Andy Jassy: No On AWS Spinoff; Yes On Facial Recognition Regs

‘I can’t speak to what the government is thinking or will do, but at the end of the day, we operate in the United States, and we will follow the United States laws,’ Amazon Web Services CEO Andy Jassy said regarding a potential federal antitrust probe requiring a spinoff of AWS. ‘If we were forced to do it, I guess we would have to do it. We don’t spend a lot of time talking about it.’


Amazon Web Services CEO Andy Jassy sees no need to spin off AWS from parent company Inc. unless federal antitrust regulators force it, but welcomes government regulation of facial-recognition technology, he said yesterday at Recode’s Code Conference in Scottsdale, Ariz.

Jassy’s remarks followed a Washington Post report on a Federal Trade Commission and Department of Justice agreement that puts antitrust oversight of Amazon under the auspices of the FTC, and a presidential campaign call from Sen. Elizabeth Warren (D-Mass.) to break up Amazon and other large technology companies.

“I can’t speak to what the government is thinking or will do, but at the end of the day, we operate in the United States, and we will follow the United States laws,” Jassy said at the Code technology conference. “If we were forced to do it, I guess we would have to do it. We don’t spend a lot of time talking about it.”

Sponsored post

When it comes to the digital economy, DOJ antitrust enforcers must determine whether tech companies are growing due to superior price, quality and innovation, or whether some transaction or business practice is, on balance, anti-competitive in purpose and effect, U.S. Assistant Attorney General Makan said during a speech at the Antitrust New Frontiers Conference today in Tel Aviv, Israel.

“The Antitrust Division does not take a myopic view of competition,” Makan said. “Many recent calls for antitrust reform or more radical change are premised on the incorrect notion that antitrust policy is only concerned with keeping prices low. Harm to innovation is also an important dimension of competition that can have far-reaching effects.”

Jassy, meanwhile, doesn’t object to government regulation of facial-recognition technology. Speaking about law enforcement’s use of Amazon Rekognition software and potential civil liberty infringements, Jassy said AWS strongly recommends that its law enforcement customers don’t use results unless they are backed by at least 99 percent confidence levels and then only as “one piece of a human-driven decision like any other piece of evidence.”

“People are looking for those extra sets of protections around the federal government explaining how they want the (facial-recognition) technology to be used and having real ramifications if you misuse it,” Jassy said. “And I wish they would hurry up, because if they don’t, what’s happening is you’re going to have 50 different laws in 50 different states.”

Here’s a deeper look at what Jassy thinks about AWS separating from and possible government regulation of facial-recognition technology in addition to his thoughts on working with Amazon CEO Jeff Bezos, starting AWS, the current competitive landscape of cloud computing and what he views as the most promising technologies. His remarks were taken from an interview with Recode editor-at-large Kara Swisher at the technology summit.

Jassy Speaks About Federal Antitrust Issues And An AWS Spinoff

Jassy: When you’re able to build multiple customer experiences in different business segments that people really respond to, your business is going to get larger. And I think as your business gets larger, there’s going to be more scrutiny, and we expect it. You have to run the business to be comfortable with that, which is the way we try to do it.

Jassy said he doesn’t see a reason for spinning off AWS or any benefits.

Jassy: I would never say never about anything, I just don’t see it. Typically, when companies spin off subsidiaries, it’s either because they want to get that company off their financial statements, or they just can’t afford to fund the new business the way it needs to be funded. Neither of those are the case, so I just don’t see it. I think AWS customers shouldn’t want us to spin it off, because what they really want us focused on is providing highly secure, highly operationally performant services where we’re iterating at the rate where we are now – over 2,000 services and features a year we deliver. They want more capabilities. They don’t want us to be distracted with building new financial systems or ERP (enterprise resource planning) or HR (human resources) systems or doing analyst calls or earnings calls.

There’s a distraction in spinning off an organization. We’d like to stay focused on providing capabilities for customers. I can’t speak to what the government is thinking or will do, but, at the end of the day, we operate in the United States, and we will follow the United States laws. If we were forced to do it, I guess we would have to do it. We don’t spend a lot of time talking about it. We’re very aware of it. We just talk about the fact that while we don’t necessarily know anything imminent, just as you get bigger, you’re going to get more scrutiny. We just have to make sure as a group and as a leadership team that we run the company such that if people want to inspect more closely, we’ll be proud of it. We have been a business through several administrations, and the thing that we have tried to stay focused on is customers and doing what’s right for customers over a long period of time. In any short period of time, there are lots of things that could happen. But if you do things right over a long period of time for customers, usually things work out.

Concerns About The Use Of Facial-Recognition Technology

Jassy: Facial-recognition technology allows you to look at a set of faces and compare them to a data set of other faces…and see if there’s some kind of match. Each result that you get has a confidence prediction level. Some results the system is very confident about and will have a 99-plus percent confidence prediction. Some it will have less confidence about, and it will have a lower prediction number. I think the issue around facial recognition technology is a real one, and I understand why people are concerned about it. We’re concerned about it, too. I really strongly believe that just because technology could be misused, doesn’t mean we should ban it and condemn it. Look at what happened with Sony a few years ago, where they had their email systems hacked into and all their corporate emails exposed. If you banned computers and servers, think about what a different world we would live in today. You could use a knife in a surreptitious way. There are things that you could do that you have to trust people to act responsibly with.

When I look at facial recognition technology and our Rekognition service, in the 2.5 years that we’ve had the service out there, there’s been a lot of societal good done. The organization Thorn is using it to fight human trafficking, and we have organizations that are reuniting missing kids with their parents and (creating) better security solutions or identity solutions or education solutions. In the 2.5 years we’ve had the service out there, we have not had one report of misuse by law enforcement. There was a reporter that did a report on a police organization that used Rekognition, but they didn’t use it inappropriately, and there was no misuse report. They were interested in how they were using it, and they were experimenting with it.

All that said, I actually understand why people are worried about it. We give very clear guidance to any of our customers who are law enforcement organizations who may use the service in such a way where you might impinge civil liberties. We strongly recommend that they don’t use any results that aren’t at least 99 percent confidence levels and then only as once piece of a human-driven decision like any other piece of evidence. And if people violate the terms of service, we suspend them from using not just that technology, but from all of AWS.

But if you want more protection – which I think is totally reasonable – the federal government should regulate it. People are looking for those extra set of protections around the federal government explaining how they want the technology to be used and having real ramifications if you misuse it. And I wish they would hurry up, because if they don’t, what’s happening is you’re going to have 50 different laws in 50 different states. We believe both with pictures as well as video, that people should actually have to disclose they…actually do have some kind of surveillance. Like anything else – whether it’s private-sector companies or our police forces – you have to be held accountable for your actions, and you have to be responsible if you misuse it. We can’t put people in jail. What we can do is give very strong guidance. We also are offering police forces free training so we can help them learn more about how to use the service.

Working With Amazon CEO Jeff Bezos

Jassy: Jeff is so unusual and so talented, and there are so many things that are impressive about the guy. He’s unbelievably inventive. He’s a really big thinker. Teams bring ideas to him, and they seem big, and then Jeff looks around another corner and makes it bigger. He has unbelievably high standards. I worked for 18 months as what we then called a shadow – it’s really like a chief of staff role – and I thought I had pretty high standards before I started that job. I realized my standards weren’t high enough. But probably the thing that’s most impressive about Jeff is he’s an unbelievable learner. The way that he has evolved as a leader and as a manager and in the array and scope of what he works on over the last 15 to 20 years is very unusual.

Jeff, in the earlier days, was involved in virtually everything the company did, every decision the company made. As a books-only and then as a retail-only company, that was easier to do. But anybody who’s had a chance to build a larger business with many more people and many more market segments realizes that you can’t be involved in every meeting and every decision as the company grows. There are some leaders and managers who can make that adjustment, because you have to build different mechanisms to still be able to see what’s happening in the business without having to be in every meeting. Some don’t make that transition. He was really capable of finding a great leadership team, being clear with them about what the objectives were and then letting them go run the business.

Starting AWS

Jassy: I was working for Jeff in this chief of staff role, and we had this idea to explore a business where we built the infrastructure and technology services that together comprised a platform that allowed other companies to build their technology applications on top of. There were a number of things that made us think that the idea might be a good idea, but probably the biggest one for us was that we felt like we were moving slower in our consumer business in delivering software than we wanted. When I went and spoke with a lot of the different product leaders, they said, “Look, I know you guys think these projects should take two to three months end to end, but we’re spending two to three months just on the storage or the compute or the database and the analytics. Nothing we’re building scales beyond our own initiative, and we’re all reinventing the same wheel.” Even though that’s fairly obvious in retrospect, at the time, that was a big realization for us, because Amazon is a very strong technology company. We figured if we had that problem, lots of other companies did, too.

We were working on hiring the team, defining the services and building them for 2.5 years, in between when we had the idea, wrote the vision document and got approval, and when we launched our first service in March of 2006. It was actually quite difficult to hire the team without telling anybody what they would be working on. We were trying just to get to launch without our friends across the lake (Microsoft) knowing about it. At that time, Amazon was not known as a technology provider to companies. We felt like it was really important for us to be first to market to have a chance to be successful. I was hoping we could just get to launch without anybody else knowing and beating us to the market. In my wildest dreams, of the many surprises we had, I never imagined we would have a six-year head start. I don’t know exactly why others didn’t follow.

I think for some of the older-guard technology companies, our model was very disruptive to their existing businesses. I think they kind of wished it away – the Oracles and the IBMs. When you have an existing business that’s working, it’s hard to cannibalize it with a product that has much lower margin. I think that some of the other players probably were distracted with some of the other things they were working on. Then their initial attempt at the business turned out to be the wrong abstraction. It turned out to be a higher abstraction, when builders really wanted the individual building blocks to construct and stitch together however they saw fit.

The Current Cloud Computing Competitive Landscape

Jassy: We have a pretty significant market segment leadership position in this infrastructure cloud computing space. There are a few reasons for it. The first is we just have much more functionality by a large amount than anybody else, and we also are iterating at a faster clip. When you actually look at the details, that gap in functionality is widening. And that turns out to really matter if you’re an enterprise or a government that’s going to move all of their applications to the cloud, or if you want to be able to unleash your builders to build anything they could imagine.

The second thing is we just have a much larger ecosystem of partners around our platform. It’s not just the thousands of systems integrators who build practices on AWS. Most ISVs (independent software vendors) and SaaS (software-as-a-service) providers will adapt their software to work on a technology infrastructure platform. Few will do two, hardly any will do three, and they all start on AWS just because of our leadership position. You get to move to the cloud with a lot more of the software that you want to use.

The third thing that is pretty different is that we’re just at a different operating maturity than these other providers having been at it six years longer. It’s really different running large-scale infrastructure for yourself and your company, where you get to tell everybody the way it’s going to be, than it is for running it for millions of external customers with every imaginal use case all over the world, where they get to use you without any warning. It just forces a different type of operating discipline and rigor. You can see that borne out in the operational performance of the different clouds.

We’re the only ones that disclose our infrastructure cloud platform numbers publicly. It’s a $31 billion revenue run rate growing 41 percent year over year. We don’t spend a lot of time worrying about competitors. But, in my opinion, we’re pretty early in this space right now. In the infrastructure technology space, I don’t think there are going to be 25 winners, because scale really matters. But there’s not going to be just one. The market segments that we address in infrastructure software, hardware and data center services globally is trillions of dollars ultimately. There’s going to be several successful players. I do believe Microsoft will have a business there. They’re building the business, and they are the clear No. 2 player at this point. I think there will be other players who are successful as well. I think (Google is) working at it. I think they’re working at the business, and we’ll have to see over time how big the business ends up being.

In all of our businesses, there are startups that none of us know about today that have the ability to disrupt. If you think technology changes in the last 10 years have been disruptive -- and I think they’ve been unbelievably dynamic -- I think the next 10 years are going to be faster. There are all kinds of new technology that’ll evolve that will give people the chance to build businesses and pursue various segments. I don’t know exactly who they’ll be in our space, but I’m confident there will be.

The Most Promising Technologies Of The Future

Jassy: There are so many that we’re excited about. One of the obvious ones is machine learning and AI (artificial intelligence). Most applications in the fullness of time will be infused with some way with machine learning and artificial intelligence. Often people talk about hybrid infrastructure, which they mean today (as) some on-premises and some in the cloud. I think in 10 years, when people talk about the on-premises part of hybrid, they’re not going to be talking about on-premises infrastructure and servers. All of those are moving to the cloud. The om-premises part is really going to be devices. There’s going to be billions of these devices sitting in our homes, in our factories and oil fields and agricultural fields, in cars and planes and ships – everywhere. Those devices are really going to need to use the cloud to do large-scale data analytics and push the predictions to the edge on the devices themselves. So we’re really excited about the edge. Those sensors, by definition, have disproportionately small CPUs (central processing units) and…so they really need the cloud to help them be effective and make decisions and to analytics. We’re really excited about voice as well. When we first had apps on our phone, it was so neat to hit a few taps, and you had all of this information. When you use great voice apps, tapping on a phone feels so circa 2009. I think that voice apps have the potential to really revolutionize a lot of our customer experiences.