Amazon Buys Thousands of Its Own Truck Trailers

there are frequently backups by the warehouse in Bellevue, Washington, out onto the street. They park on the side of the road, but they are there.

Here's one mention:

https://www.reddit.com/r/Trucking/comments/3uzlzs/16_hours_a...?

At the scale amazon operates at, you probably do know in advance what the day's deliveries will be. I'm sure they could fill a truck with product, cruise a set route, and be reasonably certain it would all be delivered by the end of the day.

I think on amazons scale, the number of people buying a good within a large area would have some near term predictability I would imagine. In fact I would be surprised if Amazon does not speculatively dispatch goods ahead of time to local warehouses based on projected demand for orders that have not been placed yet.

A lot of people don't actually want or need 30minute delivery, and are perfectly happy with Amazon's free delivery that takes a few days. I can get two hour delivery here in London, but I've used it once.

Amazon could get a product to a local centre, then instead of sending a van out driving all day going to one customer after the other, they'll hit a few key locations near multiple customers and launch drones out to serve them at the same time.

Yeah, I wasn't thinking of the 30 minute delivery in this scenario, but rather an efficient next-day or same-day parcel delivery system. It'd presumably be much more efficient than door-to-door service. If we can perfect driverless vehicles also (driverless delivery trucks, anybody?), this would probably be the ideal delivery scheme. Heck, with that level of automation, if you time things right your trucks may never even need to come to a complete stop.

That said, you could probably extend the range of the 30-minute service with a strategy similar to this. Just bundle orders for 10 minutes, send vehicles 10 minutes out and then the drones fly 5-10 minutes to deliver the package.

Additionally, you could probably keep delivery trucks stocked with some of the top items as well and surprise/delight the customer with "Hey, we just so happen to have 30 minute delivery available now if you want it." I'm no SCM expert, though, so that could very well be a waste of more-valuable space.

So this is being implemented already from a local fulfillment center level, sans drones.

The actual mechanisms are actually super straightforward -- basically if a widget X sells Y amounts per day with Z variance in an area, get the low, low end of that to that warehouse/area so you only have to do last-mile delivery.

Then do it with operations research (well, linear algebra) on scale. :)

This points to what I don't understand about the Amazon financial story.

Walmart built up their massive infrastructure business while being simultaneously immensely profitable. However, at Amazon, building out infrastructure comes at the expense of profits.

What am I missing that a lot of smart people are seeing in Amazon?

Reminds me of ship shipping ships

http://twistedsifter.com/2012/04/blue-marlin-giant-ship-that...

I would change how the trailer docks.

Right now the weight is still on the wheels and each additional load strains the suspension. Most trailers have pneumatic load balancing, so the bed drops with the load and then slowly pumps back up. Because of this a free-floating hydraulic ramp is needed. These interfaces are finicky and an unnecessary bottleneck. I've managed to get a 5 ton lift stuck on the ramp before.

I would put anchor points akin to trailer hitches on the back corners of the bed. The driver backs up to the bay, aligns the posts over the hitch, and hydraulics lift the hitches until they mate with the posts. Next they would lift up the truck bed, pull the truck back towards the bay, and set it down in a grove so the bed and warehouse floor are perfectly aligned. It would be similar to how a u-haul ramp mates with the truck bed (https://youtu.be/Ch7ukPPtZOI?t=46s), except you would be mating the truck bed with a building.

You could also have a giant drawbridge style ramp that lowers down and anchors to the back of the truck bed. It could double as the bay door. You could also add in outriggers that hold the bed at one height during loading. Alignment would be tough. Possibly the outriggers could sink into divots in the ground and shift the bed into position.

Both of these would also add a full width mating, which would allow for multiple bots to load at the same time.

Heck, I like the idea of having 4 hydraulic lifts, one for each corner of the bed, and have them lift the trailer up until it is flush with the warehouse. This would allow for the added perk of making taller, 2 story trailer beds and loading in two sets of shelves.

There are infinite possibilities, but I would be many of them are cheap enough to be worth the investment. One this is for sure, operating a lift is boring and I have way to much time to think on the job.

Well they also have planes and delivery people... It's not that far off. They also have distro centers all over the place so orders frequently come from within a few hours away (or less!) and not a few states away.

http://motherboard.vice.com/read/a-secretive-air-cargo-opera...

That episode is a work of genius. With poop jokes.