Gregory T. Huang
If you want to understand Amazon’s culture of innovation—and the lessons it holds for startups, big companies, and the tech world at large—you could do a lot worse than talk to Andy Jassy.
Jassy is senior vice president of Amazon Web Services (AWS). In his 16 years at the Seattle-based Internet giant (NASDAQ: AMZN), he has worked on every part of Amazon’s business.
Back in 2003, while serving as founder Jeff Bezos’s “shadow” —basically his chief of staff—Jassy wrote the original proposal for AWS, which offers computing resources and storage as a service over the Web. Amazon rolled out the program in 2006, and the rest is … well, the history of cloud computing as we know it.
In seven years, AWS has accumulated hundreds of thousands of customers. Amazon doesn’t break out its AWS revenues, but analysts say it’s been a billion-dollar business for at least the past year—with the potential to become Amazon’s biggest revenue center over time. Not bad for an internal startup (albeit a well-financed one with tons of infrastructure).
I saw Jassy speak on Wednesday in Boston, at venture capitalist Michael Skok’s “Startup Secrets” event series at the Harvard Innovation Lab. Skok, a general partner with North Bridge Venture Partners, brings in business leaders to talk to local entrepreneurs and students, in part to bridge what he calls a “mentorship gap” in helping startups “figure out how to get through critical milestones” in building a company.
Jassy, for his part, is a Harvard Business School alum from the ‘90s, and he has plenty of business in Boston now. At the event, he talked about everything from the early internal debates about AWS (it was controversial), to the inner workings of Amazon’s innovation “experiments,” to one prominent early customer.
It shed a lot of light on how a big company like Amazon can innovate, as well as how startups might try to operate.
Parts of the AWS story have been told before, but here are seven key points from Jassy that I hadn’t heard previously. A few are specific to Amazon’s cloud computing efforts, but most apply much more broadly:
1. “We write the press release and FAQ before we write any code.” For any new project, Amazon starts by laying out the long-term effect and why it’s meaningful, in a hypothetical PR release. The FAQ then helps determine the nuts and bolts of the proposed service or product, and how the development team should build it. In Amazon’s early years, Jassy says, they had “great high-level ideas,” but they’d get deep into a project only to find the result wasn’t going to be that important. Hence the new approach.
2. “There’s perpetual angst inside Amazon to move faster.” This was part of the impetus for starting AWS—to help software teams become nimbler through IT infrastructure services—but it also drives the company overall. It does create some tension with No. 1 (above), because the company has to balance letting its teams race ahead and build prototypes with the need to flesh out the long-term impact.
3. “To be successful, we had to be the first mover.” Keep in mind Amazon didn’t invent cloud computing; ideas from grid computing and utility computing had been around for years. The company wasn’t necessarily the best-positioned to lead the movement, either (there were far bigger IT companies with more Internet resources). That’s probably why AWS was so disruptive—no one suspected an online retailer would come out with it. “We felt like we were sitting on a secret,” Jassy says.
4. “We picked customers we knew well.” For startups, there’s always a question of who should test their early products. One of the first beta users of EC2 (part of AWS) was none other than former Microsoftie Paul Maritz, who was running the startup Pi at the time. Pi was bought by EMC in 2008, and Maritz went on to lead VMware and now Pivotal (both AWS competitors). As usual, there’s a personal element: Maritz was good friends with a member of Amazon’s AWS team in South Africa, so they trusted him.
5. “The operating system becomes the Internet.” Amazon pursued the broader AWS vision because its experience in running IT services convinced the higher-ups (OK, Bezos) that there was a huge opportunity here. The broad goal was to disrupt 30 years of IT/computing pricing by offering a flexible service with lower prices. In the “titanic shift” that is companies’ software and data moving to the cloud, Jassy says, “this is going to be a high-volume, low-margin business” —precisely what Amazon has relentlessly pursued in retail. To that end, AWS has cut its prices 31 times since 2006, he says.
6. “If you’re doing something big and new, you don’t know how customers are going to respond.” There’s always internal debate over how and whether to proceed with a risky project. AWS had its dissenters, Jassy says, even after it was approved. (He says he had hour-long “therapy sessions” with a senior technical team member about staying the course.) The key, he says, is to “try lots of experiments” with users, and ensure you “don’t have to live with the collateral damage of failed experiments.” In other words, fail fast and move on.
7. “We funded AWS without knowing how big a business it would be.” This one speaks clearly to startups and VCs. Amazon Web Services started with just one salesperson. Jassy’s original proposal called for 57 people total—an ambitious ask at the time. He says the first write-up didn’t even have a financial model. “It could be a $1 million business or a $10 billion business,” he says, depending on which levers were pulled. “Projections were always wrong.” The key was hiring the right people (tenacious, good learners, “not pickled yet”) and letting them run. Still, he admits, “We had to buy a lot of infrastructure before we could monetize. That was a little scary.”
Gregory T. Huang is Xconomy's Deputy Editor, National IT Editor, and the Editor of Xconomy Boston. You can e-mail him at firstname.lastname@example.org or call him at 617-252-7323.Back to all News