Jeremy Legg is two months into his role as CTO of AT&T, and he has been tasked with a big mandate: transforming the company into a software-driven business, with 5G and fiber as core growth areas.
This isn’t Legg’s first CTO gig, just his biggest one. He’s an entertainment biz guy who’s now at the center of the much bigger, albeit less glamorous, telecom business. Prior to joining AT&T in 2020, Legg was the CTO of WarnerMedia, where he was the technical architect behind HBO Max.
He sat down for a conversation with Protocol in June at Toronto’s Collision tech conference to discuss the 5G race, transitioning a legacy company to the cloud and the war for tech talent.
The conversation has been edited for clarity and brevity.
What do you think are the biggest challenges when it comes to growing and expanding 5G and fiber?
Just for rough order of scale, there’s about 130 million homes in the U.S., and we have about 70,000 cell towers. So when you start talking about deploying fiber, you actually have to dig up people’s front lawns: We’re deploying in the neighborhood of 3.5 to 4 million homes [a year], so we’re deploying about 10,000 homes a day. That’s an enormous construction effort, from the sides of highways to holes to digging up lawns, so the sheer scale of that is a bit mind-numbing.
What is your team developing over the next 12 months that you’re most excited about in terms of software products for businesses as well as consumers?
When you think about cybersecurity and you think about the challenges that consumers face, as well as businesses and small businesses, the traditional way that security is delivered is endpoint software. Or it’s a firewall that may sit in a local network that tries to block certain traffic. But the bad guys are already in the arena.
When you think about network-level security, particularly the scale of an AT&T, we see most internet traffic. We partner with our competitors around peering points. Why aren’t we stopping the bad guys before they even get in? Why aren’t we preventing malware from ever getting to your phone? We’re in a unique position to prevent bad guys from delivering malware. We’ve got a lot of other things that you can begin to think about: autonomous cars and autonomous driving, for example. You don’t want to be sitting in a car that has really high latency and needs to make a left turn. So in the build-out of edge locations, this is something that’s really important because of how networks are architecting. You want these cores to be as close to the device or car or object or person as possible. That’s not generally how they’re architected today, because they didn’t have to be. They mostly delivered voice or video, and consumers will tolerate a certain amount of latency there, but they won’t tolerate that with an autonomous car.
I'm not interested in being in the computer storage business: Azure does that. But I want to make it really easy for developers to get on our networks.
There’s going to be multitudes of other use cases that arise: manufacturing plants that automate the way robots and machinery work; in health care, being able to deliver an MRI to an iPad. What we’re building is a platform that is much more consumable. You’ve probably been to an AWS portal. Did you ever notice that you can’t really provision network service there? You can reserve a virtual machine, you can reserve storage, but you can’t really provision network services. So as you think about what’s happening across the telecommunications industry, the networks have always been very proprietary and very closed because we spent billions of dollars on spectrum, and we’re like, “No, we bought it.” But what we haven’t done as good of a job [at] is making spectrum accessible to either consumers or businesses. We need to meet developers where they are. I’m not interested in being in the computer storage business: Azure does that. But I want to make it really easy for developers to get on our networks and consume a low-latency API, consume a hyper-local API, whatever it might be.
Every big company is in its cloud transition right now, including AT&T. You’ve mentioned that it's tough from a talent perspective to communicate to folks, “You’ve been working hard, but all this stuff that you’re working on is old, and we want to get rid of it.” How do you communicate that in a way that makes them feel inspired and aligned with your mission?
This has nothing to do with technology. It’s all people. People eat strategy for breakfast if it’s not delivered correctly. I characterize it in the following. The first is that you have to define a mission that is exciting. I don’t think people work for companies so much anymore as it is that they sign up for missions. And we need to sell a mission to new people we bring in, because you have to bring in a certain amount of fresh perspective and thinking to a 100-plus-year-old company.
[AT&T’s] mission is, well, there’s several. You think about what we do with first responders and how we save people’s lives from forest fires, hurricanes. We run and maintain parts of the network that actually make those things possible. Other things that you get them to sign up for is, “How do you actually roll out a set of technologies to change people’s lives?”
Autonomous driving, for my mother, would be a really big deal. She can’t drive anymore. So I know that car won’t really be safe until our network enables it to be powered. You’ve got to make people realize what they’re doing. And the same thing doesn’t resonate with everybody. Everybody’s buttons are a little bit different.
How do you convince them the cloud is key to fulfilling this mission?
You’re super transparent about the costs. You’re running it on-premises? You go, “We’re running the business, right? We may have noble intent, but we also have a P&L. Here’s what it costs to buy a server here.” And then what you also do on the legacy side — this is something I’ve done several times through these cloud rodeos now — is you take people and we make them accountable for the old stuff and the new stuff. One of the big problems is when you create the cool kids and the legacy kids, and the legacy people feel left out. No, no, no, you’re running the physical data centers, and I’m also making you accountable for our cloud.
One of the big problems is when you create the cool kids and the legacy kids, and the legacy people feel left out.
You mentioned the cost aspect of it, but cloud isn’t always cheaper in the short term.
For sure, I don’t declare “digital transformation” and [then] all the old stuff goes away overnight. There’s investments and then there’s a hump that you get over when you literally turn things off. For example, we’ll shut down half of our data centers at the end of this year. Then you’ll start seeing the costs go down. The other part that you do, though, is really invest in skill sets, whether they’re new or existing. We went from relatively few cloud certifications to having more cloud certifications than any Microsoft partner. So you’ve got to get them invested in the future as well.
You’ve also said in the past that AT&T has a history of building everything itself, but that we’re not living in a time where it can do that, and it doesn’t make sense to anymore. So what in your mind, from a tech standpoint, do you feel is better for the company to be outsourcing, and what do you think should continue to be built and managed internally?
There’s not a lot of point, for example, in building custom storefronts. There’s a lot of storefronts, so you don’t have to start from scratch on something like that. You don’t have to start from scratch on inventory management or billing systems. And we historically did that. I have a computer language that Bell Labs invented, and no one knows. So as we begin to migrate toward the newer things that are associated with this, it becomes the consumer experience, the service layer and how you actually plug things into the network. You want to own all the connectors inside your network, you want to own what the consumer experiences, you want to own the intellectual property around your products and not just resell stuff. But I can run all that on an Azure x86 Linux computer. It’s really the software layers that sit on top of the hardware. And in some cases we’ll partner with SaaS providers like Salesforce. We don't need to reinvent Salesforce.
We've done, I would argue, a poor job of articulating the opportunity. We are one of the largest holders of ML and AI patents, and literally nobody knows.
Now, that’s not the way our company’s been wired. This company’s been around forever.
How do you convince employees that outsourcing certain things is not going to harm them?
Well, they either make the transition, or they’ve got to go. You give them every opportunity to make the transition, but not everybody can cross that river. The ones that do cross that river, some of the most valuable people understand how everything works. And they begin to take on a new skill set.
On the talent front, how do you convince talented tech workers to come work for a legacy telecom company like AT&T instead of an exciting new startup in Silicon Valley?
I was a media guy, and I worked at some of those Silicon Valley companies. Let me just give you the fun facts. We probably have one of the largest IoT businesses, and nobody knows. When I go out and talk to people in Silicon Valley, I discuss the kinds of things that you can do when you actually own the network: Their heads spin.
Now that the world’s shifted a bit, with Silicon Valley equity not what it was, there’s a lot of people looking up over the horizon, going, "Wait a minute."
So we’ve done, I would argue, a poor job of articulating the opportunity. We are one of the largest holders of ML and AI patents, and literally nobody knows. So when we go out and start talking to people about the kinds of things that we can build that you can do in our network, people get really excited. Now, look, are we going to be providing equity exits? No. But you always hear about the people that made all the money; you never hear about all the people who didn’t. It’s not going to get better in the near term, so we provide an awful lot of opportunities, and we want to change some of the brand image to put more shine and polish on it.
We have so much raw material. Like, if you’re interested in cybersecurity, we are, I would argue, the foremost cybersecurity company in the world. And people don’t realize it. So we have to get out, articulate to these communities deeply and do a much better job than we have in the past. And you also have to recognize that they may not stay. The average tenure of a developer: If you get three to five years, that’s pretty good. And traditional AT&T, when you come here, you work forever and then retire. You can’t have that mindset.
I hired over 600 people from the outside, and we’ll continue to hire externally at very high rates. And that new blood is very useful. But I will tell you, I have not had a ton of trouble getting really good people, because Amazon’s a hard place to work. And now that the world’s shifted a bit, with Silicon Valley equity not what it was, there’s a lot of people looking up over the horizon, going, “Wait a minute.”