The Amazon flywheel: part 2, Amazon and the global supply chain

As a nine year old I used my parents’ dial-up connection to make my first e-commerce purchase: the Mulan soundtrack, via It took three weeks to get from Amazon’s Seattle warehouse to our address in Bellevue, WA. I know because I agonized over the days, and how each one, in succession, delayed my reenactment.

Today’s youth can expect a similar purchase from Amazon to arrive within two days.

All of this is to say, the Amazon supply chain wasn’t always the swift, efficient, ever-expanding machine that it is today.

In last week’s post I covered Amazon’s flywheel and how, through engines of new growth and scale, the company reinvests earnings back into infrastructure. One of these infrastructure investments was Fulfillment by Amazon (FBA), Amazon’s inventory and logistics overhaul.

In this week’s post, I’ll review how Amazon regained control over its supply chain through  the FBA initiative and how, moving forward, the company will continue to expand both downstream -- through continued investment in same-day, urban delivery -- and upstream -- through an expansion in cross-border infrastructure.

As a quick refresher, here’s my version of the Amazon flywheel, which breaks out improved efficiency (and the faster delivery times that result from it) as a part of the wheel and a key differentiator for Amazon’s customer experience.

Amazon Supply Chain Beginnings

In the beginning Amazon sold books. They bought them from distributors, stored them in their warehouses, and sent them off to traditional logistics providers when they received a customer order.  

If we lay this out (and abstract for simplicity’s sake,) we get the diagram below. This entire process was manual.

Growth and Amazon’s Logistics Overhaul

In 2000, Amazon launched Marketplace, bringing 3rd-party sellers into an Amazon supply chain that had grown to include more than just books. The added volume from 3rd-party sellers and additional product categories outstripped the manual processes within their warehouses. Customer orders were often late or unfulfilled (e.g. my Mulan soundtrack.)

Amazon handled this in two ways.

First, automation.

Bezos and his executive team realized that if they couldn’t improve on delivery, they’d be at a serious disadvantage to in-store retailers. In 1998, if I’d gone to Tower Records instead, I would have had my Mulan record that day. Instead, I waited three weeks. The burden of the early adopter.

Amazon poured investment into revamping these warehouses, automating merchandise retrieval and routing. The output of this investment was the massive Fulfillment Center network.

Second, logistics-as-a-service.

Once Amazon built its Fulfillment Center network to a point where it could handle additional capacity, they started offering it as a service to third-party sellers.

All of this fed the flywheel, enabling Amazon to reinvest third-party service income back into their logistics infrastructure, furthering their goals of improved cost structure and delivery time.

If we look back on Amazon’s early day supply chain and evaluate how it changed through the FBA years, we land here:

The distributors were bypassed through Amazon’s direct negotiations with publishers/manufacturers/sellers, and Amazon manual warehouses evolved into largely automated Fulfillment Centers. (That link jumps to a video tour of an Amazon Fulfillment Center. It’s worth your time. The conveyor belt speed at 2:22 is intimidating.)

With all of this we see Amazon gaining cost structure control over its own supply chain components (the warehouses) as well as those components that lived upstream -- i.e. the distributors.

Amazon: Downstream

This takes us into Amazon’s downstream supply chain expansion.

If we look at this through the lens of the flywheel, we know Amazon will expand downstream as long it enables them to a) gain more control over their cost structure and/or b) reduce delivery time for customers.

This strategy started as early in 2014 with the Amazon Prime Now and Last Mile initiatives, projects aimed at building a downstream logistics infrastructure for shipping direct from Fulfillment Centers to customers.

“As a prelude to the U.S. moves, Amazon has been testing a delivery network in the U.K. "We've created our own fast, last-mile delivery networks in the U.K., where commercial carriers couldn't support our peak volumes. There is more invention to come."” -- Bezos in his annual letter to shareholders, 2014

As it stands today, Amazon can only pursue this same-day, direct-to-consumers strategy for those customers within reasonable distance of an Amazon Fulfillment Center. Any farther and Amazon depends on its full-service logistics providers -- UPS, FedEx, USPS -- to fly them to customers.

Amazon had started addressing this Fulfillment-Center-range issue in 2014 through the development of sortation centers: smaller extensions of the Fulfillment Centers that get packages closer to customers.

Now, Amazon is replacing their logistics providers altogether.

Last week brought the official news (everything prior was hearsay) that Amazon would be leasing twenty Boeing 767 freighters from Atlas Air, signaling Amazon’s entry into a space that they’ve historically left to the traditional logistics companies. The response from Amazon management was one of deferral, along the lines of “we’re just doing this to have greater supply chain capacity during the holidays, not to eat anyone else’s lunch.”

I think they’re pursuing a larger opportunity. Here’s an excerpt from an Amazon job listing that does a good job of outlining the company's goals in downstream logistics:

"Amazon is growing at a faster speed than UPS and FedEx, who are responsible for shipping the majority of our packages. At this rate Amazon cannot continue to rely solely on the solutions provided through traditional logistics providers. To do so will limit our growth, increase costs and impede innovation in delivery capabilities." -- Amazon job listing, as cited here.

The twenty freighters Amazon leases will constitute a tiny portion of their shipping volume, but the long-term trends are in Amazon’s favor. They’ll use these first planes to learn the ins-and-outs of air freight, with the long-view goal of adding it as an owned and differentiated link within their end-to-end supply chain.

Taking the air freight and same-day delivery initiatives into account, the Amazon supply chain starts to look something like the diagram below, wherein the role of shipping companies gets smaller and smaller over the long-term.

Amazon: Upstream

What might we expect from Amazon in regards to their upstream supply chain?

Global logistics analytics startup Flexport largely answered the question at the start of the year when they broke the news that Amazon had registered to provide ocean freight services.  It’s a great read and one that I’ll defer to for explaining why this is such a big move for Amazon.

“Ocean freight is cheap right now. As of January 2016, Flexport’s ocean freight customers were paying less than $1300 to ship a 40-foot container from Shenzhen to Los Angeles. More than 10,000 parcels can fit in a single container, so the price for the ocean freight leg could be as low as $0.14 per parcel. Here’s another way to think about that figure: Right now it costs under $10 to ship a flat screen television across the Pacific.

With ocean freight itself so low, a considerable portion of logistics costs come through labor costs—particularly compliance and coordination of cargo handoffs between different players in the chain. It’s here that automation, something no traditional freight forwarding company can do even one percent as well as Amazon can, becomes the key competitive advantage over legacy freight forwarders. By using software to eliminate additional transaction costs associated with government filings, status updates, pricing, booking and more, Amazon will be able to cut their costs significantly. At the same time, fulfilling products directly from China to consumers in the U.S. will cut handling costs at U.S. warehouses.”

To bring that quote back into the context of this blog post, Amazon will extend their reach upstream and into the realm of international manufacturers. It’s a move that would bring Amazon’s ruthless efficiency into an ocean freight industry that’s largely remained an outsider in the world of automation.

Once Amazon starts ocean freight services, it will be moving product from foreign manufacturers directly to its automated fulfillment centers in the U.S., which, at that point in time, will likely have direct routes to Amazon customers through either domestic air freight or urban same-day delivery.

With all of this in mind, the future of the Amazon supply chain:

With this end-to-end visibility and control of their supply chain, Amazon will be able to further throttle efficiencies out of their process, reduce their cost structure, and give lower prices back to the consumer.

Amazon and Global Trade

All of this adds up to a big long-term vision: Global Fulfillment by Amazon, a one-stop shop for international manufacturers to insert themselves directly into a U.S. logistics network. There’s a loose parallel here to the Stripe Atlas model, in which a U.S.-based company provides international businesses with access to U.S. consumers, enabling those businesses to focus on their core competencies (and not on the paperwork associated with entering the U.S. marketplace.)

This cross-border ease of doing business (if Amazon can pull it off) will be a huge advantage for e-commerce over traditional retail. When cross-border trade is involved, it’s much easier for a company to sell to its international customers online than it is for them to establish a storefront presence in their customers's country of origin. This means more retail market share for Amazon. 

Amazon won’t be the only e-commerce company looking to go global. They’ll be competing with Alibaba (et al), who, having conquered Chinese e-commerce, will look to expand to the U.S. market. But Amazon has the better hand here. They cater better to regulators -- if Alibaba wants to operate in the U.S. at scale, they’re going to need to clean up the counterfeit issue -- and, of course, Amazon has their massive FBA network.

There are larger global macro drivers in effect. The U.S. is raising the ceiling on the amount of foreign goods U.S. citizens can import without paying duties.  All signs point to global trade and an automated global logistics network to support it. Few companies are in as good of a position as Amazon to win this huge opportunity.

Amazon Echo and the New Runtime

Amazon Echo and Smartphones in the Home

Last month Amazon announced that owners of their voice-controlled speaker, Amazon Echo, could now use it to control Nest thermostats. This is a big win for Amazon, who gains a sizable user base with Nest and cachet as a hub in the integrated home ecosystem with other manufacturers. (Honeywell is also included in the announcement.)

The Amazon Echo has been a surprise hit with consumers. Upon its initial release, it earned phenomenal customer reviews, and was heralded as a success by a tech cognoscenti that was, until recently, largely convinced that consumers would just leverage smartphones for the use cases presented by Echo (e.g. adding items to grocery lists, asking for the weather.)

Instead, Echo users are finding that, when they’re home, they use Echo more than they reach for their smartphones.

This is the biggest surprise that the Echo presents in its success. Not is it just an alternative to the smartphone, it is a superior alternative when a user is stationary within their kitchen or home. Ben Thompson articulated this well in his post covering the Amazon-Nest announcement.

"What makes smartphones the center of our lives is the convenience of them being mobile and always-connected; at home, though, we are rather stationary yet still connected. In other words, while we know that nothing matters more than convenience, what makes a smartphone so convenient most of the time — the fact it is pocketable and always with us — actually makes it less convenient at home, where getting something from your pocket is a pain, presuming it’s not plugged-in in the other room."

This is a bigger development than it might appear on the surface. In the mobile era, in which popular opinion states that “mobile is the sun” and that all content discovery is driven through it, the success of Amazon Echo signals a broadening in the runtime clients available to end consumers for discovery and general interface with the web.

“We are looking for a new discovery runtime.”

To understand the significance of Echo, it’s important to define what analysts mean when they talk about ‘discovery runtime’. In short, discovery runtime is the client (or interface) a consumer uses when they are looking for content on the internet.

In the web era (pre-smartphone, post-general adoption of the internet) discovery runtime was synonymous with Google. If you needed to find x on the internet, you opened up your web browser and searched for x on Google, regardless of what x was. This is why Google owned the last decade. They controlled the gateway to content discovery. What’s more, because users could only access the web through the browser, and because Google owned browser search, all content searches came through Google, regardless of what was being searched for. The browser-based discovery runtime was content-independent.

With the rise of the smartphone and its fractured app ecosystem, discovery runtime changed. Users started leveraging specialized apps to find things, instead of typing them into Google. This meant that the mobile discovery runtime was content-dependent, or contextual. That is, if you’re looking for a restaurant, you use Yelp. You don’t open up your chrome app and search for the restaurant using Google.

As this shift from web to mobile has taken place, companies (including Google) and investors have been looking for the one mobile discovery runtime that might consolidate the fractured app ecosystem users interact with today -- Google Now? Siri? notifications? messaging apps? As Ben Evans says below, investors are looking for the next Google, the mobile model for directing users to a single discovery runtime that is content-independent.

“Really, we’re looking for a new run-time - a new way, after the web and native apps, to build services. That might be Siri or Now or messaging or maps or notifications or something else again. But the underlying aim is to construct a new search and discovery model - a new way, different to the web or app stores, to get users.” -- Ben Evans

Others say there won’t be a clear victor in mobile runtime, that mobile search will remain contextual and we’ll use different applications based on the context of the situation we’re in. Here’s Fred Wilson in response to Ben Evan’s take on mobile runtime:

“I agree with Ben but I think there won’t be one runtime in the mobile era. I think what is emerging is multiple runtimes depending on the context – “contextual runtimes.” If I’m building a lunchtime meal delivery service for tech startups, that’s a Slack bot. If I’m building a ridesharing service, that’s going to run in Google Maps and Apple Maps. If I’m building a “how do I look” fashion advisor service, that’s going to run in Siri or Google Now. If I’m building an “NBA dashboard app”, that is mostly going to run on the mobile notifications rails.”

What's interesting about both of these takes on the future discovery runtime is that they both assume it will take place on a smartphone. And this is an assumption that has pervaded the tech community since the smartphone gained general adoption. It’s why every integrated home solution of the last five years -- Nest, Wink, et al -- was built off the assumption that its user would be operating their home devices through a smartphone interface.

The New Runtime: Content-Dependent, Environment-Dependent

Now with Amazon Echo, we’re starting to get visibility into a future in which discovery runtime isn’t just content-dependent, it’s location-dependent as well. In this future, there will be a number of devices, each with their own discovery runtime, that we use to find services and products. These runtime we use will be both content and location dependent.

An example of a location-dependent runtime. If I’m at home, cooking with raw chicken germs on my hands (a thing,) it is not convenient to thumbprint-unlock my smartphone and find the next step in my google-searched recipe. I’m going to holler at Amazon Echo ("Alexa") instead. That’s location-dependent runtime.

An example of a content-dependent runtime. If I’m at my computer at work and I need the address of the restaurant I’m headed to that night, I’ll just open a chrome browser tab and find the restaurant. If I need a list of restaurants in the area nearby, I’ll pull out my smartphone so I can filter the search by proximity. That’s content-dependent runtime driving the device I use.

Expand this logic to include other use cases, and suddenly there are lot of potential devices available to users, depending on both the content of the query, and the location in which the user is making it. Mobile will likely win the majority of these use cases, but the point is there’s a fringe available to new (and old) runtimes. Whether you’re driving in your car (smart car operating systems), sitting at your desk at work (browser), or looking for 360 degree content (VR).

For the meantime, Amazon will look to strengthen and expand their position as the runtime of the home. Last week, in addition to the Nest partnership, Amazon announced the release of two additions to their Echo product line, Amazon Tap and Echo Dot, signaling a continued investment in the space. I’d expect more home product manufacturers to announce partnerships with Amazon Echo soon.