This past weekend, just in time for Cyber Monday, Amazon CEO Jeff Bezos revealed plans on 60 Minutes for delivery drones for Amazon Prime shipments weighing up to five pounds. While the target of getting drones involved in the next five years sounds ambitious, there are many questions and technological obstacles that need to be overcome first. Taken at a high level, Amazon states that the drones could handle up to 86% of all Amazon shipments, getting product to your door as quickly as 30 minutes after you place your order. If that sounds too good to be true, it probably is for most of us. People that live in densely populated areas would be the first target I would guess, with the service potentially spreading to other areas as it becomes feasible. Amazon would need to have warehouse locations within 10 miles of your office/residence to be within range, and they would need dozens if not hundreds of the drones at each location to handle the packages.

The timing of the broadcast is particularly telling, as Bezos noted that they have around 300 items ordered every second on Cyber Monday. That creates a lot of work for the shipment side of the business, but even if Amazon only tried to ship 10% of the packages by drone on such a busy day we'd be looking at 30 packages per second, an average delivery time of perhaps 20 minutes (1200 seconds), probably another 10 minutes for “refueling” (600 more seconds), and thus a drone fleet numbering 54,000 would be needed. If all orders were to be delivered by drones, we'd be looking at ten times that number – over half a million drones.

Even assuming the physical presence could happen (at least for some areas), there remain many other obstacles: weather, operating cost, reliability, potential for vandalism/theft, FAA regulations, etc. The drones are technically octocopters, and they're already being used for taking pictures and filming. Pricing for an octocopter large enough to carry a five pound package is going to be pretty obscene as well – around $10,000 seems like a reasonable baseline, though with mass production it might be lower. Of course there's still the need for the facilities and personnel to run the operation, so $20,000-$30,000 per drone might be a more reasonable estimate.

I know Amazon ships a lot of packages, but the changes in infrastructure alone make this something that will likely take much longer than five years before we see it widely used. I suspect more likely is that the first use of the service by Amazon will be as an optional shipping method that will cost a premium. Amazon Prime members currently get free 2-day shipping on qualifying orders, with discounted 1-day shipping as well. How much would people pay for 30-minute shipping if it were available? In some cases, it might be $100 or more. If Amazon were to charge $100 for drone shipping, and a drone could make on average 15 deliveries per day (seven days per week), each drone could potentially pay for itself within a month...or at $50 per delivery, two months. If on the other hand this is a “free for Amazon Prime” service, we'd likely be looking at a year or two just to cover the cost of the drone (and assuming no equipment failures).

Regardless of when or how drone shipments take place, there's no arguing with the fact that it's a really cool idea. It's the sort of thing we see and read about in sci-fi, and as is often the case it's more a question of “when” rather than “if”. Having just traveled over 2000 miles via car for Thanksgiving to be with family, it's in the same category as fully automated vehicles. I personally hope to live to see the day where I can hop in a car, tell it to “take me to my mom's house”, and then sit back and relax (or work) as the vehicle zips along at 100MPH, coordinating travel with satellite monitoring and nearby vehicles so as to avoid slow-downs, accidents, and other potential problems. I think it's inevitable that the day will come when computer-controlled vehicles take over for humans, and Amazon's drones are yet another herald of such advancements. I for one welcome our new electronic overlords. :-)

Comments Locked

75 Comments

View All Comments

  • 3DoubleD - Wednesday, December 4, 2013 - link

    Oops, by "Driver is not liable" I meant "taxi passenger is not liable"
  • Hubb1e - Tuesday, December 3, 2013 - link

    That is my worry as well, but the supporters of this technology point to lawmakers who are opening the door to this technology and will likely sheild the developers from some liability except for gross negligence. If the accidents are reduced 10X then as a whole that is something the government can support and will create laws around.
  • dishayu - Tuesday, December 3, 2013 - link

    I'm not even talking about the legal angle yet. The bad publicity that happens due to a tiny TINY number of accidents will stop it from gaining traction in the market... or at least that's what I fear. I mean, put yourself in the place of an accident victim. Would you later on say that it wasn't the computer's fault? Most likely not.
  • Zoomer - Tuesday, December 3, 2013 - link

    It is far more likely to be the human's fault.
  • jdvorak001 - Tuesday, December 3, 2013 - link

    Whatever the incident rate with human drivers, there is always someone to be held responsible: a person who was at the wheel at the time of the incident. I wonder if people would go for even a sizeable reduction in the incident rate, while losing this someone-to-blame feature?
  • nathanddrews - Tuesday, December 3, 2013 - link

    As the saying goes "Build a better road (or car) and they'll build a better idiot."

    Data shows very clearly that human error (behavior) - exclusively - is responsible for the vast majority of crashes. Engineering decisions (roadway design, vehicle defects) are sometimes factors, but not often the cause of crashes. Roadway and vehicle safety countermeasures are reaching their limits in terms of cost effectiveness. From a data-driven perspective, the humans must be removed from the driver's seat eventually.
  • Zoomer - Tuesday, December 3, 2013 - link

    Workaround: They could always retake control by grabbing the wheel, tapping the brakes, etc. They were not paying attention? Too bad.
  • dishayu - Wednesday, December 4, 2013 - link

    That seems like a reasonable workaround.

    Counter argument : A manual vehicle (let's say a motorcycle) and a computer driven vehicle come together and the motorcyclist dies in the incident. Who takes the blame?

    There are a lot of things to be sorted out before we can start using auto-drive cars... Technology-wise, we're very nearly good to go.
  • CharonPDX - Tuesday, December 3, 2013 - link

    I think the big difference is that when Dominoes did it, they fully acknowledged it was purely a stunt, and that they couldn't get actual regulatory authority to make it a real service. (And it did receive an appropriately-high level of attention considering it was fully acknowledged as just a publicity stunt.)

    Amazon has stated it firmly as "we're going to do this - we're going to work with the government to make this a reality." That's the difference.
  • nathanddrews - Tuesday, December 3, 2013 - link

    Thoughts that immediately came to my mind - acknowledging that it's only a demo:

    1. How does Amazon intended to get its little plastic snap case back?
    2. I don't want 50 drones buzzing over my house every day. I don't want 10.
    3. Shoot down a drone - free prize inside!
    4. I'm really happy with 2 day delivery and the fact that they are already talking about Sunday deliveries - no more weekend delay.

Log in

Don't have an account? Sign up now