How Jeff Bezos aims to conquer the next 

“trillion-dollar market.” The inside story.

In the blistering heat one recent afternoon, a crowd of people gathered down a narrow side street in Govandi, one of Asia’s poorest and most sprawling slums, on the eastern edge of Mumbai. Here, goats and cows jostle for space with nearly a million people amid the fetid, trash-strewn paths and crumbling buildings. Many have no toilet at home, and thousands make their living by scavenging in the nearby municipal dump. But on this particular day, the crowd of men jammed into a storefront in the side street had arrived to try their hand at something entirely new in their lives: online shopping. At the door, a man handed passersby leaflets explaining that the store owner’s laptop connected to hundreds of thousands of items that could be delivered to his establishment within a few days. “I bought a mobile phone,” says Ansari Jameel, 24, amazed at the swift transaction, as he emerged from the store, adding that he had scraped together the money from his job packing vegetables in a local street market. “The prices are good, and I find a lot of things I cannot find other places, like clothes, shoes, and watches.”

Jameel does not know it, but he has just become part of a crucial experiment for one of the giants of the tech world: Amazon.com  AMZN -5.64% . His surprise discovery of Internet commerce is like found treasure to Amazon—and in CEO Jeff Bezos’s estimation, a key part of his strategy for future growth.

The setting for these ambitions is modest, to say the least. But that hardly matters. Or so we discovered when Amazon agreed to let Fortune spend a week on the ground exploring its barely two-year-old operation in India—a journey that took us through four cities, from north to south, veering from razzmatazz glitz to grinding poverty, sometimes within a matter of hours. In pulling back the curtains, Amazon, one of the most private public companies in the world, revealed how it is racing to piece together an immensely complex puzzle—much of which it is having to build from scratch, at giant expense and with painstaking attention to the minutiae, as it tosses out assumptions that American customers have taken for granted for decades. In doing so, the company, an upstart here, has thrown itself into a knife fight with two privately owned and much more established Indian competitors—Flipkart Internet Pvt. and Snapdeal, owned by Jasper Infotech Pvt.—as well as a clutch of smaller Indian startups that are nipping at all of their heels.

Amazon’s 280,000-squarefoot fulfillment center in Hyderabad is capable of storing and shipping 2 million items.Photograph by Benjamin Lowy for Fortune Magazine

It is a fight that Amazon is far from certain of winning, yet one it cannot afford to sit out. The company predicts that India will be its biggest market after the U.S. within a decade and that the Indian e-commerce market as a whole will ultimately be gigantic. “The size of opportunity is so large it will be measured in trillions, not billions—trillions of dollars, that is, not rupees,” says Diego Piacentini, Amazon’s senior vice president for international retail, who oversees operations in Asia and Europe and who is Amazon’s biggest shareholder after Bezos. Over lunch in New Delhi, Piacentini tells me he now spends more time, by far, in India than any other country in his portfolio, shuttling back and forth from Seattle; he jokes that his last two haircuts have been in India. He believes a big payoff from the India push is “unavoidable.” But it will not come cheap. “We know that in order to win in India we need to do things we have never done in any other country,” he says. “We need great people, a great platform, and honestly, a lot of money.”

It is not hard to see why the battle for India is this fierce, nor why Bezos, famously obsessed with analytics, would see it as essential for Amazon’s future. The numbers alone are dizzying. India’s population of 1.25 billion is four times as big as the U.S.’s and more than double Europe’s. And since the median age is 27—a full decade younger than Americans’—the trajectory will be steep. India will overtake China as the world’s most populous country in just seven years, according to the UN. It is now the world’s fastest-growing major economy, and the IMF projects 7.5% growth next year. The roads and railways might be creaking under the strain. Many laws governing business are a confounding tangle, including a law forbidding foreign companies from selling products directly to Indians. That law effectively renders Amazon India a platform for vendors—akin to its “fulfillment by Amazon” program in the U.S.—rather than the company so familiar to Americans, which buys wholesale items in bulk and sells them directly in a Brobdingnagian online store.

“The opportunity will be measured in trillions, not billions—trillions of dollars, that is, not rupees.”

Still, one factor makes India vastly more accessible to global tech companies than that other mega-nation, China: The language of business here is English. “You have a market that is rivaled by only the U.S. and China. And China is very difficult,” says Jayant Sinha, minister of state in India’s Finance Ministry, whose résumé seems tailor-made for his high-level post: He spent years as a partner at McKinsey & Co. in Boston and at the Silicon Valley investment firm Omidyar Network in Mumbai. On the wall of Sinha’s Finance Ministry office, in the grand British-era government buildings in New Delhi, hangs a portrait of the national hero Mahatma Gandhi, who led the 1940s independence struggle, preaching simple living—a far cry from the talk within these walls these days, which is all about investment opportunities and profit-making. For e-commerce, Sinha says that after the U.S., “India is the logical No. 2.”

One reason for India’s huge potential is that it starts from such a small base. Barely one-quarter of India’s population has access to the Internet at home, whether on a smartphone or computer, and only a small fraction of those have ever shopped online. For e-commerce, that means explosive room for growth. Morgan Stanley estimates that revenue could soar to $137 billion by 2020, more than 10 times the 2013 level of $11 billion. That is partly thanks to plummeting prices for smartphones (there are cut-rate deals online for Chinese models like OnePlus and Xiaomi), which about 350 million Indians now own. Google  GOOG -2.25%  and Facebook  FB -2.33%  are among those investing heavily in increasing Indians’ online access, having concluded that their future growth—like Amazon’s—depends on massively expanding the world’s Internet users from the current 1 billion or so.

“The mandate is wider than just selling some products online. We’re building a whole e-commerce ecosystem,” says Snapdeal co-founder and CEO Kunal Bahl, who launched his company at age 27, in 2010, after earning a management and technology degree at Wharton in Philadelphia. He returned home to New Delhi, he says, only because the U.S. refused him a work visa to stay—a spectacularly good twist of fate for the now-32-year-old Bahl, it turned out. His company, built in part with financing from Alibaba  BABA -5.64%  and SoftBank, is valued at an estimated $5 billion.

Bahl is not surprised by the sharp growth in India’s e-commerce. “We know the tectonic shift is already happening,” he says. “We just need to explain to the outside world that it is happening.”

Though indian entrepreneurs saw the shift coming, U.S. companies were slow to arrive. That gave local executives like Bahl a headstart, which Amazon, for one, is now struggling to overcome. In 2007, six years before Bezos launched Amazon India, Binny Bansal, a computer engineer from northern India, co-founded Flipkart with his childhood friend Sachin Bansal (no relation) when they were in their mid-twenties, after the two had worked in Bengaluru (a.k.a. Bangalore) as software engineers for … Amazon. They concluded that Indians would love an Amazon-style shopping site. Just like Bezos, they began by selling books online. Eight years on, Flipkart, valued at about $15 billion, is now India’s biggest e-commerce company, with 44% of the market—nearly triple Amazon’s 15%—and the two Bansals are each worth $1.3 billion. Tiger Global Management, Accel Partners, and South Africa’s Naspers are key investors. Binny Bansal says they plan to take the company public in the U.S. by 2019. When I ask him if Amazon is hampered by having begun years after Flipkart, he says, “Absolutely. When you start late, you have to crunch whatever others have done in eight or 10 years into two years. That is just messy and crazy.”

The craziness was on full display traveling around India in October, when Amazon, Flipkart, and Snapdeal—on perhaps their busiest shopping week of the year—launched a frenzied marketing blitz to kick off the season of Diwali, the Hindu festival of lights, when Indians engage in a massive spending spree. A forest of billboards, the majority for Amazon, sprang up on roadsides and bus stations, promising blowout sales on furniture, clothes, smartphones, and other items; Amazon even ran a lottery for customers with a prize of a gold brick. Eight-page jacket ads for the three companies wrapped newspapers. Within hours of its five-day sale ending, Amazon switched its billboards for new ones announcing another three-day sale just a week later. Flipkart, meanwhile, announced it had sold $100 million in items within the first 10 hours of its five-day sale.

To an outsider’s eyes, Amazon’s ads seemed frenetic and breathless, maybe even a little desperate. But the strategy comes straight from Bezos. To head Amazon India, Bezos had picked Amit Agarwal, a Mumbai native who joined Amazon in Seattle in 1999 after graduating with a computer science degree from Stanford. Agarwal shadowed Bezos for two years as his technical adviser before heading to Bengaluru, in southern India, the closest thing the subcontinent has to a Silicon Valley. “I spent more time with Jeff than my wife,” he jokes over breakfast. Then in 2013, Agarwal and Piacentini met with Bezos in Seattle to present their detailed business plan for India. The boss was unimpressed, they say, regarding their ideas as too cautious and methodical, given Amazon’s late arrival in India. He believed they had to go in with guns blazing. “He challenged us to think like cowboys, not like computer scientists,” Agarwal says. “We needed to move very fast.”

Bezos had good reason to worry. He did not want a repeat of Amazon’s past mistakes—in China, that is. There, Amazon trails far behind Alibaba and other Chinese companies, with no sign of ever catching up. “China has been something of a debacle,” says Benjamin Schachter, Internet analyst for the Macquarie Group in New York City. Amazon has only about 1.1% of China’s market and is so overpowered by Chinese competitors that it finally launched a store on Alibaba’s Tmall platform last March. But what went wrong in China is a matter of dispute. Schachter says investors believe Amazon “has invested a ton of money that has not really paid off.”

Amazon sees things differently. “We should have spent way more in China. On everything,” says Piacentini, an Italian who ran Apple’s Europe operations before joining Amazon in 2000. “And that is why we are not shy in India.” He and Agarwal tell me Bezos has given them wide freedom to run India without his intervention, accepting that it could take years for the billions of dollars in investment to show returns. “His solicitations are never, ‘When will we make money?’ ” Piacentini says. “It’s always, ‘Are we investing enough?’ ”

Motorcyclists line up to receive orders and addresses at an Amazon delivery center.Photograph by Benjamin Lowy for Fortune Magazine

To signal just how much money Amazon was prepared to plow into India, in September last year Bezos flew to Bengaluru in a visit filled with Bollywood-style bling, arriving one day after Flipkart had closed a $1 billion round of financing. He posed on an ornately painted delivery truck outside Amazon’s Bengaluru headquarters dressed in a white Indian wedding suit and handed a $2 billion mock check to Agarwal. The message was clear: He would spend whatever it took to win.

By some estimates the company is spending nearly $25 million a month in India already. “Investors are prepared to give Amazon leeway,” says Schachter. “But they don’t want to see a repeat of what happened in China.”

Whether unbridled investment alone is enough for Amazon to catch up to rivals and dominate in India remains far from a settled question, of course. And some of those competitors, indeed, don’t exactly sound nervous at the prospect. “We’ve also got deep pockets,” says Flipkart CEO Binny Bansal, who adds that he has no worries about staying on top. “This is the last big frontier of e-commerce left. Nobody wants to lose.”

I got a close-up look at Amazon’s no-holds-barred plan late one afternoon in the Janakpuri neighborhood in western New Delhi. Down a side alley sits a warehouse that serves as one of the 4,000 motorbike delivery centers in the country—the first such service in Amazon’s 20-year history. There was a frenzy of action inside as each bike rider filled a large black backpack with packages, grabbed a list of addresses, and raced out the door. Straddling on the back of one of the bikes, I clung to the sides as we weaved through rush-hour traffic, dodging cars and rickshaws, women in saris, and the occasional cow. Experts in their districts, the riders are essential to Amazon’s ability to make speedy deliveries, since the country has an arcane address system, patched together haphazardly over the decades, with many addresses containing descriptions reading something like “behind the mosque, across from the stadium.”

Amazon faces a far bigger logistical hurdle than addresses, however: getting paid. Barely 60% of Indians have bank accounts, and only a fraction of those have credit cards. So Amazon’s payment systems in India are drastically different from any the company has attempted before and involve the kind of hand-holding that would be unimaginable to U.S. customers. About half the customers pay cash only when their purchases are delivered. Amazon has partnered with thousands of small shop owners across the country to act as pickup points in exchange for receiving a small commission per package.

Whisking deliveries through New Delhi’s legendary gridlock.Photograph by Benjamin Lowy for Fortune Magazine

When we finished our deliveries by motorbike that October afternoon in Janakpuri, we dropped by a tiny convenience store in the neighborhood, where a few Amazon packages sat on shelves between the rice and cooking oil. Lokesh Maggo, 33, whose father runs the shop, says he calls customers when their Amazon packages arrive and then collects the cash to give to Amazon the next time they swing by. In the process, his own business has grown. “People come here and then buy other things they need,” he says.

For now, Amazon’s strategy is to persuade Indian stores to sell their products through the company, even if it currently costs Amazon more money than it makes on the transactions. That could be inevitable right now, since its U.S. revenue model—buying in huge bulk at wholesale prices, then selling retail online—is not an option in India, whose laws forbid foreign companies, Amazon included, from direct sales to Indians.

Amazon boasted in its earnings call with analysts in October that the number of Indian sellers on its site had increased 250% in a year. Left unsaid, however, was that recruiting them has required staff to help on almost every aspect of the sellers’ business. One afternoon we watched an Amazon team set up lights and cameras in a small furniture store in Bengaluru so they could photograph a few leather couches and upload them to the site. The store owner tells me he doubted he would sell a single item on Amazon, but decided he “might as well” post his high-end furniture on the site. It was, after all, free.

Much could still go wrong with Amazon’s strategy. The legal obstacles it faces are not ones that affect Indian competitors, including the country’s retail giant, the Tata Group. In February, Tata—late to the game—is launching its own e-commerce platform, effectively turning its thousands of retail stores into an Internet behemoth. Up in the colonial-era whitewashed Tata headquarters in Mumbai, the e-commerce team show me their plan, designed, they say, to put a serious dent in Amazon, Flipkart, and Snapdeal’s trade. “Indian growth is on a tear,” says K.R.S. Jamwal, executive director of Tata Industries. “People like to say the winners have already been chosen. I don’t agree. This is a 10-, 20-, 30-year journey.”

Beyond even its prodigious financial investment, though, Amazon is clearly going all in. Witness its practice of filling out tax forms for some sellers, letting the latter avoid one major bureaucratic headache that is a perennial gripe here; each of India’s 29 states has a separate tax code, making cross-country shipping a laborious process. Amazon will even pick up items from sellers and deliver them—something the company does only in India. Another practice unique to its subcontinent operations: Amazon takes telephone orders for goods from sellers across India, packs the products into bags with Amazon logos, and delivers them straight from the sellers’ stores on Amazon motorbikes. That’s a far cry from the Amazon Americans know—whose core business is built on shipping millions of items, including groceries, from its own warehouses rather than collecting them from outlets. But then that’s a model that would severely limit Amazon’s business in India, where mom-and-pop stores dominate.

For Indian businesses, to be sure, the partnership with Amazon is an unalloyed boon. Sales for local businesses rocketed eight- or 10-fold during the Diwali promotion in October (at least according to Amazon); the company also exports items like saris and handicrafts to 25 million Indians in the U.S. “This is a land of small shops, with people who have probably never accessed the Internet,” Agarwal says. “If you want fast, cheap delivery, you have to build the whole infrastructure yourself.”

Workers in the Hyderabad warehouse, an endless hive of activity, ready packages for shipping.Photograph by Benjamin Lowy for Fortune Magazine

The morning after zooming through Janakpuri by motorbike, I flew to Amazon’s biggest Indian warehouse, outside Hyderabad, 1,000 miles south of New Delhi. Set about a half-hour’s drive from the city’s glassy new airport, through fields and rice paddies, the 280,000-square-foot warehouse is capable of storing more than 2 million items—making it a relative piker compared with Amazon’s gargantuan hangars in the U.S. The Pampers diapers, Dr. Seuss books, math textbooks, Nikon cameras, and Ferrero Rocher chocolates stacked on the shelves represent perhaps something more precious to customers in India: the chance to buy merchandise they have never seen in their towns—including remote villages in the Himalayas and communities in Kerala reachable only by boat.

Propped against the walls of the warehouse are signs with the slogan that captures in a handful of words what is unfolding at hyperspeed around the workers in Hyderabad: “Transforming the way India sells, transforming the way India buys.”

To truly disrupt the shopping habits of 1 billion people, however, Amazon, Flipkart, and others will need drastic transformation of the country itself—change that can come, perhaps, only at the hands of government, not companies. There are signs that it is slowly happening (see “The Last BRIC Standing”). In 2014 Narendra Modi, a charismatic Hindu nationalist, came to power as Prime Minister partly on a sweeping “Digital India” campaign, promising to resuscitate the decrepit manufacturing industry, build highways and railroads, streamline arcane tax rules, create dozens of new cities with high-speed connectivity, and provide banking services to hundreds of millions of people.

Those are all things e-commerce badly needs, and Modi is promising big changes. Even for the few Indians with credit cards, online shopping is still “painful,” in Agarwal’s word, with obligatory two-step authentication and cumbersome interstate trading laws. Parliament is debating whether to introduce a national sales tax, and the government has launched a fund to finance venture capital, modeling it on Israel’s Yozma Fund, which has helped power the rocketing success of Israeli startups. After years of stagnation, Modi seemed like a godsend to Indian entrepreneurs. Little surprise, then, that during his September visit to Silicon Valley, where 15% of startups are founded by Indians, 18,000 people jammed the San Jose Convention Center to hear him speak.

India Fashion Week brings cachet and plenty of exposure for Amazon, a lead sponsor of the event.Photograph by Benjamin Lowy for Fortune Magazine

But California is 8,000 miles—and a galaxy away—from India. “In South Mumbai or Bangalore it’s easy to feel you are in New York or Silicon Valley. We drink the same coffee, listen to the same music, watch the same TV,” says Sinha, the minister of state for Finance. “But step 10 or 20 yards outside that zone, and you are in India. We are a poor, developing country.”

No matter how much Modi transforms his nation, millions of Indians will continue shopping with cash for years to come, the roads will remain muddied and potholed, and thousands of factories will remain dysfunctional. Until now, Sinha says, tech companies have zeroed in on the “top 100 million”—hardly a small number, but a small fraction of Indians. But in creating cash payment systems and pickup points for customers and even allowing those with no Internet connection to shop on Amazon, the company has the possibility of reaching a customer base several times that size. “Amit [Agarwal] is Indian to the core. He totally gets India,” Sinha says. “He’s come in and reinvented Amazon’s model.”

For Amazon to succeed in India, it will have to straddle those two worlds: the wealthy few and the poor masses. If it does, it could have a shot at something else too: helping to create a new model for other emerging markets. If e-commerce takes off across India, the industry could replicate its model for India in other vast developing countries, such as Indonesia and Nigeria. Then, says Sinha, the tech industry would have two global centers: Silicon Valley and Bengaluru. “Just like the U.S. is the economic hub and innovation engine for the top 1 billion people on the planet, India is going to be the innovation hub for the next 5 or 6 billion people,” he says.

In the ramshackle slum of Govandi, where raw sewage clogs the gutters, Amazon’s storefront operation is one sign of how that new hub might look. Here, few people have ever logged on to a computer, and many do not yet have a smartphone. Instead, the store owner guides them through Amazon’s site, then writes down their order in pen in a ledger and takes the cash once Amazon delivers the item to the store, passing the money along to the company, minus a fee.

On a good day, more than 30 people purchase items in his store. It is a small start, but one Amazon believes will rise quickly. “What we do in India will affect Amazon’s future in a very, very big way,” Agarwal says. “If Jeff writes a book on Amazon, India will certainly be a chapter in it.” It will be some time before we know whether that chapter will be titled “Bezos’s Folly” or “The Billion-Customer Bounty.”