Thank you. Welcome everybody. It's already been a really exciting day. So, we've got an amazing conversation here about the promise and the risk of AI. We've heard a lot in these sessions about the promise of AI. We have a great panel here to discuss the risks and you have two very different businesses, but you both handle the risk of artificial intelligence in different ways. Haras, I want to start with you. Booz Allen Hamilton is the largest provider of AI and cyber security to the US government. I know what that means, but I know you love to answer the question, what is it exactly that you guys do?
Well, it's a long answer, so let me simplify it. Think critical missions and critical areas, national security, healthcare, critical infrastructure, and then think AI, cyber, 6G, quantum. We work at those intersections. We build technology that can be deployed into those missions, which is actually pretty different from deploying them into more traditional private sector missions. So we leverage what's out there from the best companies in the world. We partner with a number of them and then we build on top of that. We add it to our technology stack and we deliver for the American people.
From the seabed to space, you operate pretty much. You say that the world is not ready for AI cyber security risks. What are you seeing out there in that broad mission that you have that concerns you?
So 2026 is going to be a highly complicated year at the intersection of cyber and AI because AI as an attack vector, both as its own attack surface and also what it can do to people's networks and infrastructure, is further ahead than what the defense side of it looks like. We tested a tool in our lab that can actually crack a network in under 10 minutes. If you think about it, it takes two weeks. The CIS standard for patching is two weeks. We're talking minutes. And then we have built a set of agentic tools to defend against that. But the agentic tools to defend are not as far along. And the attacker has an advantage right now because if you're a nation state that doesn't follow international law, and there's plenty of those, or if you're simply a criminal enterprise, you can put something out there and if it causes harm that you didn't intend, I guess you don't care. If you're on the defense side, you do care. You're not going to put something on your network until you're absolutely sure that it works. And a lot of these tools are too new.
So, you see AI as essentially a force multiplier for bad actors in the cyber world.
Over the long run, AI, I'm a techno-optimist and I actually believe that AI is going to transform the world and our national security for the better and we're beginning to see that too. But at this point the cyber risk is quite significant and we're in a race with China and a race we simply cannot afford to lose.
Let me bring you in here Greg because you handle AI from a different... you are a risk company and a people company but a risk company, risk management for the AI data centers. Tell me about your business.
Well, listen, step back a little bit. Christine, it's great to be up here with you. Looking forward to the conversation. And listen, very much appreciate your comments on the cyber front. Absolutely true. From our standpoint, we see massive opportunity here. The opportunity in AI, as Haras was saying, is phenomenal. But it's got its challenges. First among them is the cyber opportunity or risk. If you want to think about it, by the way, just for reference, Aon, if you don't know Aon, well, we match capital with risk to reduce volatility in the world. And we do that in every way you can imagine it and physical assets with people, etc. So that's what we do. We're matching capital with risk to reduce volatility. In the case of AI, we have an opportunity around sort of the infrastructure that's being built just to support it, which is not just substantial, it's the most significant we've ever seen in the history of the world. And that brings with it a whole series of challenges. The construction process that's going on right now has never been seen before. The risk challenges against that are astronomical. The cyber pieces that might come out of it, just for reference, cyber market from an insurance side's about a $15 billion market against a potential trillion dollar loss exposure. So, the markets don't fully understand it, which is why capital doesn't come. Capital doesn't come unless it thinks it can earn a return. It doesn't understand risk, it doesn't come. That's why it's a $15 billion market. From our standpoint, we're looking at what's out there and essentially saying, 'Look, this isn't just about the hyperscalers. This is the scalers, the builders, the money that's coming in, and the banks.' That ecosystem is sort of what we are looking at on an integrated basis. If we can help those get built and operate, remember operate when business interruptions could be measured in millions of dollars a minute. If we can help operate with less volatility, that's value creation. So from Aon's standpoint, we see tremendous opportunity with accompanying risk. But tremendous opportunity from a standpoint of looking at it from an integrated risk perspective. If we can accomplish that, we can do some things that would be quite remarkable. And that's how we look at it.
I'm interested that citing these data centers must be just a phenomenal undertaking. There's risk in that and there's also political risk also in the sighting of these data centers. How do you factor all that in?
Well, listen, from our standpoint, this is kind of what we do. We've been looking at all aspects of risk from an analytic standpoint really for decades. That database really gives you a view on where you build these things matter massively. By the way it matters from the standpoint of just physical risk, catastrophic risk, you know, wind blows, earth shakes, all these things, wildfires, all these things matter by the way at a zip code level literally down to sort of siting around the world. How you build them matters. Not just what's inside but the box and all the things that come with it. All the things that surround it and the community, the sentiment, all the pieces, all those things matter. So all those things have to be factored in and then understand we are talking about an ecosystem I described before, the scalers, the builders, the money, the banks and wrapped around energy and wrapped around water, all the things that come with it. And if one can actually reduce volatility in the context, you're actually creating value. Who gets the value? Scalers want the value. Builders want the value, money wants the value, banks want the value. How do you think about that? These are all the things we do as you sort of think about how this is coming together. So, it really is an integrated risk view that's really, last point I'd make on this, has to be done from the beginning. If you do this part way, you've lost the plot. You've got to do it from the beginning to get the value out of it.
I know that you guys in your workforce for years and years have been using some of these tools. But in terms of where we are with the data center buildout, this is all still new. There's not really a corollary is there in history to this kind of investment.
There's no corollary whatsoever. I mean from the standpoint we're talking trillions in a period of time we've never seen before. That's going to stress every aspect, the building piece, the operating piece, the insurance marketplace that we live in and we're at the beginning of the beginning. Again there are thousands of data centers in the world. We probably do 40 plus percent of the data centers around the world, maybe more than that. But this is different. This is at a scale and a pace that's fundamentally different. And we've got to get the integrated risk view right or we actually run a tremendous amount of risk of the operating characteristics that are going on after they're built. So this is the overall piece and we've loved this gig from the beginning. We put the first NVIDIA chip into a pathwise application of Aon in 2009, right? So, we've been seeing this from an opportunity standpoint forever, but this period of time is different.
I want to talk to you a little bit about China. You mentioned China and you service clients all over the world, Greg, at Aon. The United States government is your biggest customer and you are unapologetically pro-American about making sure the US wins the AI race. Is the US winning the AI race? And what happens if it loses to China?
Look, I just want to grab the opportunity theme because I think we have to have the risk conversation. We have to have the two horse race conversation but we have to have it in the context of the opportunity that's in front of us. The opportunity that's in front of us is for every company to transform itself into a better, more successful company. The things that we're doing with AI we're identifying now were unimaginable five years ago. And it's opening entire new markets and entire new doors. The work that Greg and his companies do to enable all of that is absolutely central which takes me to the point of the race between the US and China. The most critical question is how do you measure it and I think at the end of the day all of these new technologies get measured by the rate of adoption. If you look at the input side of it the US is clearly still ahead. Our models are stronger and better. The gap, I get a report every month about what our models are doing in the US and what models are doing in China. That gap is closing but our models are still ahead. We have a clear advantage in chips because we have an advantage in lithography and that is going to be an enduring advantage at least for quite some time. It's a good debate about who has an energy advantage and who doesn't. It's probably cheaper to build in China than it is to build in the US and that creates an advantage there. But at the end of the day, this is going to be a race for adoption. And the thing that I worry about is China and the US have taken two different paths in terms of this technology curve. The US companies, our frontier labs, are trying to build the strongest, most powerful models. Trying to get to AGI first if you will. China's focused like they always do in technology on delivering the cheaper models, the models that actually cost less to operate. The token values are now six to one between a US frontier model and a Chinese model and so that is going to drive adoption around the world and that is not a race we as America can afford to lose. So I'm super excited when I see us as a country come together into a national cyber strategy, come together as consortia of companies that don't naturally fit together if you don't think about it differently to actually advance American industry in this area. The biggest advantage that we have in this race is not the chips, it's not the data centers, because those are all going to be obsolete within a decade. The biggest advantage that we have is the workforce, the quality of our talent and ultimately the strength of the American innovation engine and economic system which I think is the way which has led us to win races so far and it's going to help us win this one too.
What about government oversight and regulation?
