Watch Now


What external freight makes sense for Amazon’s network?

Image: Jim Allen/FreightWaves

This is an excerpt from Thursday’s (9/9) Point of Sale retail supply chain newsletter sponsored by ArcBest.

During peak holiday season 2013, Amazon ran into major roadblocks. The company had built a foolproof e-commerce platform where you could buy pretty much anything, and it had a warehouse network to fulfill millions of daily orders. But its transportation providers at the time, namely UPS, FedEx and the Postal Service, were unable to handle Amazon’s peak volume surge. 

The following year, Amazon began building its own global transportation network from the ground up. In the seven years since, Amazon Logistics (AMZL) has delivered more than 10 billion packages and has grown to over 400,000 drivers worldwide, 40,000 tractors, 70,000 trailers, 30,000 delivery vans and a fleet of more than 70 planes. One of its biggest infrastructure outlays, the $1.5 billion Amazon Air hub in northern Kentucky, became fully operational in August. According to SJ Consulting Group, Amazon is now shipping 72% of its own packages, up from 46.6% in 2019. Similar growth has occurred with its warehouse fleet, which has added more than 100 million square feet of space in less than three years. Currently, 77% of the U.S. population lives within an hour of an Amazon warehouse, up from just 51% in 2018, according to UBS. 

By now everyone knows the Amazon Web Services (AWS) story: Amazon outgrows its providers, builds a best-in-class solution for itself and begins selling excess capacity as a service to others. For years, the logistics industry has watched with shock and awe as Amazon plows tens of billions each year into its logistics infrastructure just waiting for the day it dusts off the AWS playbook for AMZL. 


In a limited capacity, Amazon already offers a variety of shipping options for outside merchants. In the U.K., Amazon has a logistics-as-a-service program — a business model that researchers from DePaul University predict Amazon will launch in the U.S. in the next 18 months, while Morgan Stanley predicted it could happen this year. According to an investigation by The Information, Amazon has been quietly transporting cargo on its planes for the Postal Service since at least 2019. 

This week, CNBC reported that Amazon is now moving cargo for outside customers in the U.S. The details, as they often are with Amazon, are sparse. We know Amazon offers LTL space at discounted rates via Amazon Freight. And we know that Fulfillment By Amazon (FBA) services orders not made on Amazon.com — that’s why some orders from eBay, Etsy and even Walmart.com arrive at your doorstep from an Amazon-branded truck, often in Amazon packaging. In this scenario, all of a brand’s inventory is stored in and fulfilled by Amazon warehouses but sold via multiple channels including various marketplaces, the brand’s own website, social media and all the other places consumers shop these days. 

Where does it make sense to inject external freight into Amazon’s logistics network? 

The most glaring inefficiency in Amazon’s transportation network is its backhaul capacity. Through independent contractors, Amazon operated more than 40,000 trucks primarily in a one-way traffic network. When Amazon trucks complete a move from inbound center to another node in the network, it often returns empty. Amazon knows exactly where its backhaul capacity is, and on regularly trafficked lanes, I suspect the capacity is already being sold to other shippers. 


Leveraging this capacity is by far the least complex for Amazon, given it could be injected into the system and handled with only a truck. The “cost of disruption,” as Andrew Lockwood, senior manager of solutions design at Suddath Global Logistics, put it, must be carefully calculated when selling excess capacity. Should there be delays at the dock or traffic or poor weather, the cost of disrupting Amazon’s existing freight flows could be more costly than the incremental revenue opportunity. In other words, Amazon trucks are tightly integrated in a loop between facilities with telematics integrated into the warehouse management system and dock scheduling. When you take someone else’s freight, you lose control over the origins and destinations, the facility efficiency, the detention of the asset, etc

Extending its excess courier capacity to external parcels will be far more complex than leveraging its fleet of trucks to fill backhauls. If AMZL can find the right shippers around its network, it could fill its backhaul capacity without ever touching another node in the Amazon network. With parcels, however, packages must flow through at least one more node in the network besides the vehicle, either sortation center and/or delivery center. The cost of disruption, or at least the chance of it being more impactful, is higher when leveraging excess parcel capacity than truckload. 

But the courier opportunity is viable, and Amazon is surgical in its selection. “They’re not going to be just this blanket carrier that will deliver whatever package that you want them to, to whatever address,” said Dan Romanoff, who researches Amazon for Morningstar. “Amazon is sort of cherry-picking their routes. They want to run and sort the parcel sizes they want to deliver.”

Satish Jindel, president of SJ Consulting and ShipMatrix, estimates a 4.7 million package-per-day gap between parcel supply and demand for the upcoming holiday season. Beyond the peak, there are still millions of units sold on Amazon.com that AMZL never touches. Of AMZL’s volume, Jindel and his team estimate 55% is for FBA customers and 45% is products sold by Amazon. The third category of goods, those sold by merchants but not fulfilled by Amazon, makes up 15-20% of total units sold on Amazon.com, according to Jindel. 

Amazon has rich data on these sellers — where their product is located, detailed demand and supply data, and how quickly these packages are typically delivered. The end destination is crucial to adding density rather than adding stops. Should packages be headed to addresses already in Amazon’s routes, it makes a ton of sense to inject external parcels into Amazon’s delivery network. 

What types of customers are most likely to use Amazon shipping services? 

It’s without argument that Amazon has built an efficient, low-cost transportation network without unionized labor and a fleet operated primarily through contractors at hourly wages without benefits. CNBC spoke with Amazon seller Keith Gregory, owner of vitamin and supplement company Highland Laboratories, and he said his company saves ~$1,700 on routes from Oregon to Southern California when using Amazon Freight. 

Marc Wulfraat, president of supply chain consultancy MWPVL International, believes it is small to midsized shippers, like Highland Labs, that are likely best suited for Amazon’s offerings, whether it be LTL or parcel. His reasoning is founded on Amazon’s demand schedule — Amazon generates one-third of its annual business in Q4 and has built its network to ensure smooth sailing during peak season. But this capacity cannot be easily flexed up and down, so from Q1 to Q3, Amazon has much more excess capacity available to sell. In Q4, “Amazon needs every ounce of capacity it can get to meet customer expectations,” Wulfraat said. 


So what happens to the shippers that left FedEx or UPS for Amazon in Q4 when AMZL limits — or outright stops — shipments for your brand? (Insert joke about FedEx doing the same thing last year.)

In all seriousness, the brand would look to FedEx, UPS or another regional carrier to provide needed transport, and it would be paying full price plus the neverending surcharges. For smaller shippers, this is what they would have been paying anyway — i.e., smaller shippers don’t receive the large volume discounts their larger counterparts get from FedEx and UPS, so not using them for most of the year won’t be penalized as much as a large shipper that jumps to AMZL. 

Wulfraat added, “If you’re a serious shipper with lots of shipments year round, you need a carrier that can deliver all year.”

Is now the right time for Amazon to do this? 

Despite the Herculean growth, there are still huge gaps in Amazon’s logistics networks. Wulfraat estimates that Amazon currently has about 70% of the population covered, but there are still 10 states in which Amazon has no presence at all. These are of course very sparsely populated areas in places like the Dakotas, Montana and New Mexico. 

At this pace, the gaps won’t last. Amazon currently has 465 delivery stations with another 267 in the works. The rumored target is 1,500, according to Wulfraat. Amazon now has 71 regional sortation centers, with 31 more planned/in construction. At the same time, Amazon Air is adding flights weekly. Amazon Air’s flight activity grew by 15% between August 2020 and February 2021, despite the slowdown in retail shipments that typically occurs after the start of the new year. The airline now regularly makes an average of 140 flights daily, and DePaul University researchers expected that number to reach 160 by June 2021. 

“What took UPS 120 years to build out, Amazon is attempting to do in less than a decade,” Wulfraat stated. Given the current pace, Amazon is expected to reach 85% of the population within the next three years. 

Jason Murray spent 19 years building out Amazon’s fulfillment and distribution network before leaving to start Shipium. Murray told me by email, “Given that there are increasing volume constraints with carriers, Amazon probably looked at the available capacity in their LTL lines and calculated that now is the time to formalize making it available to retailers.” 

Murray continued, “It’s probably similar to the way Prime is managed: What extra pieces of value can be delivered to keep incentivizing customers to use us versus competitors? Prime adding a video streaming service is an example. With the continual explosion of marketplaces, they probably see opening up AMZL as a competitive differentiator to other marketplace options.”

Final thoughts. 

Amazon built its logistics infrastructure not with the idea of competing with UPS and FedEx but to push forward its core e-commerce business. The playbook is very similar, but the product is vastly different to web services. Unlike cloud services, transportation (at least not yet) is not a commodity. Transportation takes assets, people, timing, geographies, etc. that make it a far more complex service to sell. 

So, to think Amazon will in the near future be a major competitor to FedEx or UPS the way AWS is to Microsoft Azure or Adobe Cloud is a bit far-fetched at the moment. Murray likened Amazon’s move to selling sawdust, which speaks to how lumber mills realized every part of their process created output that could be valuable to someone else, including their non-primary markets. Sawmills saw sawdust as an incremental revenue stream, but sawmills didn’t disrupt their core business to sell more sawdust. Neither will Amazon. 

But will it allow Amazon to build density in its network, bring down costs, eliminate inefficiencies and ultimately generate cash flow to further reinvest in building out a bigger network? Absolutely. It will take assassins accuracy, but Amazon has all the data to make aiming an afterthought. “What Amazon is able to do right now is sort of pick off probably the most attractive routes, the most attractive packages and deliver to the most densely populated areas,” said Romanoff. 

If you enjoyed this piece, try Point of Sale, my twice weekly newsletter about the rapidly evolving retail supply chain covering consumer trends, retailer strategies, supply chain challenges and more. Join more than 3,500 supply chain nerds like myself for trends, news, and analysis of every corner of the retail supply chain. Sign up for free here: https://freightwaves.com/pos