by Brad Stone
“What do you mean, an airplane? Are we talking about a plane ticket and an overhead compartment, or an actual plane?” Indresano asked.
“Just get an airplane,” Clark replied.
Indresano asked Mike Roth whether Clark was serious, then enlisted a member of his sortation center team, Scott Ruffin. Ruffin called an old friend who operated an air charter service and rented two Boeing 727s. Amazon had the planes filled with the stranded orders and merchandise from its fulfillment centers in Southern and Northern California and flown to Sea-Tac Airport. As a joke, Indresano bought a Santa hat that was stored in the Ontario, California, FC and had it gift wrapped and delivered to Clark’s home with a holiday greeting.
That was the unofficial birth of the service that would come to be known first as Prime Air, then Amazon Prime Air, and finally, Amazon Air. The confusion was born from Bezos using the former name in his infamous 60 Minutes interview in 2013 to announce that the company was working on aerial drones to ferry individual packages to people’s backyards. Several operations executives told me they were embarrassed by that stunt (which seven years later, had progressed no further than private tests). They used an old internal Microsoft term to describe it: “cookie licking,” or the act of claiming to do something before you actually do it, in order to capture notoriety and prevent others from following.
Unlike the drone program, the effort to lease planes delivered immediate results. After the first two charters in 2014, Ruffin was put in charge of developing the business case for creating an air cargo network. He recruited a half-dozen colleagues and holed up in a windowless conference room in the Ruby-Dawson buildings in South Lake Union, where they met regularly with Clark.
In the resulting white paper, they argued that owning air capacity would allow Amazon to shorten its delivery times and pay only the true cost of transporting cargo in the air rather than the public rates charged by UPS and FedEx Express. Bezos was dubious about buying planes outright and wondered what Amazon would do differently from other logistics companies if it operated a freight airline. Executives were also undoubtedly aware that operating an airline would come with plenty of actual baggage: it could expose the company to a potentially belligerent pilots’ union, a surfeit of regulations, and an oversight authority, the FAA, that had a dim view of Silicon Valley–style corner-cutting and innovation.
The solution they came up with, which Clark successfully proposed to the S-team, allowed Amazon to avoid those drawbacks. Like Amazon’s parallel initiatives on the ground, the plan called for controlling air freight while not necessarily owning it or exposing the company to the dangerous messiness of the aviation industry. The details provide a fascinating window into the self-amplifying advantages of Amazon’s size and power.
Over the spring of 2016, Amazon announced it was leasing forty Boeing 767 freighters from a pair of airlines: ATSG, based in Wilmington, Ohio, and Atlas Air, based in Westchester County, New York. The airlines would continue to maintain and operate the aircraft, but the planes would be rebranded with the Prime Air logo and requisitioned into Amazon’s service for a period of five to ten years.
As part of those deals, Amazon purchased warrants to buy 19.9 percent of ATSG stock at $9.73 a share and 20 percent of Atlas’s parent company at $37.50 a share. Amazon knew that investors in those companies would be excited by the partnership with the e-commerce giant and want to participate in the upside.
Sure enough, after each company announced the news, their stock prices soared—49 percent and 14 percent respectively over the month after each deal was announced. “Fantastic job—that is how it’s done!” Bezos crowed, replying to an update that Clark emailed the S-team. By the second anniversary of the deals, based on the numbers in each airline’s 8K public filings, Amazon had simultaneously achieved its goal of securing exclusive access to air routes and earned nearly half a billion dollars on its investments.
The planes, emblazoned with Prime Air on their sides and the Amazon smile logo on their tails, made their debut at a Seattle air show over the summer of 2016. The press was enthralled by the implications; could Amazon disrupt FedEx, UPS, and the USPS and their hold over the domestic e-commerce delivery business? On an earnings call with investors, FedEx CEO Fred Smith called such talk “fantastical,” specifying that he chose the word carefully.
Publicly, Clark answered questions about the initiative by saying that FedEx and UPS were “great partners” and that Prime Air was supplemental. But Clark’s antagonism toward Fred Smith, borne from the tension in the companies’ relationship, wasn’t very well concealed. “Ho!Ho!Ho! Have a Fantastical Holiday everyone!!!” Clark tweeted later, along with a photo of a Prime Air model airplane in front of a Christmas tree.
Amazon added ten more aircraft to its fleet in 2018, twenty more in 2019, twelve in 2020, and another eleven as of January 2021. The Prime Air fleet allowed Amazon to exploit its increasing sales volume and pack the planes full of bulky yet lightweight items that would be more expensive to ship via traditional carriers. It also allowed Amazon to cater to its peculiar around-the-clock cycle—routing the aircraft from sort centers at 2 a.m. on a Sunday, for example. UPS and FedEx Express, which served thousands of other companies, couldn’t offer that kind of flexibility to a single customer.
Now that he had airplanes, Clark also needed something else: air hubs, to load and unload cargo before and after flights. In January 2017, Amazon announced it would build a Prime Air hub at the Cincinnati/Northern Kentucky International Airport, the location of DHL’s international hub, which Amazon would pay to use while its new facility was being constructed. The $1.49 billion deal was negotiated by Holly Sullivan, Amazon’s director of economic development, who extracted $40 million in tax incentives from local and state governments.
But if Clark thought he would get another commendation from Bezos, he was wrong. Amazon’s new air hub was going to create around two thousand new jobs. In contrast, electric automobile maker Tesla—run by Elon Musk, Bezos’s chief rival in the private space industry as well as for public adulation—had secured $1.3 billion in tax breaks a few years before for a battery plant in Nevada, dubbed the Gigafactory. Tesla was projecting it would create 6,500 jobs. It had earned about thirteen times more in tax incentives, per job, than Amazon.
Bezos, of course, had spotted the difference. Three employees independently recalled his response to the news of the Kentucky deal and considered it a fateful harbinger of the belabored public process, nearly two years later, that would go by the infamous name HQ2. Bezos’s email went something like this: “Why does Elon Musk have this superpower of getting big government incentives and we don’t?”
* * *
In the midst of the Amazon Logistics buildout, the prospect of Bezos agitating for additional government succor was ironic. Amazon had already displaced significant costs onto the public. It didn’t provide healthcare coverage for its drivers, maintain the congested roads to and from its FCs and sort centers, or sustain the temporary employees who got jobs in its FCs over the holidays and then were unemployed and collected public benefits for the rest of the year. In building out a transportation network with considerable speed, Amazon had avoided many of the risks that often accompanied the unsexy business of moving items from one place to another.
But those challenges hadn’t disappeared. And in the press and court of public opinion, at least, Amazon wasn’t going to be able to dissociate itself from them altogether.
On December 22, 2016, an eighty-four-year-old Chicago grandmother named Telesfora Escamilla was struck and killed by a white Nissan NV1500 cargo van emblazoned with the Amazon logo on its rear door. It was operated by a company called Inpax Shipping Solutions, one of the independent delivery firms that existed almost solely to deliver packages for Amazon.
Inpax had a shady record, according to subsequent media accounts. The Department of Labor later found the firm had underpaid dozens of employees and violated labor laws by failing to pay overtime. Its driver, t
wenty-nine-year-old Valdimar Gray, was fired from a previous job at another Amazon delivery company for what was later characterized in legal filings as a “preventable hit and run.” He did not possess a commercial driver’s license but nevertheless got a job with Inpax and worked there for two months before the tragic incident.
Gray was charged with reckless homicide—and acquitted. A Chicago judge agreed with the defense that it was an accident. But a civil lawsuit brought by the Escamilla family against Amazon and Inpax alleged that Amazon had contributed to her death by putting excessive pressure on drivers to make deliveries on time. A week before, Amazon had emailed Inpax and other DSPs in the area about “some struggles the last few days with route coverage and route performance” and proclaimed: “Our #1 priority is getting every package to the customer on time.” Amazon and Inpax quietly settled the suit in March 2020 and agreed to pay the Escamilla family $14 million.
There were similar tragedies, many exposed by investigative reports in ProPublica and BuzzFeed News over the course of 2019. The publications revealed that drivers delivering Amazon packages had been involved in more than sixty serious crashes, including at least thirteen that resulted in fatalities. In 2018, a sixty-one-year-old legal secretary named Stacey Hayes Curry was killed in her office parking lot by a driver delivering Amazon packages for a firm called Letter Ride. The driver told police he never saw Curry and thought he’d hit a speed bump. Her son later wrote (but never sent) a letter to Jeff Bezos, saying, “I think this attitude of reckless speed stems from the top and trickles down,” according to ProPublica.
Amazon executives told me that safety was their top priority and that the company met or exceeded all public safety laws. In an interview, Jeff Wilke said that Amazon’s transportation partners had collectively driven 700 million miles in 2019—and asserted that Amazon had a better safety record than the national benchmark. This claim was impossible to check without specific data about Amazon’s accident rate, which the company declined to provide. “I have not seen a deterioration in the safety culture; in fact, I’ve seen just the opposite,” Wilke said. “I just don’t think it’s true that we had to sacrifice safety in order to build this capability.”
Many former operations employees agreed that Amazon did not deliberately dilute its safety standards, noting that delivering packages is an inherently dangerous business for all companies. But they added that the act of contracting with delivery service providers, instead of acquiring them or simply hiring drivers directly, limited Amazon’s ability to control what was happening on the road.
In its FCs around the world, Amazon was accustomed to wielding almost tyrannical authority and could deploy that power to prevent accidents. Workers began each day with lectures on safety and stretching exercises, for example; drivers, however, received no such support from Amazon, lest they be reclassified as its employees. “You had this weird dichotomy,” said Will Gordon, a senior manager on Amazon’s operations team for three years. “Safety meant different things when you were talking about FCs, versus the contractors who made last-mile deliveries.”
Unable to impose the same culture of safety and efficiency on its delivery partners, Amazon used technology to try to guide them. The company distributed an app to drivers, internally called Rabbit, that scanned packages, displayed addresses, and algorithmically generated the fastest routes for delivering packages.
Rabbit was originally developed for a separate, Uber-like delivery service called Amazon Flex, which allowed individual drivers to sign up online and pick up and deliver Amazon packages for an advertised $18 to $25 an hour. Drivers could set their own schedules but had to supply their own vehicles, fuel, insurance, and smartphones. Flex was meant to deliver Prime Now orders (thus the Rabbit name, which was part of Stephenie Landry’s taxonomy of magical brands). But it quickly became the de facto tool for all drivers delivering packages for Amazon.
Former employees who worked on the app said the development of Rabbit was haphazard and rushed. At first it was missing features that might remind drivers to take breaks or that would choose routes in such a way as to minimize the statistically dangerous act of making a left turn—a key safety feature long known to UPS and FedEx. “Like a lot of things at Amazon, it was ready, fire, aim,” said Trip O’Dell, the former head of the design team that worked on Rabbit. “They just rapidly expanded it and assumed they would fix it later.”
O’Dell said that he and members of his team worried that the app was distracting drivers on dangerous city streets and complained to their superiors. They pointed out that the presentation of information on the app was dense and difficult to read in sunlight, and that constant notifications for new workloads vied for drivers’ attention when their eyes should have been planted on the road ahead.
Issues with the app cascaded, O’Dell said. Routing was poor, and drivers found it difficult to progress from one delivery to another. Fraud was prevalent, with drivers finding loopholes to get paid for work they didn’t do. One problem was that dueling two-pizza teams in Seattle and Austin worked concurrently on versions of Rabbit for iOS and Android, magnifying the confusion among delivery firms. “All of the things were wrong with it all at once,” O’Dell said. “It was not a good app.”
But employees said that Clark was more concerned with resolving the tricky economics of Amazon Logistics, like maximizing the shipments per truck and getting the pay for drivers just right. He made decisions based on cold, hard data, while employees’ initial concerns about safety were based on anecdotal evidence. The Rabbit team occasionally observed drivers and watched them skip meals, rush through stop signs, and tape their phones to their pant legs so they could easily glance down at the screens, all to meet the challenging delivery deadlines.
Without quantifiable evidence of a widespread safety problem though, Clark and other execs largely dismissed the complaints. “I don’t think safety was the number one problem or priority,” said Will Gordon, the senior manager. “It was productivity and cost-effectiveness.”
After the damning articles in ProPublica and BuzzFeed, Amazon cut ties with several delivery firms, including Inpax and other companies linked to accidents. A spokeswoman said that Amazon had “a responsibility… to ensure these partners meet our high standards for things like safety and working conditions.” Implicit in the statement—as close as Amazon ever got to a mea culpa—was that some of the partners it had previously chosen to work with did not meet those safety standards.
There would be no comparable admission about Amazon Air, despite similar circumstances. On February 23, 2019, an Atlas Air cargo flight ferrying Amazon and USPS packages from Miami crashed into a muddy marsh while approaching George Bush Intercontinental Airport in Houston, killing two pilots and one passenger. Atlas had grown quickly over the past three years, in large part because of its deal with Amazon. It employed around 1,185 pilots at the start of the relationship and 1,890 by the time of the accident less than three years later—a 59 percent jump, according to the airline’s public filings.
In the weeks leading up to the crash, the news website Business Insider talked to thirteen pilots who worked for Amazon’s airlines; all thirteen said their pay and benefits were below industry standards, and twelve said the pilots there were less experienced overall than their counterparts elsewhere. One of the pilots alleged that Atlas Air overworked its pilots and that the situation was “a ticking time bomb.”
* * *
By the fall of 2017, Amazon Logistics was delivering about 20 percent of its own packages, a larger share than FedEx, and was on the verge of permanently surpassing UPS, according to the research firm Rakuten Intelligence. After years of effort, it had spurred the creation of hundreds of new small delivery companies and generated a fair amount of chaos and negative headlines. It was also just starting to meet some of the lofty goals executives had set for it.
But Amazon still needed more drivers, particularly during the busy holiday peak. Clark then made two additional moves to attra
ct more delivery firms. In June 2018, he introduced a new program for delivery service providers with fewer than forty vans, offering discounts on Amazon-branded vehicles, uniforms, fuel, and insurance—while still requiring them to operate independently and provide healthcare and pay overtime. The addition of multiple, smaller firms in each city would guarantee Amazon had plenty of partners to work with and ensure that it had yet another kind of leverage: the ability to dictate terms and cut ties with any disruptive or poorly performing companies without compromising the level of service for customers.
A few months later, Clark then authorized the purchase of twenty thousand dark blue Sprinter vans from Mercedes-Benz and leased them at discounted rates, along with blue-and-black Amazon uniforms and caps, to the new delivery firms.
The gambit worked, helping over a thousand new delivery businesses get off the ground. And in early 2019, Amazon eclipsed both UPS and the U.S. Postal Service to become its own largest carrier in the U.S. It was a momentous achievement, even if there had been significant costs along the way.
Clark’s decade-long retrofit of Amazon’s fulfillment centers and its last-mile delivery network had inexorably changed Amazon’s retail business. It allowed the company to place its merchandise closer to dense population areas and to lower its transportation costs, since Amazon no longer had to pay the inflated retail rates of the big carriers. It also aligned Amazon’s delivery expenses with its growth; the more customers signed up for Amazon Prime and Amazon’s grocery delivery services, Prime Now and Amazon Fresh, the more efficient and cost-effective it became to send drivers into those neighborhoods.
