Amazon.com Chief Executive Jeff Bezos has always been a controversial figure. His big-spending early on to expand Amazon (AMZN) from an online bookshop to a Web department store backed by a massive distribution system led some analysts to predict its demise. Not only has Bezos' company—once derided as "Amazon.toast" and "Amazon.bomb," survived, it has been profitable for the past three years.
But no sooner had Amazon printed black ink than Bezos began spending anew on technology to expand Amazon's horizons even further. That has profits sagging. In 2006, thanks to a 52% jump in technology spending so far this year, Amazon is expected to earn only half the profits of last year, on $10.5 billion in sales.
Now, the outlines of Bezos' next big idea are starting to come to light. Essentially, he wants Amazon to run your business, at least the technical and logistical parts of it. Using the massive computer system and physical-distribution network that powers its retail business, Amazon is now starting to offer those services over the Internet to other businesses (see BusinessWeek.com, 11/13/06, "Jeff Bezos' Risky Bet").
They include an on-demand computing service called the Elastic Compute Cloud (EC2), a virtual disk drive called the Simple Storage Service (S3), and something called Amazon Mechanical Turk, a marketplace for online piecework where businesses can parcel out small jobs such as transcribing snippets of podcasts.
His booming laugh intact, Bezos recently sat down with BusinessWeek Silicon Valley bureau chief Robert D. Hof to talk about his thinking behind these strange new ventures and what they say about the future of Amazon and the Internet at large. Following are excerpts from that conversation:
Many people don't know what the heck Amazon is doing with this new stuff. They say, "I thought this guy was a retailer. Why's he doing this?"
The answer is, we've been doing it the whole time. The only difference is, we're now exposing it, making it available to others. A huge fraction of our effort historically has gone into building the infrastructure that lets a Web-scale business run. We have all the scar tissue to [show people we] know how to do it.
We think it's also going to be a very meaningful business for us one day. So our motives for doing it are very straightforward: We think it's good business.
Still, it's unusual for companies to serve both consumers and businesses in such different ways, isn't it?
I think it is fairly unusual. But we have three businesses today, and they're all customer-facing. There's the business that has the 59 million active customers, it's our consumer-facing business, and that's the business that we're best known for. Then there's our seller-facing business, so that's the second set of customers. That business has grown very rapidly, and we have everything from tiny used bookshops and sellers on our platform up to companies as big as Target. We have over a million sellers.
Then there's the third business, which is the most nascent of these, our developer-facing business. We have over 200,000 developers in that business now. The first two businesses are already financially meaningful businesses for Amazon.com, and we believe the third one can be too.
I don't know. But it's small and growing rapidly. What we've historically seen is that the seeds we plant can take anywhere from three, five, seven years.
Still, these are very different kinds of customers. Is there a uniting vision for all of this?
The uniting vision is that the philosophy is the same: customer-centricity, starting from the customer and working backwards, doing innovative things—the sort of cultural biases inside Amazon. Some of these things, we don't even claim that they're the only right way to do business.
There are lots of examples of close following being an excellent strategy. It just happens not to be us. In a fast-moving area like the Internet, that kind of close following doesn't work as well. We're willing to go down a bunch of dark passageways, and occasionally we find something that really works.
But is there a little more specific vision of where you want to go?
We're trying to leverage an existing asset, skill, or competency—something we think we're really good at. So while these businesses are different, they aid each other. We cannot operate our consumer business without these pieces of Web-scale infrastructure.
Can you give me an example of how you extended some Amazon skill to become a service others can use?
Mechanical Turk was invented for Amazon because we needed to do things like detecting items in the catalog that look like they might be duplicates of one another. You can create computer algorithms that can get you 80% of the way there, 90% of the way there.
But to get that last 10%, it effectively requires artificial intelligence, and that's too hard of a problem to solve. So we decided instead to have artificial artificial intelligence and to figure out how to let a developer call on a human being for help without having to set up a lot of infrastructure.
The really simple way to think about this is, let's say you wanted to let people openly post photos on your Web site, but you didn't want pictures of naked people. Well, it's really hard to program a computer to tell, does this photo have a naked person in it or not? But it takes a human being less than a second to make that determination. The phrase that people are starting to use is "software as a service." Think of Mechanical Turk as "humans as a service."
So you're saying that with services like Mechanical Turk, which seem so far from Amazon's core business, you're actually tapping into what you're already doing for yourself?
In the case of these Web services like EC2 (Elastic Compute Cloud) and S3 (Simple Storage Service) and Mechanical Turk, and so on, we're especially well-positioned to do those things because we had already been doing them. It's a part of the business that historically we had worked very hard to keep invisible from consumers. Consumers don't care about the back end, they just want it to work. The fact of the matter, of course, is that a huge fraction of our effort has always gone into the back end.
In fact, that's one of the enduring criticisms of Amazon.com—how much you've spent on all that.
One of the reasons we have delivered well for our consumer customers is because of that. If you go back to the 1998-99 kind of bubble years, there were a lot of e-commerce companies that set up, and they put all of their energy into the front end. We had done extensive research and found out that people actually want to receive their products.
So that energy and attention that you have to put into the back end is kind of the price of admission. You have to do that stuff well. But it's not really differentiating. So we're already doing it, we have to be doing it at this scale that supports a roughly $10 billion business. And the way you do those things is figuring out how to do those things efficiently, inexpensively, with high availability, is very, very hard. And having done all that, it makes sense to open it.
What about your business aimed at helping other sellers—same deal?
The seller business is the same way. These things, while they're very different businesses, aid each other. They're allies. And the seller business, for example, is of great aid to the consumer-facing business because it, for example, makes really obscure products available to the customers of that consumer-facing business.
So an unusual out-of-print book for which there are only used copies available, we would never be able to offer that in our retail business if we didn't have this big network of sellers who are helping that consumer-facing business by bringing selection to the table.
Likewise, the retail business of Amazon helps the sellers because it keeps the consumers coming back to Amazon, so that then we're providing customer acquisition and customer flow for the sellers. So the three things are actually tied together in some very important ways.
You've been relatively quiet the last few years. Why are you getting more public now about these services?
There's an evangelism effort that has to occur, because this is a new way of doing things. If you look at things like Mechanical Turk and EC2 and S3, these are very, very powerful tools. A lot of people see this right away. In the blogosphere, you can kind of sense the excitement that people have when they look at these things. But it's a different way of doing back-end infrastructure and building these Web-scale applications.
Some people wonder what this says about the retail business. Maybe it doesn't have the promise it originally had, so now you're going into this other stuff.
The answer to that is no. The retail business is growing rapidly, in multiple geographies, in multiple categories. We keep adding categories, we keep innovating in and around it with things like Amazon Prime [a two-day-shipping membership program]. The adoption rates on Amazon Prime are going extremely well. So the retail is stronger than it has ever been.
The main objection I still hear is the ongoing investment the retail side seems to take.
No, if you look at our business, if you compare our inventory turns and capital efficiency with traditional retailing, all those metrics are much better. The return on invested capital in our business is high, and can ultimately be even higher. So we love the economics of our retail business. From a business-model point of view and a capital-velocity point of view, it's a terrific business.
Is it frustrating that so many analysts and apparently investors out there don't believe that?
No, we're very comfortable being misunderstood. We've had lots of practice. Back in 2001 or maybe 2000, the predictions were that Amazon was supposed to be bankrupt by a certain month. It didn't make any sense to us.
If you want to make seven-year bets, if you want to be able to plant seeds that can grow for seven years, you have to be willing to have a thick skin. I honestly think that many of the investors that invest in Amazon really like our approach and our strategy. We won't be right all the time when we make these investments. But we hope this longer-term approach, in and of itself, is a competitive advantage in a world where many companies either don't have the opportunity to do that or are unwilling to do that.
By providing services such as EC2, aren't you enabling many more potential competitors by giving them these services to get a company up and running much faster? And even if they don't compete with Amazon, you could be losing the opportunity to hire great engineers and programmers who are doing their own company rather than joining Amazon.com.
I think you're right. But my point of view is if something is a good idea, it's going to happen. And so, you can fight it, or you can do it. We're just exposing the guts of Amazon. And people are going to take those things and they're going to do stuff that surprises us with those pieces that get pulled out and exposed. We're excited about that, and we think we can make money doing that.
By the same token, enabling people to much more easily create new companies also creates more competitors for your competitors. Did you think about that?
Absolutely! The fundamental point is, it's going to happen, and if it's going to happen, you might as well participate in it.
You're also expanding your distribution services to smaller businesses, right? Perhaps to thousands of eBay (EBAY) sellers?
One of the things I also think is extremely exciting, just released in beta, is Fulfillment by Amazon. We have this big network of fulfillment centers around the world, 10 million-plus square feet, and what Fulfillment by Amazon does is allow third parties to think of our fulfillment network as a giant computer peripheral.
They can make Web services calls to Fulfillment by Amazon to notify us that we're about to receive inventory from them. Then we receive and stow that inventory. And then they can make other Web services calls to tell us to pick certain items of inventory and ship those to particular addresses.
One of our booksellers who has been in our marketplace program for a long time is so excited about Fulfillment by Amazon that he said, "I'm going to sell my house, travel around the world permanently, buying books."
Does this also help your fulfillment efficiency, by filling up warehouses that may not be at capacity most of the time?
Absolutely. Let's say somebody buys two books in the same shopping session—they buy one from a third party and one from us. The way that would work today is you, the customer, would get two boxes, because the seller is shipping it—and that's two transportation charges.
If that seller keeps their inventory at Amazon, then we can use our fulfillment infrastructure to sort those two items together, get them into one box, and then it's one transportation charge. So there's a systemwide cost savings there, for the consumer, the third-party seller, and us.
Social sites such as MySpace and YouTube are becoming the places for people to congregate and create things. How can Amazon keep the attention of potential buyers who may be going elsewhere to hang out and even buy things?
If you're thinking about community-generated content, it's something we've always done. Listmania [customer lists of related products they like], customer reviews, wish lists. [We] may be inspired by these techniques, but maybe social networking is always a modest piece of what we do.
Everyone wanted to be a B-to-C company for a while, then everyone wanted to be a B-to-B company, and then everyone wanted to be a P-to-P company. You know, those things come and go. I guarantee that five years from now, no one will want to be a social-networking company.
One last thing: What's with all the Darth Vader stuff on your Amazon Wish List?
I have three boys. Star Wars is very big in our house. I do love being a dad. I find plenty of time for it. It has cut into my TV viewing. Life is full.