It started at an off-site meeting at Jeff Bezos’s house on Lake Washington in 2003, the first year Amazon reported a profit. The online bookseller was expanding its online store to include more items. Annual revenues had passed $4bn and were growing fast.
The off-site’s purpose was to identify Amazon’s strengths and discussion turned to its ability to run technology infrastructure, which it had developed to operate its website and those of other retailers. The germ of Amazon Web Services was formed.
One person championing the idea was Andy Jassy, a business school graduate working as a “shadow” to Bezos, a role somewhere between technical assistant and chief of staff. He saw that AWS could help solve a problem for Amazon’s software engineers, who were spending too much time figuring out the computing infrastructure for each new project. They felt “they were all reinventing the wheel on the infrastructure pieces and nothing they were building scaled beyond their own projects,” Jassy recalls. AWS, which provides on-demand computing power in the cloud, could fix that.
Today Jassy heads AWS and one of his tasks is to maintain the open-mindedness that enabled Amazon to say yes to AWS those years ago. “A lot of companies really will only pursue businesses that are adjacent or look like an extension of their current business — which by the way is a completely rational strategy.” Amazon is different, he explains: it considers the size of the opportunity, whether it is underserved and whether Amazon can do something new. “When we like the answers to all those questions, we pursue the business even if it has hardly anything to do with the other businesses that Amazon is in.”
When AWS first got off the ground, it was not obvious it would be a hit. It proposed the radical idea of allowing any programmer with an Amazon account the ability to rent computing power whenever needed, whereas previously most companies bought and ran their own servers.
“Ten years ago, the idea that you would run critical applications in the cloud, and that ‘Amazon’ was a word that would even be associated with that, would have seemed like an impossibility,” says Rich Wong, a software investor at Accel Partners.
After launching its first service 10 years ago, AWS grew up alongside many of today’s most prominent start-ups — and its existence enabled these companies to operate cheaply and scale quickly, because they no longer had to buy expensive servers. From Airbnb to Yelp to Slack — even Netflix’s online video collection — all run on AWS. Over time more mature businesses started using AWS too, and a turning point came in 2012 when AWS won a contract with the CIA (beating incumbent IBM in the process).
AWS has become the biggest technology infrastructure provider in the world — and it is also the fastest growing and most profitable part of Amazon. Its annual sales, at $8bn, are less than a tenth of Amazon’s retail revenues but some Wall Street analysts calculate that it will become more valuable than Amazon’s retail business as soon as next year.
All this was gestating behind a veil of secrecy until last year when Amazon began disclosing financial results for AWS. As it stepped out of the shadows, so did Jassy. He might seem an unlikely leader for the business, in that he lacks a technical background — he is not a programmer. A native of New York, he attended Harvard College in Massachusetts and his first job after graduation was project manager for a collectibles company, MBI. It was a great business training, he says, even if he was less than interested in the figurines the company was selling. He went on to Harvard Business School and joined Amazon in 1997, the year that it went public. It had just a few hundred employees and had reported $15m in sales the previous year. Jassy says he joined because Amazon clearly had ambitions to be more than a bookseller.
Today Jassy comes across as a dyed-in-the-wool Amazonian and often reaches for a company slogan to illustrate a point. Some of these might sound odd at first to an outsider, such as the principle of “being right a lot”. He explains this simply means having good judgment. Another favourite saying is that “there is no compression algorithm for experience.” This turns out to be a personal slogan and his way of explaining why AWS has such an advantage over later competitors such as Microsoft and Google. As a first mover, AWS has gained experience for which there can be no substitute.
He is a firm believer in the Amazon principle of “having backbone” and says it is important for employees to respectfully challenge each other.
One of his top lieutenants confirms that he runs meetings this way. “Andy is totally comfortable with contention,” says James Hamilton, a vice-president and senior engineer who joined AWS from Microsoft eight years ago. “It is respectful contention and eventually we reach a decision based on the data, but meetings are hotly debated. There is never an opportunity when you can just sit back and observe.”
He recalls one meeting where Jassy proposed slashing the price of a service by 80 per cent, while everyone else on the team felt that it was impossible. “The fact that he is the only one arguing the point, and everyone else has a different perspective, doesn’t bother Andy one bit.” After hours of discussion, the team agreed on the price cut.
Hamilton says such debates do not prevent the company from being nimble. “It’s getting to be the size of a business where it seems impossible it could keep running like a start-up. But to me it seems like the world’s largest start-up.”
Amazon takes maintaining that culture very seriously, even when now with over 200,000 employees it continues to grow. Jassy says certain practices help codify a start-up culture. One is hiring people who are “builders” and like to tinker. Another is avoiding long roadmaps and rigid project schedules; those of AWS are “changing constantly”. A third is organising people into small, autonomous teams that combine technical and non-technical people.
Yet another is the near-mythical status of Amazon’s six-page planning documents, also known as “working backwards” documents. These are the set pieces around which any decision on a new initiative is taken; they begin with a press release about the proposed idea, and include an FAQ and appendices.
“We often will go through several versions of those ‘working backwards’ documents before we write the code,” Jassy says. When he was writing the six-page plan for AWS, he went through 31 drafts.
On Wall Street, it is hard to overstate the enthusiasm for AWS. Amazon’s share price almost doubled in 2015, largely because of tremendous investor response after the first financial details of AWS were revealed.
Yet the risks to AWS are widening. Microsoft and Google, are investing heavily in their cloud offerings. In February, Google poached one of AWS’s marquee clients, Spotify.
In Silicon Valley, some are starting to grumble that AWS is becoming “too big to fail”, and they worry about systemic risks in the event of an outage. Analysts fear that AWS could lose revenue as Apple starts building its own data centres; Morgan Stanley estimates this could affect around a tenth of AWS revenues. Neither AWS nor Apple would confirm that Apple services are hosted on AWS.
These concerns have not prevented several Wall Street analysts from building colossal growth projections into their financial models; several predict that AWS will soon be as valuable, or more valuable, than Amazon itself.
Jassy says he tries to ignore all this. The public perception of Amazon is often out of sync with what is happening inside the company, he argues, pointing to the obituaries that were being written about Amazon during the dotcom bust as an example. “You can’t feel 30 per cent smarter when the stock is up 30 per cent, because then it means you have to feel 30 per cent dopier when the stock is down 30 per cent, and usually neither is true,” he says.
He refers back to the shareholder letter that Bezos wrote in 1997, the year that Jassy joined the company. It tells Wall Street that Amazon would not be trying to please the financial analysts and would instead focus on long-term growth. Jassy says this is why he can ignore the stock market gyrations.
“We’re trying to build a business that outlasts all of us,” he adds, again quoting an Amazon slogan.
The task ahead is to make sure AWS does that.