AWS Summit Washington, DC 2025 | Shannon Kellogg, Amazon

Clips
News
More from AWS Summit Washington, DC 2025

Shannon Kellogg

VP of AWS Public Policy - Americas

Amazon

Inside the AI shift in government: 11 perspectives on public sector ops transformation

Public sector transformation is entering a critical phase, driven by the rise of agentic AI and digital labor that doesn’t just automate, it adapts. But the biggest obstacle isn’t the technology. It’s the systems, silos and culture built for a slower, pre-AI world.TheCUBE’s John Furrier discusses public sector transformation.What’s really holding back progress in the public sector isn’t a lack of innovation — it’s the difficulty of adoption. From aging infrastructure to deeply rooted bureaucracies, systemic inertia remains the biggest obstacle. Meanwhile, AI is rapidly reshaping the role of government, raising the stakes for timely transformation, according to theCUBE’s John Furrier

play_circle_outline AWS Expands Data Center Clusters in North Carolina and Pennsylvania with AI and ML Investments

play_circle_outline Strategies for AI Ethics: Expert Recommendations on Legislation and Data Center Location Optimization

play_circle_outline Government support for streamlining permitting and enhancing energy availability for data centers.

play_circle_outline Focus on innovation, guardrails, and federal standard in AI policy and regulation.

play_circle_outline Daily responsibilities include managing generative AI, infrastructure expansion, and energy issues.

Info
Transcript

Shannon Kellogg, Amazon

Shannon Kellogg

VP of AWS Public Policy - Americas Amazon

Shannon Kellogg, vice president of public policy, Americas, at Amazon.com Inc., joins theCUBE’s John Furrier and Scott Hebner during the AWS Summit Washington, DC event to discuss AWS’s infrastructure growth and public policy outlook. The conversation centers on AWS’ $30 billion investment in new data center capacity across North Carolina and Pennsylvania, driven by increasing demand for machine learning and AI.

Kellogg outlines the strategic approach behind these expansions, including efforts to optimize energy use and build partnerships with state ... Read more

explore Keep Exploring

What is the reason behind the 30 billion dollar capital investment announced in North Carolina and Pennsylvania? add

What factors are taken into consideration when deciding the location of data centers, particularly in relation to the increasing importance of machine learning and artificial intelligence? add

What is the discussion surrounding the possible provision in the BBB bill to limit fragmentation of AI regulations across states and how does it relate to supporting large data centers in terms of energy policy? add

What measures does the speaker believe should be in place in relation to generative AI technology? add

What are some of the challenges and considerations related to AI policy and regulation? add

bolt Powered by CUBE AI

Shannon Kellogg, Amazon

search

>> Welcome back everyone to theCube. We are here, live here at the AWS DC Summit, formerly the Public Sector Summit. But as the world's gotten big, it's gone beyond public sector. I'm John Furrier with my co-host and Cube researcher Scott Hebner. We got a great guest here, friend of theCube. Cube alumni going way back now. He runs the show for America's AWS Shannon Kellogg. He runs all the government affairs policy for AWS. Huge job, lots of news. North Carolina, 10 billion, 20 billion, throwing billions around. Shannon, great to see you. Thanks for coming on.

>> Well, it's great to be back and to be with both of you. I love theCube and so I've had that Cube itch for a while and it's great to actually be back on the show.

>> Yeah, great to come on. Well, first thing I want to let you know is that we have theCube data center for sale in case you're looking for more power.

>> All right, let's talk-

>> Throwing billions around. You guys are spending billions of dollars. I mean, obviously everyone been following theCube. Dave Vellante is breaking analysis. Of course, we're tracking the CapEx. Everyone's talking about the hundreds of billions of dollars in CapEx. This past week you guys announced 10 billion in North Carolina, 20 billion in Pennsylvania.

>> That's right.

>> These aren't small things. These are big. I talked to Prasad last week, I just shipped a video this week. There's a systematic design element that Amazon has taken to the market, well documented. It's not just a data center, it's a system. Now with that becomes the work around it, you got to get power. Power is run by sometimes the government or affiliated to the states. That's your job. So take us through the news and then we'll get into some of the things you're working on. But I think the big news is the 10 billion in North Carolina and 20 billion, how does that deal go down? Like, "Hey, we need some CapEx"?

>> Well, I have a lot of very talented colleagues working on both of their deals. We have a world-class economic development organization working with world-class economic developers in each of those locations in North Carolina and Pennsylvania. So there are a lot of people who put a lot of time and effort into these projects, which sometimes can take years to develop. Now in the case of both North Carolina and Pennsylvania, these are new states for us in terms of putting large clusters of data centers and computing power in those locations. And a lot of this is driven by machine learning and artificial intelligence. And so in this era that we're living in, companies like ours have to find a way to expand our infrastructure. And that includes in new areas, new states, new countries where we can put a lot of infrastructure in one place fairly quickly. And so that's why you're getting this level of capital investment, literally 30... Oh boy, I can't even say it.

>> Billion.

>> 30 billion.

>> Billion, yeah, not million. Billion, the big B.

>> 30 billion announced over the last week. And I was in both of those states. I co-hosted basically both of those announcements with colleagues from economic development and infrastructure. And we were in a very rural county in North Carolina where we're developing the infrastructure down there in Richmond County. And there is significant power available from Duke Energy in that location. And so we're very confident that we're going to get at least a gigawatt, access to a gigawatt, that is already available. This is not new generation. And so we're making that 10 billion investment there in North Carolina. And then yesterday I drove up to Pennsylvania to announce with the governor there and also Governor Shapiro and then US Senator McCormick we're both involved in that announcement. And this is a big deal for Pennsylvania. It's a big deal for us. $20 billion investment there. Also in a fairly rural county in Northeast Pennsylvania, we announced this in Berwick. The infrastructure will actually initially be put in two different townships in two different counties, but in that Northeastern area of Pennsylvania. And that will include some proximity to the Susquehanna Nuclear facility that is in that area. And of course that is owned and operated by Telen Energy. And you may recall that we did an announcement with them last year, and so we're putting data centers again in Pennsylvania-

>> Is that nuclear on the table for you too? Is that going to be available for you or is that next step?

>> Yes, we will have access based on the original agreement and that partnership is going to continue to expand.

>> So that was in the deal, the nuclear option?

>> That wasn't in this stage of the deal in terms of what we announced yesterday, we announced the partnership with Telen last year,

>> Okay so-...

>> and then as we move forward, you will see us continue to expand that partnership. So stay tuned.

>> All right.

Scott Hebner

>> Do you generally go out to these regions or the regions coming to you competing for data centers and the economic investment?

>> It's a really good question. So we have a lot of activity across the US right now, and you have governors and states that do proactively call you and you also have energy companies that proactively reach out to you. In the case of North Carolina, the conversation actually started with Duke Energy and they reached out to a colleague of mine who runs economic development for us. And they said, "You know what? We may have a really good site that meets your requirements. You've been talking to us about what you need as you move forward and this site of North Carolina could be really good for you." And so that started the conversation. There's a lot of back and forth with Duke Energy and the Economic Development Authority in Richmond County. And then of course conversations started to happen with the state of North Carolina as well.

>> So my daughter would say, is there a Tar Heels energy or because they Duke Tar Heels on, I had to get that in. Are you Tar Heel fans?

>> By the way, I wore Tar Heel blue at the announcement last week. So your daughter should be happy about that.

>> I love the North Carolina. I love that area. And again, this speaks to the mindset people think data centers has got to be in Virginia, that's the MA east, MA west, these areas that have been access points for connectivity, not necessarily the case. You're starting to see a lot of distributed data center architectures in the US. Are these decisions intentional? You mentioned AI machine learning and GenAI, was that involved in the decision making selection process and/or what is the intentional strategy for these regions?

>> It is intentional and it's intentional in a couple of ways. So first of all, when you look at how AWS and Amazon built its business around Cloud, for the first 15 to 20 years, we really had a very deliberate, actually first 15 years, we had a very deliberate architecture in how we established regions around the world that were close to customers. So we would put data centers in places like Virginia and Dublin and Frankfurt and in Ohio where we had big customer sets and then we put a lot of services in the data center infrastructure in those AWS regions. And you had a really good latency experience as a customer. So if you're a customer in Tokyo, our infrastructure in Japan is your first option to use that infrastructure for a whole range of services. However, over the last several years with machine learning and artificial intelligence really taking off, including GenAI, it is not as necessary to have the infrastructure located as closely to customers to ensure that latency experience. Because what you need is you need large-scale compute. And so you need to be able to develop these large language models and use-

>> Compute, not networking as the optimization focus... And energy, energy and compute.

>> And so you could put these clusters in a lot different places. And then bingo, power access to energy becomes the other big factor because even though we are architecturally able to put clusters in other areas now, we still need to ensure that we have line of sight to energy and power for indefinite period of time.

>> Let's talk politics for a second, not politics in this the way normal people do, but tech politics. You have been very active. You've been a pioneer. I've been following, we've been known each other since 2012, 2013, I think the world's changed, CIA deal, Cloud hits the scene. Now we're in a whole nother level. You start to see a tech savvy governing body, younger legislators, younger leaders. What's the temperature like on the political spectrum around things like the national grids, power windmills? I talked to Prasad, he said your data centers are air-cooled. Like what? How do you design that? That's incredible. So there's innovation happening on your side. Is the innovation moving faster on the government side?

>> Yeah, I would say a couple things. So as industry and players like us are innovating, making the data centers more efficient, innovating the way water is used in the systems that we operate and then redesigning the cooling systems, all that innovation is going on, the chips are becoming more efficient. There's just a ton of innovation going on in the industry. Government realizes that it has to do a few things in order to enable the innovation in the US leadership that people in this country want to see, right? And so there was some continuity between the Biden administration and the Trump administration in some areas around data center development. So one of the concepts that is carried over is how to designate federal lands for data center development and then to use the ability of the federal government to streamline permitting on those federal lands to fast-track projects. And so that's something that has been a focus of both administrations, and this current administration has put the gas on the pedal-

>> It's like a national park for data centers. It's a national service.

>> But it's a great idea in terms of looking around the country and looking at what land could be available, where they could actually use the powers of the federal government to streamline the permitting processes around the use of that land. And then to ensure that it has a power availability nearby. And so it's a great concept and one that both administrations have pursued. This administration is also really expanding what it's trying to do around the availability of energy both in the next few years as well as long-term. And so there's a big focus on in all of the above strategy for energy generation. And then there's also a big focus on what can be done to expand transmission across the country. So energy is a huge focus and that includes nuclear energy, both what can be done to modernize current nuclear infrastructure, and then what can be done to also advance the development of next-generation nuclear, including small modular reactors. That's a big focus of this administration, which there's a lot of innovation going on in that sector. And so I think that that is a very positive sign what this administration is trying to do. And then last but not least, and David Sachs talked about this earlier when he did a fireside here at the summit today, and that was, it's really important as we try to maintain and advance US leadership to not overregulate AI. You have, and he pointed this out, states now are considering just about a thousand different legislative proposals and my team has to focus and work on all of those across the country. So we see it, and there's a lot of attempts to try to regulate the technology. It's very early days when you think about it.

>> Yeah, don't regulate, it won't grow fast.

Scott Hebner

>> You just stole my question.

>> Yes, sir.

Scott Hebner

>> I was going to ask you about the regulatory landscape is getting easier or harder to navigate. And if I understand correctly, in the bill, the so-called BBB, there is a provision in there that's going to try to limit all the fragmentation of the AI regulations across the states. Is that part of the energy policy to support these large data centers?

>> Right, right. So there has been an effort by Congress, certain members of Congress to put a moratorium on state level legislation. And we haven't taken a public position on whether or not that language should be in the bill. But look, the concept of trying to figure out how to advance a federal standard is important, right? Because there should be a federal standard if we're going to have some level of regulation, let's create one federal standard to do that. But in the absence of a federal standard, states are going to legislate and are going to pursue different types of regulation. And so to me, it's a joint responsibility. One, for states to maybe slow down a little bit, take more time to figure out what makes sense and what doesn't make sense to preserve US leadership in AI so as to not over regulate. And by the way, to also enable states to enjoy the benefits of AI development and innovation in the US because as we were talking about earlier, states are attracting in lots of investment from companies like ours to develop AI. So states have a real benefit of creating a regulatory environment that is a reasonable regulatory environment. But congress needs to act too. The moratorium proposes a ten-year moratorium, but should we wait 10 years? I mean, can states-

>> What's your take? You've seen the internet. I mean the internet did a good job. The US did a good job with the web. The policy was solid. Let innovation, keep it open. What's your view? What's the general's common sense view for ai? Let it rain, let chaos rain and reign in the chaos?

>> We've always acknowledged that when you look at generative ai, there needs to be guardrails in place. So we put out ideas. In the past, we were one of the companies that signed up early for the White House commitments during the Biden administration around generative AI. So we believe that there should be guardrails. Now, there's a lot of law that exists today already both domestically and internationally. So let's make sure that we're leveraging current law and enforcing current law before we start developing a whole patchwork of new ones. Right?

>> Shannon, I got to tap your expertise. I know you got the Amazon position, which is cool, but as an individual, take your Amazon hat off, put your Shannon Kellogg hat on. There's a lot of young people on X, on Reddit, they're not experts like you. Where could you point them to look for information on where to get more? I like Stax's argument. I think he's right on this one. Let it go. Don't get too heavy-handed on the regulation. Let's see what develops. Keep an eye on it-

>> But he didn't say no regulation, right?

>> Huh?

>> He didn't say no regulation.

>> Keep an eye on it. Guardrails.

>> There should be some guardrails.

>> Guardrails are fine, but a lot of people are interested in this. It's a huge generational interest. And it may have been in the circles years ago for legal and policy folks, but there's a lot of young people where can they go look? Threads, they can go look at sources. What would you share for folks who want to get informed on how to bone up on AI and the ethics and the legislation policies around it?

>> Yeah, so I mean, first of all, watch your show. I mean, that's obviously the answer.

>> Sounds like we got a sponsor, AWS.

>> But other than watch your show, I think there are lots of different sources of information out there. Lots of people are writing about this subject both in the press, but also you have a lot of industry groups and trade associations. There's a group in town, it's called the Center for AI Competition, which was started out at a trade association and they're kind of like a trade association and a think tank. And they've produced a number of reports and recommendations, collecting information and analysis from a range of stakeholders in their work. And that includes voices in industry, voices in government. And by the way, the industry voices are not just cloud providers or AI developers, it's the whole ecosystem. And so that group has done a really nice job of assembling, and they also have tapped into the energy sector. So there's a lot of different views, but they've put together, I think a lot of really solid recommendations and thoughts on AI policy and regulation. When you look at the data center part of the equation, and there's a lot of activity right now around developing data centers, and there's also a lot of states that are also considering regulations of data centers, localities are considering it too, and putting restrictions on data centers. And so there's a group called the Data Center Coalition, which is a nationwide organization that has lots of really good reports and a lot of neutral analysis that's coming out of that organization as well.

Scott Hebner

>> Yeah, it seems like it's coming from everywhere. I know recently, last couple of years, the judiciary ruling in Delaware where corporate officers can no longer hide behind, "Well, my AI told me to do it," but they're actually accountable for what their AI does and they can be personally accountable. So it's coming from all different angles.

>> I don't know if I like that one, but that sounds like... Well, there's a lot to do, Shannon, I got to ask you final-

>> But there are a lot of various ideas out there. And that's the thing is if there isn't some kind of federal preemptive legislation, you're going to get a thousand flowers blooming there. Well, maybe 10,000, right?

>> No, we're going to do a lot of content. I'm just kidding about the sponsor. But we are going to have a whole show on... Not AI ethics, I don't even know what they call it, but global AI because there's legislation input, there's community input, and now everyone's a lot more interest. It's not like some corner tech lobbying thing. It's transparent, it's wide open.

>> Well, I have a colleague, her name is Nicole Foster, who's our global AI policy and regulatory director. And so she spends most of her day, not all of her day, but most of her day on issues like that globally, including getting into some of the AI safety and ethics issues. So she'd be a great person to interview as well.

>> People will throw a lot of money at this thing from many different not optimized angles as inherent incumbent. Okay, final question. You got a lot going on. You wake up every morning, now you're the head of America's Global Affairs and Policy. So much going on. You have stakeholders in multiple areas. You got an ever-changing markets, growing a lot of threads. What's your day? Is it crisis every day or is it some big meaty things you work on every day?

Scott Hebner

>> Yeah, I mean, I joke that I wake up every morning and there's a flashing neon sign sitting above my bed and it says, "Generative AI equals more infrastructure, equals more energy." And those three buckets make up a lot of my day. AI policy. So not just what happens in terms of spurring the development of AI, but what could be the regulation of AI. Because if you regulate AI as we're talking about in a way, it then actually quashes some of the infrastructure development and economic development, right?So that's a real healthy bucket of issues that we deal with every day. And then infrastructure expansion, we're doing a ton of it. I can tell you yesterday my day was I had to drive four hours from my house to Berwick, Pennsylvania. I landed there and basically started interviews at the moment I landed with press before our announcement. We then met with the governor of Pennsylvania and the US senator from Pennsylvania, a number of other stakeholders. We did the announcement, which was over an hour long. We did some more conversations and meetings and interviews. And then I drove four hours back to DC yesterday. And then I hosted a dinner with the press last night where we talked about AI policy, infrastructure expansion, including our recent announcements and the third leg or the third flashing box, which is energy. So those three issue areas make up a lot of my day every single day.

>> It's certainly eating into your yoga.

>> Yeah, my wife would agree with you.

>> Zen likes to energy more data centers, AI. Shannon, great to see you. Thanks for coming on theCUBE.

>> Great to see you.

>> As always, pleasure and congratulations on all the continued success. A lot more work to do.

>> Thank you.

>> We're looking forward to keeping in touch and of course, covering all the AI greatness.

>> Thank you. Great to see both of you. Really appreciate it.

>> Okay, we're bringing all the action here. A lot of policy, government interactions. This is really where technology's crossing over with money and impact all happening at once. It's happening right now. Of course, we're doing our job here in theCUBE. Thanks for watching.