Andy Jassy's 5 Boldest Statements At AWS re:Invent 2018

Bold Moves

Amazon Web Services CEO Andy Jassy shared his insight into a number of topics in a wide-ranging Q&A session with reporters at AWS re:Invent 2018. He gave more detail into two of the biggest product announcements from the show—AWS Outposts, which offers integrated data center hardware running either VMware or native AWS environments; and Glacier Deep Archive, an Amazon S3 storage class that offers secure, durable object storage for long-term data retention and digital preservation, eliminating the need for on-premises tape libraries. He also touched on machine learning and the ethics of the technology as well as the highly anticipated HQ2 decision, in which the company just revealed two new locations in New York and Arlington, Va.

Here are five of Jassy’s boldest statements from the Q&A session.

AWS Outposts

We have a number of enterprises that said, ‘We love the fact that you've made it easy to run on VMware software and tools that we were using to run our on-premises infrastructure on top of AWS. People love that VMmare cloud-native offering and it has lot of momentum.’ But we also have a number of customers that say, ‘Well, some of my applications can't easily move or can't move for the foreseeable future because they have latency requirements.’ They want the ability to use the same AWS services—compute and storage—on-premises but in a very consistent way, and that is why we built Outposts. It's based on customer demand.

We are going to extend VMware Cloud on AWS on Outposts, and then we have customers that want to run the exact same control plane that they use in AWS with AWS native.

We use a lot of different hardware today in our data centers and Availability Zones and we can change out that hardware whenever we want. In the beginning of AWS, we primarily used Dell’s and HPE’s, and over time, we did more of our own designing of our chips and hardware. We are always open to whatever hardware at the right price point because we try to pass on whatever cost savings we have to customers in the form of lower prices so they can build as many apps as possible.

Glacier Deep Archive Advantage

The reasons I think people will use Glacier Deep Archive very substantially is if you’re managing your own tape libraries on-premises, it's really hard to do. They break and degrade all the time, it takes a lot of effort, and it’s a lot of work that people don’t like doing—it doesn't really differentiate their business in any meaningful way. If you can give that to someone that can do it more cost-effectively—it’s one-tenth of one cent per month—that is a very attractive value [proposition].

What has changed over the last several years is you see this incredible linear acceleration that companies are wanting to make use of data like they never have before. A lot of this data isn’t accessed very often, but even if you don’t access it often, there is a lot of learning in that historic data that you want to reside where all the rest of your data is so you can run analytics and machine learning on that data. Glacier Deep Archive is going to allow you to not have to manage a tape library at a lower price point than you can do on-premises and where it can reside with the rest of your data, which I think is really compelling.

AWS’ Machine-Learning Strategy

We hardly have a conversation with companies without machine learning coming up. We are entering this golden age of what is going to be possible, and in five or 10 years' time, I think virtually every application will have AI and machine learning infused into it.

What really drives what we build is what our customers tell us matters. The problem with machine learning today is even though each year the incremental process year over year is astounding, it's still pretty early for most companies in terms of knowing what they want to do with machine learning and finding people that know how to build these models and train them and tune them and deploy them, and that’s why you see us spending so much time and so much energy and investment in the [machine-learning] and AI space. Just in the last year, we've released 200 significant services and features, and that's just because we know there is a real thirst and hunger from builders to be able to do it more easily, and that’s just what we want to enable them to do.

The Ethics Around Machine Learning

If you look at the advent of machine learning and the potential for the problems that it can solve, it's huge—like the ability to help fight human trafficking, and education and security services. There is a huge amount of good happening in the world based on the use of machine-learning services. Even though we haven't had a reported abuse case of our services, we are very aware that people will be able to do things with these services, much like they can do with any technology, that could do harm in the world. I think you always have to think about what's the right way to have people responsibly use it. In the last two to three years, think of all of the evil things people have done with servers and computers, and yet, we would live in a very different world without them.

I think the answer lies in a combination of things. First, I think the algorithms that different companies use have to constantly be benchmarked so they are as accurate as possible, and then it has to be clear how you recommend people using those services. For instance, with things like facial recognition, if you are using it for something like matching celebrity photos, then you can have the confidence level be about 80 percent. But if you're using facial recognition to do law enforcement or impact people's civil liberties, you need to have very high confidence levels at least 99 percent thresholds before use in those case. And then even then, it has to be a responsible decision by a human being.

The HQ2 Decision

It's true across Amazon in general that we've had a lot of people across all locales for a while. As much as we have built a great team in Seattle, you can't convince everyone to move to Seattle. We need talent and builders all over the world, so just at AWS alone we have a lot of people in Seattle, Boston, Virginia, New York, D.C., Dublin and Vancouver—we have people all over the world. When you think about HQ2, the priority of the company was to pick locales where we thought had access to the greatest amount of talent as possible. And while we are competing for talent in the marketplace, we wanted to make sure we were situated in places we could convince people to move to.

Because AWS continues to grow at such a rapid rate, I think it’s a fair bet we'll have a lot of builders in those two new HQs and I expect we will continue to have a lot of people in a lot of locales because we have an insatiable appetite for talent right now.