenger mile, sometimes an operator is the last part of the labor fraction to take out. Mapping is a similar price per pixel type operation.

Smallsats are cool, but they are really only attacking price and reliability of satellite--they aren't even close to say Pleaides or WorldView3 constellations on performance, let alone matching RapidEye for agriculture.

Regarding the disposition of the civilian mapping data market, I think if you look at what the military did in Iraq and Afghanistan the unlimited defense budgets of the 2000s might give an idea of what civilians of the 2020s will be doing. The satellites mostly got used by division and theater level staff to answer long term questions, say 20% of questions. Hand launched drones, which were ubiquitous in company storage rooms, but not in operations answered a few micro questions say 5% of aerial data needs. The remainder of planning questions got answered by air-breathing aircraft of one type or another. Not sure why the venture community has not largely not noticed this.