Amazon Unbound: Jeff Bezos and the Invention of a Global Empire

by Brad Stone

But Houdini was the relatively easy part. Copperfield, the initiative to sell affiliated grocery store products online for the stores, was arguably the more important initiative; people didn’t mind waiting a few days for Beats headphones or a pair of slippers, but they usually wanted their groceries right away. Seeking a partnership, Prime Now executives and their counterparts in Amazon’s business development group visited the headquarters of Kroger in Cincinnati, Safeway in Pleasanton, California, and Gelson’s Markets in Los Angeles. The grocers were all afraid of Amazon, indifferent to Prime Now, and concerned about Amazon Fresh, even though it only operated in a few cities.

Numerous private-label employees later admitted to taking a shortcut out of this predicament: they exploited Amazon’s massive treasure trove of data. Years later, this fact would become a significant focus of attention for regulators in the U.S. and Europe. They would demand to know: Did the company take advantage of the unique tools and proprietary information at its disposal as a retailer to give its house brands an unfair advantage? Did Amazon, in effect, cheat in its effort to compete directly with its own vendors and sellers? One of Amazon’s databases, Heartbeat, had all customer reviews across the website; accessing it allowed employees to look for revealing patterns that might indicate how they could improve well-established products.

Speaking on the condition of anonymity, several private-label managers admitted to exploiting a resource that was even more precious than product reviews—prominence in Amazon’s search results. When they introduced a new brand, like Mama Bear diapers, a practice called “search seeding” allowed the brand managers to pin the initial relevancy score for the new product to the score of an established product, such as Pampers, at least for the first few days. The Amazon product would then appear at the top of search results, rather than starting on the unseen last page with other new brands.

Amazon’s critics and some of its sellers accused the company of exploiting another significant advantage. By looking at the sales data in the company’s third-party marketplace, Amazon’s private-label managers could rapidly identify new consumer trends and determine what products were selling well and should be copied. Amazon executives claimed they had safeguards in place to prevent this kind of data snooping from happening.

the U.S. and Europe. The question, ultimately for regulators, was whether all this gave Amazon unfair advantages. Back in 2017, when Doug Herrington’s private-label expansion was in full swing, the data and internal tools almost certainly helped amplify the team’s efforts and meet their ambitious S-team goals. But much of the data they gleaned was also easily available to competitors, either by scraping the Amazon website or via research companies that collect data on consumer trends, like Nielsen. At least in consumables, many of the private-label products from that time—from Happy Belly peanut granola bars to Wickedly Prime roasted almonds—didn’t appear to hurt rival brands, at least any more than the similar efforts of other large retailers.

But Bezos intended to keep trying. He had two additional ideas for how to forge a connection with grocery shoppers and solve the quandary of CRaP. They remain among the strangest projects in Amazon’s history and provide an additional glimpse into its odd corporate rituals. The first, which Bezos proposed in a free-flowing brainstorm session in 2014, started as a notion he called “the steak truck.” Imagined as “an ice cream truck for adults,” the original suggestion was to stock a van or truck with steaks, drive into neighborhoods with lights flashing and horn blaring, and sell them to residents, as Doug Herrington remembered it. It would be convenient and a great deal for customers, since the meat was being sold in bulk. Eventually, the company might even predict demand and eliminate the inefficiencies and wasted food of supermarkets. The idea was perhaps just kooky enough that an executive was assigned to write a PR FAQ, the Amazon memo that kicks off the development of a new project by imagining its press release. The document gave the idea an official name: Treasure Truck. Bubble machines and digital displays would convey a carnival-like atmosphere while the truck sent out text alerts to the smartphones of nearby customers, announcing the item on sale that day.

The team commissioned new trucks, not as garish or pricey as the original, and expanded to twenty-five major U.S. cities. But the service was never as ubiquitous or as endearing as Jeff Bezos and Doug Herrington had hoped. Internet critics were baffled by the project and sneered at some of the more inexplicable deals (“bidet sprayers for $19.99, 33% off!”).

In August 2015, the Washington Post published an unappetizing article about how a single hamburger might contain the meat of up to a hundred cows. Sourcing a burger from just a single cow could theoretically produce a superior-tasting patty but that “would be hard and expensive,” a meat distributor told the paper.

The project was assigned to a new culinary innovations team inside Amazon Fresh and immediately established as an S-team goal—a high-priority benchmark monitored closely by Bezos and the leadership council. A product manager named Megan Rosseter was then charged with finding a way to actually produce it. The meat vendors she initially contacted told her that such a thing was totally impractical and would in fact be disruptive to their operations. “I felt like I was always getting crazy daunting goals that seemed almost impossible,” she said. Somehow, Rosseter and her colleagues found a ranch in San Diego County, near the Mexican border, that could produce the burger.

Over the next few years, Amazon methodically redesigned the Kiva robots and moved Kiva’s software to AWS. Then it introduced the machines into its newer FCs, with profound results. As Clark had hoped, they magnified worker productivity and decreased the rate of growth of Amazon’s seasonal labor needs relative to its sales. They also allowed Amazon to build denser fulfillment centers, with the shelf-toting robots swarming over the ground floor as well as a series of reinforced mezzanines. In a 2014 TV interview, Clark estimated that Amazon was able to get 50 percent more products per square foot into new fulfillment centers than the previous generation.

The day after the 2013 holiday meltdown, Clark called Michael Indresano, a former FedEx executive who had joined Amazon as a transportation director, and asked how many “sortation centers” they could build before the next holiday peak. These were the facilities that consolidated packages by zip code and injected them into the U.S. Postal Service for last-mile delivery to people’s homes. Indresano estimated that he could open sixteen by the end of 2014. “Build ’em all!” Clark replied. Inside Amazon, the rapid creation of sort centers in cities like Atlanta, Miami, and Nashville was dubbed “the Sweet 16.” During this buildout, Amazon packages also began showing up to customers’ homes on Sundays. Bezos, said an ops executive, was frustrated with UPS and FedEx’s refusal to deliver over the weekends. Clark and his colleagues had found an ingenious solution: a deal with the USPS to deliver on the proverbial day of rest. The arrangement quickly reinforced the post office as Amazon’s top carrier by volume and allowed Amazon to achieve a total lower cost per delivery than it could achieve by using UPS and FedEx.

A primary concern, said executives who were in those meetings, was not whether Amazon could effectively build such a complex network. Rather, it revolved around whether getting into the transportation business would increase Amazon’s exposure to unions. Delivery stations would have to be placed in the urban areas where most of Amazon’s customers lived—places like New York City and New Jersey that were the locus of the organized labor movement. Clark and his colleagues assuaged themselves by considering the nonunion workforces of FedEx Ground, DHL, and pretty much every other ground delivery firm that competed with UPS and its unionized delivery workforce. Amazon would use the exact same model in creating Amazon Logistics, its new transportation division. It wouldn’t directly hire anyone but instead create relationships with independent delivery companies—DSPs (delivery service partners)—that employed nonunion drivers. That would allow it to pay lower labor rates than UPS and to avoid the prospective nightmare of drivers bargaining collectively for higher wages, which could destroy the already fragile economics of home delivery.

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.

The last-mile network also protected Amazon from the exigencies of UPS and FedEx, and the political winds that battered the USPS. When Donald Trump accused Amazon of ripping off the U.S. Postal Service and threatened to raise rates, the company disagreed, but the result of the dispute hardly mattered. Amazon now had the leverage to shift that volume to its own network and to another shipping partner, UPS. The carrier had recognized that Amazon had changed the dynamics of the industry. In 2019, UPS announced that it would finally deliver on Sundays, bowing to customer expectations and the pressure that Amazon’s around-the-clock cycle was exerting on its e-commerce rivals. As the post office had years before, UPS had to renegotiate its contract with the Teamsters to allow it to create a new category of drivers who would work weekend shifts at a reduced pay scale.

Blue remained secretive, struggling with the dysfunction encoded into its genetic makeup by Bezos, who had otherwise succeeded in nearly everything else he had created.

Bezos let all those plates spin on their own, returning to them only occasionally and usually without warning to generate provocative new ideas, clamp down on costs, and whack at the gathering bureaucracy. On his own time, he tinkered with the business and technology of the Washington Post, supervised the new management at Blue Origin, and exulted in launching test flights of the New Shepard rocket from his ranch in West Texas. He also considered plans for charitable work in the wake of public pressure to give away a fortune that had surpassed $100 billion. And as always, he contemplated Amazon’s long-term future—not just the dramatic new things it could do, but where it would do them.

Instead of developing numerous offices in multiple cities, or negotiating privately with one location for a satellite office, Amazon would announce its intention to create a second headquarters—an equal to its home in Seattle. It would then open the site selection process to all cities in North America and allow them to compete for a prize of some fifty thousand jobs and $5 billion in capital investments over a span of fifteen years. Such a process, Bezos argued, could highlight what communities coveted from the company instead of what its critics feared about it. “Part of it was a cheerleading exercise,” said a member of the HQ2 team, who like many others involved, declined to speak on the record for fear of retribution. “Who wanted us? That would come through in the process itself.”

Idealism largely pervaded the effort. Members of the HQ2 team earnestly believed that any city had an opportunity to win. At one point, they put their predictions for the finalist cities on pieces of paper and sealed them in an envelope. Holly Sullivan, Amazon’s eloquent and well-connected economic development director, came closest to guessing the outcome. “We genuinely thought we were working on the most important economic development project in a generation and were going to change the lives of hundreds of thousands of people,” said one of the HQ2 employees. In early January, Sullivan and finance director Bill Crow presented all of the data and applications to the S-team. Colleagues say Bezos, cognizant of the responsibility of giving every candidate a sincere look, read through the applications of all 238 regions. The review lasted for hours.

In the days before the finalists were announced, the HQ2 team divided the list of more than two hundred cities that hadn’t made the cut and placed phone calls to local officials, alerting them to the bad news. Most asked why, while expressing disappointment with the amount of time and resources they had put into the failed effort. Amazon employees responded with blasts of data. “Your metro area only has 375,000 people and of those only 10 percent have advanced degrees,” went a typical explanation. “Sorry, that’s not enough of a labor pipeline.” City officials mostly agreed that Amazon’s outreach was conscientious, and that Holly Sullivan in particular spent more time than she needed answering questions and preserving relationships that might be helpful to the company in the future.

Prepared in June 2018, this first paper broke the twenty finalists into three groups: “not viable,” “hotly debated,” and “top tier.” Austin, Columbus, Denver, Indianapolis, Miami, Montgomery County in Maryland, Newark, and Pittsburgh were all cast in the first category and dismissed, largely because they were too small and did not have the required infrastructure and talent. In addition, the team had concluded from their visits and from analyzing public sentiment that Austin and Denver might be hostile to Amazon’s presence.

Atlanta, Boston, Los Angeles, Nashville, Toronto, and Washington, D.C., were listed in the “hotly debated” category. The HQ2 group cited high costs and high taxes as negatives for Boston and Toronto.

The “top-tier” locations were Chicago, Dallas, New York City, Northern Virginia, Philadelphia, and Raleigh. Despite recommending these cities the most, the HQ2 team still expressed concerns about them.

“Our goal with this next HQ2 milestone is to continue driving positive press and fortifying our corporate reputation, while not giving our critics unnecessary ammunition or feeding the perception that this is an over-the-top reality show,” the document stated, before recommending three, surprising finalists from months of travel, meals, speculation, and negotiation: Chicago, Philadelphia, and Raleigh. “The locations do not have the largest concentration of existing tech talent but we believe they have the foundation for talent growth across our many businesses,” the paper concluded.

HQ2 managers emerged from the meeting with a radically different short list than the one they had proposed. In the second document I reviewed, produced in August, the team reflected that they had left the June session with a decision to follow up on five locations: Dallas, Los Angeles, New York City, Northern Virginia, and Nashville. That eliminated the three top contenders the HQ2 team had recommended, though the paper suggested they revisit Chicago, but only “to minimize potential negative reaction if Chicago does not move forward in the process.” The priorities in the HQ2 search had changed. The hunt for the most robust incentives package had been replaced by an interest in the largest cities, the best opportunities for recruitment, and the friendliest political environment.

In addition to identifying Bellevue as an immediate alternative for headcount growth, some Amazon executives concluded that HQ2 would now have to be bigger than previously planned and most likely ramp up faster than initially expected. By the time the seventeen-page August document was written, the HQ2 and S-teams were homing in on New York City and Crystal City in Northern Virginia—regions they believed could accommodate the coming expansion.

There was plenty of blame to go around. The mayor and governor had enticed Amazon to Queens without securing the backing of its local politicians. Those leaders were also at fault; they assembled the opposition atop the falsehood that Amazon was getting an indecorous $2.5 billion handout, rather than a rebate on the sizable tax contributions it would make over the course of two decades. They also played on innate fears that the character of a cherished community and its surrounding neighborhoods would change. Yet much of Long Island City had gentrified years ago, and most of the lower-income housing in the area and surrounding neighborhoods was either rent-stabilized or belonged to large public housing complexes whose residents were protected from rising rents. And the alternative to rising home prices and an increased cost of living is rarely stasis; usually, it’s falling home prices, a lower cost of living, and hopelessness. By rejecting Amazon, an outer borough undergoing its own dramatic transformation was robbed of an economic injection that may have tangentially benefited its poorer residents.

Inside Amazon, little self-reflection followed the New York City fiasco. The D.C. team did not author a “correction of error” or COE report—which is often the case when Bezos himself is partly responsible for some mistake.

The media, most observers, and even his own extended family would later condemn Sanchez for this astonishing act of betrayal. But in his own mind, at least—distorted by bitter resentments, years of feuds with his sister, and the dysfunctional dynamics of a complex family—Michael Sanchez believed he was cleverly manipulating the Enquirer. His sister and Bezos were conducting their relationship out in the open and it was only a matter of time before their families and the larger world discovered it. He was trying to “bring the 747 in for a soft landing,” as he later put it, referring to the delicate process of allowing the couple to inform their respective spouses, initiate divorce proceedings, and then introduce their relationship to the public. “Everything I did protected Jeff, Lauren, and my family,” Sanchez later emailed me. “I would never sell out anyone.” He also believed his source agreement with AMI precluded the media company from using the most embarrassing material he had provided them.

But on one issue, at least, Sanchez appeared to tell a straightforward truth. He later told FBI investigators for the Southern District of New York that he never actually had an explicit photograph of Bezos in his possession. In the meeting with Enquirer reporter Andrea Simpson on November 21, with Dylan Howard and James Robertson watching via FaceTime from New York and recording the transaction, Sanchez didn’t show them a picture of Bezos at all, but an anonymous photograph of male genitalia that he had captured from the gay escort website Rent.men.

In the midst of the crisis, Michael Sanchez innocently suggested he could exploit his relationships with editors at the Enquirer to find out what materials they had. After signing a $25,000-a-month contract with his sister to help her navigate the descending insanity, he called Dylan Howard to announce that he was acting as his sister’s representative and suggested that he come to New York to review the paper’s reporting (which, of course, he had provided). Confident in the promise of confidentiality from AMI, Michael Sanchez was now playing both sides.

Pecker and his colleagues had first faced the allegation that they were operating on behalf of Donald Trump; now they confronted the insinuation that they were in league with the Saudis. With Howard sidelined and on vacation in Mexico, the company decided that it need not honor its confidentiality promise to Michael Sanchez. “The fact of the matter is, it was Michael Sanchez who tipped the National Enquirer off to the affair on September 10, 2018, and over the course of four months provided all of the materials for our investigation,” the company said in a statement. “His continued efforts to discuss and falsely represent our reporting, and his role in it, has waived any source confidentiality.” But AMI’s disclosures carried little weight, particularly since its executives had lied so flagrantly about the “catch and kill” allegations related to Trump. Hints of an international conspiracy endured—in large part because of the improbable truth in the idea that the government of Saudi Arabia was out to get Jeff Bezos.

De Becker then commissioned an examination of Bezos’s iPhone X. The eventual report by Anthony Ferrante, a longtime colleague of de Becker’s and the former director for cyber incident response for the U.S. National Security Council, concluded that the promotional video about broadband prices that MBS had sent Bezos the previous year likely contained a copy of Pegasus, a piece of nearly invisible malware created by an Israeli company called NSO Group. Once the program was activated, Ferrante found, the volume of data leaving Bezos’s smartphone increased by about 3,000 percent.

Had MBS’s regime learned of Bezos’s relationship with Lauren Sanchez and tipped off the National Enquirer or even supplemented the information it received from Michael Sanchez? That possibility might make certain logical sense if you squinted hard enough. David Pecker had once unsuccessfully courted Saudi investors to finance AMI’s purchase of Time. To boost their prospects, AMI executives had even produced a sycophantic ninety-seven-page glossy magazine, called The New Kingdom, on the eve of the crown prince’s U.S. tour. But at least from my vantage point, at this particular moment in time, there is not any conclusive evidence to support the hypothesis that the Saudis alerted the tabloid paper to Bezos’s affair—only a fog of overlapping events, weak ties between disparate figures, and more strange coincidences.

In October, he turned up unexpectedly outside the former Saudi consulate in Istanbul to commemorate the one-year anniversary of the murder of Jamal Khashoggi. Gavin de Becker handled the intricate security arrangements. Bezos sat next to Hatice Cengiz, Khashoggi’s fiancée, and embraced her during the ceremony. “Right here where you are, you faced that street for hours, pacing and waiting, and he never came out. It is unimaginable,” he said. “And you need to know that you are in our hearts. We are here.” The trip into potentially dangerous territory was another sign to employees of the Washington Post that their owner would stand up for their journalism, whatever the personal cost. It was also an arrow aimed directly at Bezos’s enemy, MBS.

“All big institutions of any kind are going to be and should be examined, scrutinized, inspected,” he told the private equity billionaire David Rubinstein during an onstage interview at the Economic Club in Washington, D.C., in 2018. “It’s not personal, it’s kind of what we as a society want to have happen.” He sounded almost resigned toward whatever outcome might result: “We are so inventive that whatever regulations are promulgated or however it works, that will not stop us from serving customers.”

Bezos’s longtime PR deputy, Drew Herdener, urged the communications staff to consider “every blade of grass,” to never overlook even minor facts and subtle insinuations if they were perceived as incorrect and merited a response.

Bezos opted for the most aggressive plan, raising the entry-level U.S. hourly rate across the board to $15. At the same time, he compensated for at least part of the additional expense by discarding supplemental sources of worker income, such as stock grants and collective bonuses that were awarded to employees based on the performance of their facility.

The move was tactically brilliant. Amazon had surveyed its warehouse workers over the years and found a large majority were living paycheck to paycheck and would rather have the instant gratification of up-front pay than stock grants. By getting rid of the grants, Bezos not only helped to partially offset the pay increase but eliminated another incentive for unproductive or disgruntled low-level workers to stay at the company for more than a few years. The wage hike also satisfied many of Amazon’s critics, made arduous fulfillment center work more appealing to job candidates, and put the company in a position to lobby for raising the federal minimum wage, which its less affluent retail competitors could ill afford.

Lina Khan, who in her last year of Yale Law School in January 2017 published a ninety-three-page article in the Yale Law Journal titled “Amazon’s Antitrust Paradox.” The paper challenged the recent history of lax antitrust enforcement in the U.S. and invited authorities to take a harder look at the e-commerce giant, which she believed uniquely demonstrated how current laws had failed “to capture the architecture of market power in the twenty-first century marketplace.”

Hannah Arendt’s 1958 book The Human Condition, which examines how modern technologies affected democracy.

The report also struggled to establish that Amazon even had the kind of monopoly power that might make certain conduct illegal under U.S. law. The industry’s most widely cited data gatherer, eMarketer, put Amazon’s share of U.S. e-commerce sales at 38.7 percent. The success of Walmart and Target’s websites, as well as the Canadian company Shopify, which developed sites for brands to sell directly to customers, also belied the notion that Amazon had any kind of hammerlock on the industry. (Only Amazon’s utter dominance in the U.S. in books and e-books, estimated at 42 percent and up to 89 percent respectively in 2018, might merit a deeper look.) “While there is no single seller online that can match Amazon, it’s pretty hard under conventional antitrust to find someplace where Amazon has monopoly power,” said Jay Himes, a former antitrust bureau chief in the New York Attorney General’s Office.

Not mentioned in the report, to quell the chaos of the Amazon Marketplace, lawmakers could also reform the notorious Section 230 of the Communications Decency Act, which currently holds that internet providers like Amazon are not liable for the legal infractions of their users. Changes to Section 230 could force Amazon to be accountable for fraudulent or unsafe products sold on its site by third-party sellers. Regulators could also compel Amazon to verify sellers with a tax ID number or require them to put down a security deposit, which they’d forfeit with any signs of fraud (Alibaba’s Tmall website does this).

To make a credible case for breaking up Amazon, they will have to answer questions that the company goes to remarkable lengths to obscure. How do its various components interlock? How do you gauge the true profitability of the individual business units when some costs are covered by the subscription fees of services like Amazon Prime? Is it anticompetitive when Amazon increases those fees or seeks to improve its market position by dropping them altogether, as it did in the fall of 2019 when it waived a separate $15 a month charge for grocery delivery? “There is no company in the world that is more complex and difficult for outsiders to understand than Amazon,” said Kurt Zumwalt, Amazon’s treasurer for fifteen years, before he left in 2019. “This is not a typical corporate conglomerate like Berkshire Hathaway or General Electric. Almost every aspect of Amazon is built around subtly increasing its connection to customers. The power of the business model is the combination of the sum of its self-reinforcing businesses and services, enabled by world-class technology, operational excellence, and a rigorous review and measurement process.”

The company insisted that it was not retaliating against these employees for speaking out. In each case, Amazon spokespeople described an internal policy that had been violated, such as social distancing or the guidelines against talking to the media without the company’s authorization. But that was difficult to believe. While Jeff Bezos and his colleagues had bristled at external criticism over the years, they seemed to find it completely intolerable when it came from inside the company. It was as if they feared that an incendiary spark from within their own ranks might finally ignite the long-feared inferno of a disgruntled and activist workforce.

Jeff Bezos’s mission had been to stave off stasis and to keep Amazon a “Day 1” company with an inventive culture and durable customs that would outlast him. “Amazon is not too big to fail,” he once warned employees at an all-hands meeting. “In fact, I predict one day Amazon will fail. Amazon will go bankrupt. If you look at large companies, their lifespans tend to be thirty-plus years, not a hundred-plus years.” It would now largely fall to Andy Jassy to prevent that dark possibility. Among the new CEO’s greatest challenges will be retaining the company’s deep bench of experienced senior leaders, even if Amazon’s stock price stagnates; keeping the ever-growing warehouse workforce motivated and happy; and navigating the impending regulatory scrutiny inside and outside the U.S. That climactic battle, which might one day culminate in an antitrust lawsuit against Amazon by the U.S. government, was still largely ahead. But with its fourteen sacrosanct leadership principles, interlocking business units, and overwhelming momentum, Bezos had seemingly set up Jassy and the company to flourish well after he stepped aside. In this respect, his life’s work, at Amazon at least, was arguably done. Still looming, of course, was a definitive answer to a perennial question that even now is almost impossible to resolve: Is the world better off with Amazon in it?