Just about everything you interact with digitally is enabled by the cloud. Whether you’re doom-scrolling on Instagram, binge-watching on Netflix, ride-hailing on Uber, or downloading super-cool cloud podcasts (hint, hint) on Spotify, you’re using the cloud.
But most people don’t have any idea what the cloud is, where it came from, or what we, as a species, spend on it. Episode One of “Cloud Atlas” goes back to the birthplace of the cloud: a little digital bookstore called Amazon.com.
Stream Episode One:
Cloud Atlas is brought to you by CloudZero, the cost intelligence platform that offers advanced visibility and optimization for your entire cloud environment. Eliminate wasteful spending, ship efficient code, and innovate profitably — all in one platform.
The only reason you can hear me right now is the cloud.
In fact, you may not realize it, but you’re a heavy cloud user. And not just because you sync your vacation photos with iCloud or share presentations in Google Drive. Every time you stream the new Beyonce album on Spotify, every time you rewatch The Office on Peacock, every time you settle an argument with a Google search or watch a how-to video on YouTube or have Yelp find you the actual best bagel in NYC, you’re using the cloud. You just don’t know it.
And you’re not alone. Every smartphone user uses the cloud on a daily—if not hourly, if not minute-to-minute—basis. Of the nearly 8 billion people in the world, 6 and a half billion have smartphones. Biologically speaking, we may be a terrestrial species, but in terms of how we live, work, learn, and play, we’re really a cloud-native species.
But, outside of the tech community, people don’t really know that. Now, at this point, most people get that the cloud is not literally a big cloud of data floating somewhere above us. But when I asked my friends and family to tell me what they thought the cloud was, it fell somewhere between datacenter and storage service. And that’s not entirely wrong, but it doesn’t begin to scratch the surface of what the cloud is or what its impact has been.
The more I asked around, the more I saw this incredible phenomenon: Our lives are defined by the cloud, but we ourselves can’t define what the cloud is, or does.
Why don’t we understand it? Well, partly because it’s extremely difficult to understand. But it’s also because, if you weren’t paying attention during the development of the cloud, a lot has happened that no one bothered to tell you about. Mostly, it was things happening behind the scenes at a digital bookstore that was trying to become the biggest company on Earth. Software engineers messing with the way that digital machines communicate — and in the process, stumbling into a technology that would completely alter the fabric of life. Without the cloud, there’s no Uber, no Spotify, no YouTube, and certainly no Netflix and chill.
There’s an incredible gulf between how much people use and spend on the cloud, and how much people know about it. This podcast — Cloud Atlas — is designed to fill in that gap with knowledge. I’m going to show you how this thing that many of us either don’t know exists or don’t really understand is powering life as we know it. I’m going to tell you the story of the cloud.
And I’m going to have some help. Help from the people who built it —
Allan Vermeulen: My name is Alan Vermeulen, and I worked for Amazon from 1999 through 2021, and I kind of had two roles. For the first few years I was part of engineering management and helped start the AWS business, including running the engineering team there at the outset.
Help from people whose ideas inspired it —
Matt Round: So my name is Matt Round. I joined Amazon as a software engineer, and then I managed a small team of software engineers, and then I directed Amazon’s Personalization department, and then I was responsible for setting up their software operation in Scotland.
Help from major cloud investors —
Michael Skok: My name is Michael Skok and I’m one of the founders and partners at Underscore VC.
And, of course, help from people who have built businesses in the cloud —
Joe Kinsella: I’m Joe Kinsella. I’m the founder and CTO of CloudHealth Technologies and an overall cloud evangelist.
I’ll go from prehistoric times — that is, what people did before the cloud existed — all the way up to the present day. I’ll explore how we went from monolith to microservice architecture, from on-premise to distributed data centers, and ultimately, from Blockbuster to Netflix. Whether you work in tech or have only a cloudy understanding of all things tech, the story is a fascinating one.
In this first episode, I’m going to break down the perfect storm of business pressure, growing pains, and technological limitations that made the cloud necessary.
I’m Dustin Lowman, and this is Cloud Atlas.
This is probably a good time to introduce myself. I’m an English major and musician who finds himself working for a cloud technology company. I have no engineering experience whatsoever — I remember trying to teach myself to code back in 2015 and quitting after precisely two and a half hours.
Until recently, the extent of my cloud knowledge was understanding that I stored my poems and half-finished novels in Dropbox. And let me tell you, this stuff is not easy to understand. Researching the cloud is a little bit like reading Ulysses: You spend a lot of time rereading passages; it really helps to have experts walk you through it; and a lot of the time, it’s easiest to understand things in terms of metaphors.
At last, that English degree is paying off.
Another reason the average person doesn’t know much about it is because we’re not paying for it — or at least, not directly. Though an on-demand streaming service like Netflix would scarcely be possible without the cloud, the twelve-ish dollars a month we pay feels like the modern equivalent of a movie ticket.
So who is paying for it? And what does it cost?
As I record this voiceover, it’s late in 2022. The public cloud as we know it today is just barely old enough to have a learner’s permit: It was first launched sixteen years ago, in 2006. When it was three, in 2009, businesses collectively spent about $1 billion on cloud services. By the time it was a teenager in 2019, businesses spent about $100 billion on it. And this year, sweet sixteen, businesses are expected to spend a whopping $500 billion on public cloud services.
Just to put that in context, $500 billion is about double the GDP of the Czech Republic. It’s also five times the GDP of Ecuador, and 500 times the GDP of Montenegro. 186 of the world’s 212 countries have GDPs of less than $500 billion, putting the cloud comfortably in the 88th percentile.
Especially in the early days, the story of the cloud is really the story of AWS — short for Amazon Web Services. Yes, that Amazon. The Amazon that Jeff Bezos and his then-wife, Mackenzie Scott, started as an online bookstore in their Seattle garage in 1994. The Amazon whose cardboard boxes litter the entrances of every house and apartment building in America. The Amazon that’s one of five companies on Earth to have a market capitalization of more than $1 trillion.
Most people don’t know that there’s an entire line of business within Amazon that has nothing to do with retail. It originated as an internal engineering project, and all it does is sell web services — i.e., the cloud — to other businesses. Heard of Apple iCloud? That’s actually AWS (plus a sprinkle of Microsoft Azure). AWS owns about a third of the total cloud service market, and in 2021, earned Amazon about $65 billion in revenue.
Allan Vermeulen: I did not have the scale and the scope in my head that this was going to affect how everybody built software. I believe that Jeff [Bezos] did, and so he helped me, kicking and screaming toward, “Yes, it absolutely makes sense to engineer this in such a way that you know that we can scale it to these enormous numbers.”
Way back in the mid-90s, Jeff Bezos was among the first to recognize the potential for e-commerce to overtake brick-and-mortar commerce. And less than a decade later, he and his cohort were the first to recognize similar potential in the cloud. To boot, it was about eight years before they had any serious competition — an eternity in the world of tech innovation.
Part 1: The Everything Store
In one thousand nine hundred and ninety nine, Amazon dot com was just barely more than an online bookstore. But Bezos and company had a much broader vision. They didn’t just want to sell books, they wanted to sell, well, everything.
Allan Vermeulen: Amazon in late ninety nine was in. Get big fast mode, and it was an absolute land grab.
That’s Allan Vermeulen. He’s not particularly fond of titles —
Allan Vermeulen: Yeah, I don’t really care that much about titles.
— but he was about as central to the development of AWS as anybody. He had a bunch of different titles at Amazon, including Chief Technical Officer, but at heart, Allan’s a builder.
Allan Vermeulen: Well, I describe myself — in the early days, I was a simple Canadian farm boy who went to engineering school. I like to actually create things, whether those things are created out of bits and software, or whether those things are created out of boards and lumber. That’s what I like to do, and I like to make it easier for people to build things more easily, which explains most of my career.
Before launching his professional career, he got a Ph.D. in engineering from the University of Waterloo, where…
Allan Vermeulen: What you actually spend your days doing when you’re a graduate student is writing code, and I came to the realization really quickly that writing code is really really hard — much harder than it should be.
The challenge he’s referring to is one that every software engineer was dealing with at the time. Put simply: If you wanted to build a piece of software, you had to start completely from scratch, every single time.
It gets a little technical here, so it’s time for…Metaphor Number One: Building a building.
Think of a software development project like a construction project. The same way there are lots of different kinds of software applications, there are lots of different kinds of buildings. They may fall into common categories — houses, office buildings, apartment complexes, and such — but each one needs to be built to unique specifications.
Let’s think about building a house. Before you actually start building, you need to get your raw materials — stone, wood, plastic, concrete. You also need more specialized parts for plumbing, electricity, irrigation, and nowadays, solar power.
Now, imagine if instead of buying these materials from a retailer like Home Depot, you had to build each one yourself. You had to quarry your own stone, cut down your own trees for lumber, synthesize your own PVC pipe, and build your own solar panels from scratch. It would take a ridiculously long time just to get the raw materials — and it would take you forever to get into the real project — building the home.
Allan Vermeulen: And the challenge at the time was, you were given a computer and the computer could run a language and that was it. You weren’t given any sort of components or any tools to work with. So if you wanted any of those components, you had to build them yourself. So as a result, people just spent enormous amount of time writing the same basic building blocks over and over again.
That’s why it took Allan so long to write code in grad school. And in 1999, Amazon had an enormous amount of code to write.
By 1999, Amazon had already made a pretty substantial mark. It had gone public two years before, it was valued at about $25 billion, and it was starting to make the transition from “digital bookstore” to “everything store.” It had begun selling things other than books — things like clothing. But selling clothing, it turned out, was very complicated.
Allan Vermeulen: The problems with the variety of books are absolutely nothing compared to the problems of, you know, buying a sweater that’s a slightly different size or a slightly different color.
Remember, there was a time before we were sure it made sense to buy clothes online, instead of … at The Gap. If Amazon wanted to give customers the same level of flexibility that they had in The Gap, their software engineers had to code features for every element of that flexibility. They had to build features for size, for color, for fit, and for whatever other customizable features the seller offered.
Now, replicate that process for the waist size and leg length for pants. Now, replicate that process for whatever features it takes to sell a gray blender, a blue blender with extra cups, a red blender with extra cups and an extra blade…and every other permutation of every other product that a human being might one day wake up and want.
That’s not all. Consider this: If Amazon wanted people to prefer shopping on Amazon to shopping in person, they had to exceed the flexibility and optionality of shopping in-person. Their solution there? Personalization.
Matt Round: So the the personalization department at Amazon was responsible for everything in the website, and in the wider kind of Amazon ecosystem that tailored itself specifically to who you were as a customer. So we used to think about trying to build a store for every customer. I guess it’s one of the distinctive advantages of being an online operation: For a physical store, everyone has to walk into the same shop. You can’t give different people different shops. But a digital store, you can reconfigure the store entirely for every customer.
That’s Matt Round, a software engineer who directed Amazon’s personalization team. Imagine if, every time you walked into The Gap, all the shirts and pants and sweaters and beanies magically rearranged themselves so that you saw your favorite stuff first. So in addition to building tons and tons for tons and tons of products, Amazon wanted the site to magically rearrange itself depending on which user logged on.
Allan Vermeulen: And so they hired hordes of, you know, fairly young people who just wrote a ton of code that did all kinds of stuff, and it was not carefully engineered. I mean some parts of it were, but but a lot of it was, you know, just thrown together as quickly as possible — very deliberately, because the goal was speed. Not necessarily, you know, perfection.
Yes, Amazon’s paramount goal was speed — or, to use an economics term, “time to market.” Time to market refers to the time it takes a product or feature to go from an idea to something customers are actually using. The longer your time to market, the better chance there is that your competition will beat you to the punch. But the faster your time to market, the better chance you have of defining the competition.
In 1999, Amazon was definitely well-positioned, but there was still no defining e-commerce leader. E-commerce was a somewhat marginal part of the retail economy, accounting for less than 1% of all retail sales that year. But it was growing, fast — one CNN article from November 1999 read, “Few events have captured the imagination of so many so quickly as the notion of selling online, arguably the biggest development to hit retail since the invention of the cash register.”
Amazon had some competition. There were other pure-play digital sites, like Ebay and Buy.com (rest in peace). There were brick-and-mortar stores trying to get into the digital game, like Wal-Mart, The Gap, and Barnes & Noble. A big part of Amazon’s competitive strategy was to make it easy for brick-and-mortar stores to sell on their platform.
Andy Jassy: We started doing these very large partner deals on the e-commerce side that we called “Merchant.com.” We were doing these deals with Target, or Toys ‘r’ Us, or Marks & Spencer, where their whole website was powered by Amazon technology.
That’s Andy Jassy, Amazon’s current CEO. This interview was conducted by Michael Skok, an entrepreneur and venture investor (who just so happens to invest in CloudZero — you’ll hear more from Michael later). In 1999, Jassy was Jeff Bezos’s “shadow” — that was literally his job title. He was always at Bezos’s side, taking notes, observing his meetings, identifying ways for the company to improve.
Andy Jassy: And when we went to deliver the solution to Target, which was really the first big one that used all of our technology that way on their website, it was much harder than any of us thought it would be.
Part of the difficulty stemmed from the coding issue — that they were trying to do things with web construction and software that had never been done before, and if they couldn’t do them, it could threaten their existence as a company. They had an incredible number of features to build and had to reinvent the wheel anytime they wanted to build a new one.
But there was a deeper issue, too — an issue that, when it came to wrapping my head around the cloud, was probably the toughest one to understand.
Business pressure made the cloud necessary — but technological innovation made the cloud possible. In the next episode, I’ll go over the first giant leap in the direction of cloud technology: the death of the software monolith.
Cloud Atlas is written, hosted, and produced by me, Dustin Lowman, with invaluable assistance from Natalie Jones, Greg Barrette, and many others at CloudZero. Credit also to Tim O’Keefe, our sound designer, composer, and associate producer.
Thank you to Allan Vermeulen and Matt Round for their contributions to this episode, to CloudZero for trusting me to turn Cloud Atlas into a reality. And, of course, thank you for listening. Until we meet again, this is Dustin Lowman reminding you to keep your feet on the ground, and your head in the cloud.