As Amazon Web Services Inc. opens its seventh annual re:Invent conference this week in Las Vegas, it’s looking to extend its dominance in the cloud computing juggernaut that’s sweeping across all of information technology.
The Amazon.com Inc. cloud unit remains far out in front of main rivals Microsoft Corp.’s Azure and Google Cloud Platform, at least on the computing, storage and networking services that form the foundation of cloud computing. But the battleground continues to shift from startups and cloud-native unicorns, which early on embraced the cloud as a faster, cheaper and more flexible way to offer their services and get work done, to more mainstream enterprises that need to make the shift from their own data centers to the cloud more gradually without disrupting their own businesses.
In particular, it’s now apparent that, despite the insistence of AWS Chief Executive Andy Jassy (pictured) that building one’s own data centers is “kind of wacky,” many companies for many years to come must continue to run some key applications on-premises because of latency, regulatory or other issues. In turn, AWS executives have adjusted their thinking to this “hybrid” approach with pacts such as its groundbreaking deal with virtual-machine king VMware Inc.
Aiming to appeal to enterprises even more, AWS has steadily offered more, higher-level services such as artificial intelligence, with more likely to come at re:Invent. To learn more about Jassy’s strategy and suss out what’s coming next from the cloud leader, I met Jassy last week for an exclusive interview iin the sports bar he has set up in his Seattle basement, which he calls Helmet Head.
Jassy was his usual straightforward, competitive self, the conversation interrupted at times by the crazy scoring in the Monday Night Football game between the Kansas City Chiefs and the Los Angeles Rams. In this first of a three-part series from the two-hour conversation, Jassy hinted at some surprises coming at re:Invent and discussed lessons AWS has learned over the past 12 years, how artificial intelligence is opening new enterprise doors and where the company aims to expand next. The interview is edited for clarity.
Look for more strategic and competitive insights from Jassy in the second and third parts of the interview in coming days. And check out re:Invent coverage all this week by SiliconANGLE, its market research sister company Wikibon, and its livestreaming studio theCUBE, now in its sixth year covering re:Invent from the show floor.
Q: What can we expect from re:Invent?
First and foremost, I hope that people come away after the four days feeling like they learned a lot about what they can do in the cloud and what others are doing in the cloud, and what’s working well today, which things they should be careful about, as well as a lot of new services and features and what they’re going to be able to take advantage of.
While the focus is being an educational event, I think they should expect that we’ll have a few releases and a few goodies for them there. I think there are going to be some things that surprise people… because they’re totally different segments than we’ve traditionally spent time in or partnered.
Q: How do you try to maintain the edge in re:Invent as it gets bigger and bigger?
A: It’s already meaningfully bigger than most of the other tech conferences. And I think that if you fast forward five years from now, it’s possible it could be 75,000 to 100,000 people. I think we’re going to be in eight venues this year in Vegas. We made the decision a few years ago that we would rather take on the challenge of figuring out how to make the event still feel like us, and unique and quirky, and handle the logistics challenge than prevent people who want to go and learn about the cloud to do so. Especially this stage of cloud computing and people’s learning about it.
Q: What about international?
A: We have 20 Summits around the world, which are kind of mini-re:Invents. In the same way that I was nervous the first several years about having our builders be distracted for a week in prep before that, having to do that multiple times during the year where everybody has to travel, it’s a tradeoff in what you can actually build and deliver for customers. So, so far we’ve decided not to try and replicate re:Invent exactly in all those places and instead do Summits.
Q: Over the past 24 to 36 months at AWS generally, what big lessons have you learned and what would you have done differently?
A: I would’ve made sure that we hired more salespeople and more professional services faster. We launched AWS in 2006 with two salespeople. At this point, we have a very large field [organization], with a lot of account managers and a lot of solutions architects and a lot of professional services folks, but we have just insatiable demand. We have a very large enterprise business that’s growing like crazy, and we can never hire enough good people in those areas to have the coverage that we ultimately want and to help our customers build as much as they want.
Having a sell/service model was really helpful to us. So many companies, including enterprises, started much earlier, and our cost of acquisition, if you look today as well as historically, has been much lower than any other company because it’s so easy to get started using AWS. And then by the time we’re actually in sales cycles, they’re at a different point in the adoption lifecycle than if we had to cold-call and get them interested in the cloud and convince them what it was about.
[Now] you need to have the right types of conversations with the CIO, you need to have the right types of conversations with the CTO, oftentimes with the CEO. There’s a number of product groups and line of business leaders that want to be involved with it. More and more, the chief digital officer is an important decision maker.
Needle movers, door openers and roadblocks
Q: What have been the big needle movers for your business?
A: There’s some general trends — just people seeing the cloud being secure, people seeing the cloud being operationally performant, people seeing more and more enterprises moving all-in to the cloud and having significant success.
But then there are a bunch of other enablers. Containers, where companies are completely rethinking that kind of basic wrapper and unit of compute, people are really reimagining how they do compute in containers and doing it on the cloud. Same thing with the event-driven serverless computing with Lambda.
In the database space, we don’t meet enterprise customers who are not looking to flee Oracle and SQL Server. Having an option for them in Aurora, where they can get the same availability and durability and fault tolerance as the commercial-grade databases, but in a database that is compatible with the open source engines and MySQL and Postgres at a tenth of the cost, is very, very compelling. And that has been our fastest-growing service. Not just the enterprise, but across the business.
Q: Not SageMaker, the machine learning service?
A: SageMaker’s growing incredibly fast. Because it starts on a smaller base, it’s growing at a percentage that’s faster, but we measure everything at the relative evolution of where they are. Aurora continues to be the fastest-growing service at a common point of its evolution, in the history of our business. That has changed a lot of discussions in the enterprise because the database is so important to applications that enterprises are really looking to make significant moves from whom they’ve been using, and that’s opened up a lot of enterprise conversations.
Q: Is AI a door opener as well?
A: Things like machine learning and AI, analytics, edge and IoT, where enterprises are figuring out their migration over several years, there’s this real, palpable excitement about these new workloads and these new capabilities. It isn’t just a migration from what they’ve been doing, but net new opportunities, that has been very impactful in building relationships with enterprises.
Q: Biggest obstacles?
A: In the first six years of AWS, the biggest blocker, by far, was security. Enterprises would say I’m really worried about security, and we’d say, well, OK, what are you specifically worried about in security? And sometimes the answer was, well, I don’t know how I’d keep my data private. We’d remind people you get to pick where in S3 it sits, it doesn’t move unless you move it. And then sometimes they say, well, I’m just worried that I don’t control it anymore.
Now, over the last few years, security is one of the biggest reasons that people choose to adopt cloud.
Q: Is the barrier now how you stand up these new net workloads fast enough?
A: The biggest barrier is inertia. It’s people who have been doing things a certain way for a long period of time, where the people running it aren’t as excited about doing something new, and in some cases, don’t know what it means for themselves personally. You can convince yourself to go slow, that you can do it in the future, there’s plenty of time.
Of course, anybody who studies these big, on-premises footprints and how it works when you’re moving to something like cloud, the longer you stay, the harder it is. Part of our job and our responsibility is to help companies realize how they can move more easily.
Moving up the service stack
Q: What are your top priorities going forward?
A: We’re consistently thinking two to five years forward. We have a lot of things we’re working on right now that we won’t deliver for two to five years. We’ll share some things at re:Invent that are forward-looking. But you can see our thinking about where compute is evolving in what we’ve been investing in, in both containers and Lambda.
There’s going to be a generation of developers that don’t think about instances and don’t think about servers and don’t think about clusters. They’re just going to think about writing scripts and [adding sudden] triggers on what actions they want and have them be connected across every imaginable service in your infrastructure. And it’ll be a totally different programming model. And so we’re thinking a lot about that on the compute side and working very hard there.
Q: Is this where artificial intelligence and machine learning come in?
A: If you look at machine learning today, I think there’s huge progress, both in the offering as well as the usage, in the last year. But it’s still pretty early when you talk to enterprises about machine learning. And they know they have a lot of data, they know there are jewels in that data, they know they need to make it easier for more of their builders to be able to access and do something with that data. That’s why thousands of companies are standardizing on top of Sage Maker.
It completely changes the game for everyday developers and data scientists to be able to build machine learning models and deploy them and get value. But there’s so much that’s coming there. It’s the services further up the stack, where you’ll have companies that don’t want to mess with having to build any models or knowing anything about algorithms. They just want to make API calls and get what this image is or transcribe this audio or translate this into multiple languages.
Q: What you do could be powering new industries. It’s not just IT. So what’s your vision on moving into new industries and providing that service, AWS, to other industries?
A: Every company, including ours, has to think about where it’s going to play versus just serve customers and partners. And I happen to think that we are many, many years away — if ever — from companies not wanting to consume these infrastructure and platform services from folks like us. The big question we ask is how far up to go up the stack.
We try to look at if we went into those areas and were successful, could they be big? Are they being well-served today? Do we have a differentiated approach? And then do we have the confidence to pursue them? There are some areas where we say, yeah, they could be big, but they’re really well-served today. So there’s probably not as big an opportunity or as big a need for us to go in, and that’s why we don’t hear customers asking us.
Q: But customers are asking for services for some areas?
A: There are several areas, as we think about going up the stack, where the answer to all four of those questions is positive, where people want us to go. A good example of that was in the call center and customer service space, where we just had so many companies who said I need a modern, strong machine learning, artificial intelligence customer service and call center type of service, and that’s why we built Connect.
Even though it’s only about a year old and we have a lot of functionality that we’re adding to it every month, it’s been a huge early success with a lot of momentum. So when we find areas like that, I think you can expect us to continue to build a little bit further up the stack. And then, you know, the magic in all that is picking the ones that make sense. We have ideas for some of them that are areas that exist today and that aren’t being well served and the customers want us to go into.
We have other ideas. If you think about what machine learning and AI enable, it, there are altogether brand-new opportunities. Most applications weren’t built with the scale required from the cloud or machine learning or AI capabilities. I think virtually every application area is going to be completely reinvented over time, with those at the core of what people build.
Q: I’m just seeing this entrepreneurial vibe where it’s not so much starting up a startup, it’s more starting up a project. If I were working for an enterprise or I was enabling my team to be creative, to take new territory, the cloud would be perfect, so there would probably be a lot more innovation going on outside of startups.
A: That’s exactly right. I think one of the big things you see from enterprises who have largely committed to and are moving to the cloud is that the level of new ideas they get from all areas and all levels of the organization is astronomically different from what had been the case before.
With the cloud, where you can spin up thousands of servers in minutes and you don’t have to build all the infrastructure software, you get to use the 125-plus services however you want, you can get from idea to implementation in several orders of magnitude faster, just the amount of ideas that companies find all over the company is very different.
The world for nontechnical people is very difficult to get anything done technically. And we have loads of people inside Amazon who have projects they need where they need a little bit of a developer’s time, but they’re just too minor to get resources and they just never get done. There’s opportunities over time, I think, to help nontechnical folks as well.