Geospatial Data: Superpowering AI Infrastructure Planning

A Conversation With Joe Francica

Over the next few years, it’s estimated that nearly a trillion dollars will be spent on data centers, semiconductors, and power grid upgrades. And this is all just to meet the growing computational needs of AI technologies. In fact, industry estimates suggest that global data center capacity could double by 2030, but this growth is far from straightforward.

As AI continues to evolve and the demand for data centers grows, having access to detailed geospatial data will give companies a strategic advantage in making informed, forward-looking decisions.

In This Episode:

Up Next

What Makes Geospatial Data Unique for Expanding Telco Infrastructure?

Joe Francica:

From a land acquisition standpoint, it’s a game of information arbitrage, meaning all things being equal. Those with the data have a first mover advantage. It is then both an analytical tool and a marketing tool for communities and companies looking for investment opportunities.

Maiclaire Bolton Smith:

Welcome back to Core Conversations: A CoreLogic Podcast where we tour the property market to investigate how economics, climate change, governmental policies, and technology affect everyday life. I am your host Maiclaire Bolton Smith, and I’m just as curious as you are about everything that happens in our industry. Artificial intelligence, whether you hate it or love it, is powering change. This technology is changing the way we interact with people, do our jobs and conduct business, but all of this innovation needs power. Literally, AI and gen AI are powered through data centers that demand enormous quantities of power, and those numbers are only going to get bigger. Some industry analysts estimate global data center capacity to double by 2030, but fulfilling this demand is not as easy as just building a data center. Data centers need to be optimally located to ensure efficiency, efficacy, and longevity. The future of AI depends on getting this equation correct. So what do people need to think about? So to dive into this and talk about how companies can tap into geospatial data and prepare for the increasing need for data center capacity, we have Joe Franca, a principal product manager in CoreLogic’s location intelligence group. Joe, welcome to Core Conversations.

JF:

Thanks, Maiclaire, great to be back on Core Conversations.

Erika Stanley:

Before we get too far into this episode, I wanted to remind our listeners that we want to help you keep pace with the property market. To make it easy, we curate the latest insight and analysis for you on our social media where you can find us using the handle at CoreLogic on Facebook and LinkedIn or at CoreLogic Inc on X, formerly known as Twitter and Instagram. But now let’s get back to make Claire and Joe.

MBS:

Alright. Okay. So let’s just start by talking about what’s happening with AI power consumption and how much more data center capacity is going to be required.

JF:

So let’s start with a rather startling number. Goldman Sachs estimates that around a trillion dollars will be spent over the next few years on data centers, semiconductors, grid upgrades, and other artificial intelligence infrastructure. So I think the question then is whether we need all of that additional computational capacity and on top of existing cloud data centers that we’ve already have. So we will be the same type of demands on cloud computing or for search for storage, for advanced computational models, for regenerative AI and applications that leverage these things, call it large language models, or for us in geospatial 3D digital twins that we hear so much about. So these are the unknown factors that hold promise for now and therefore might require a large build out of entirely new AI data centers.

ES:

A digital twin mirrors the real world through GIS and is a virtual representation of reality. It can include physical objects, processes, and relationships, and can be scaled.

MBS:

Are we going to have to completely build new data centers or can we just grow the ones that already exist?

JF:

For me, this is the most exciting question because it requires all of the tools we typically want to leverage from using geospatial data and location intelligence software solutions. This is what those tools will meant to do. It goes to the heart of a most important where question and where will I put a new AI data center and why?

The answer is that we will need all kinds of data at our disposal, such as proximity to power plants, transmission distribution lines with greater capacity substation locations, and the location of broadband fiber backbone and available land, which goes to the heart of what we do to put these data centers and where there needs to be a balance between risks of natural hazards and environmental impact and let’s say a regulatory tax and permitting guidelines that may further impact land availability. So all of these considerations require the ability to assemble myriad data sources to analyze them concurrently.

MBS:

Okay. Well let’s dive into each of those a little bit, Joe, because I think there’s a lot to talk about here. So let’s just start about with the data center, the power consumption and the energy sources. I think that’s a big one. So I guess as a kind of precursor to that is why is AI consuming so much bandwidth, so much energy?

JF:

So the answer goes regarding the energy required for AI, which is related to additional electricity required to run and cool the artificial intelligence chips like those manufactured by NVIDIA and others. And let me just quote you something from the Electric Power Research Institute that says data centers powering advanced AI models could account for up to 9.1% of the US’ overall energy demand by the end of the decade. So by 2030, data center energy requirements in the US could range from 4.6 to 9.1% of total electricity demand generated compared to what we are today at 4%. Now, to your second question, let’s look at where we are today in the adoption of AI applications. So we are less than two years since the launch of chat GPT and new applications like Microsoft copilot, Google Gemini, apple Intelligence. Think of where we were today versus the launch of the Mosaic internet browser in 1994 that made it easy to search for content on the web. That was 30 years ago. Now, in the 10 intervening years from 94 to 2004, there were incredible strides for us in geospatial in utilizing that technology and which led in 2005 that marked the launch of Google Earth, a searchable platform for earth observation. So by 2034, we will likely see the next generation of specialized industry specific LLMs and applications of gen ai.

MBS:

Wow. So that’s such a great timeline to kind of think of where we’ve come in the last 30 years a little bit more even that it really is it because of I guess the eruption of gen AI and these LLMs, the large language models that are driving the demand for so much more capacity now because they are demanding so much more capacity than everything that’s ever happened in the past is done.

JF:

That is what we are hearing, that because of the computational requirements of ai, the cooling, the electrical demands, that’s what we’re hearing. So today we are in an energy transformation period towards electrification, and so these additional energy that’s going to require for new data centers will compete with EVs and home electrification, and therefore there will be a need to generate even more power. And let’s put aside public policy and environmental concerns and let’s just say all existing forms of energy will be needed, both renewable and non-renewable. But during this transition, we are both reducing power as we retire things like coal-fired plants and trying to replace that with other power sources, whether that’s solar or LNG or liquified natural gas and nuclear small modular nuclear or wind. I mean, just in the last few months, the US has added about 12 gigawatts of solar capacity, mostly in Florida, in Texas, and we just added a new nuclear power plant online in Georgia for an additional gigawatt.

MBS:

Wow.

JF:

So how will this impact site selection as we look towards new data centers?

MBS:

Sure. Where the sources are. Yeah, that was actually the next thing I was going to ask you is are we moving in the way of renewable energy sources? And it sounds like there’s a lot of that, but that’s not, it is really tapping into every energy source that may exist.

JF:

It is. And then when you think of the location in the optimal location of data centers, is it proximal to new energy and power sources? So even if we have the electrical capacity, that power needs to be supplied at the right time and at the right place and make certain that the power is reliable, redundant, and resilient.

MBS:

Right. And the reliability is a huge part of it, I’m sure. Okay. You also alluded to availability of broadband. How does that play into the equation too? Because I mean, not everywhere is created equally both within this country and outside of this country as well too. What are some of the things that may be optimal for locations when we think of broadband specifically?

JF:

Well, so we know today there are some fairly large centers where data centers are located. Some of the obvious ones, Washington dc Dallas, the San Francisco Bay area, the not so obvious ones are getting attention because some communities look at data centers in terms of economic development. There are designated data center opportunity zones, and the example is in the DC area, and of course they require broadband and power capacity. So these opportunity zones are closed to both commercial and government centers that are developing AI applications and they require low latency from a fiber infrastructure.

MBS:

Interesting. Okay. Something else that you mentioned, Joe, that I want to dive into is climate resilience. Regular listeners of this podcast will know that we talk about climate change a lot on this podcast. It’s something that’s really a passion of mine. Climate patterns and climate change in particular has to be a big part of this because you’ve alluded to Texas and other places that have built all this new solar Florida. Those are areas that are prone to natural hazards, and we’ve seen increasing patterns and increasing severity, whether it be increasing hail in the Texas area, increasing severity of hurricanes on the east coast. How does climate resilience come into play when looking for optimal locations of data centers?

JF:

So it’s important and it’s one to consider. So as we start to layer in all of these other factors, power, fiber, wherever the capacity relies, we have to consider where there is exposure to natural hazards. And so that’s another critical element that needs to be considered when you’re looking at different sites. So as you consider the natural hazards, you’re going to look at forecasting models, whether you’re in an area where there’s earthquake risk or whether it’s risks from hurricanes or other convective storms, you need to consider that because resilience is an issue that should be in the mix, and we want to add that in.

MBS:

Yeah, how can someone know if the site they’re specifically looking at is safe or potentially more resilient or more hazardous to natural disasters?

JF:

So the use of the hazard models that we at CoreLogic do can mitigate some of that perhaps, but not all of the risk. If you look at population migration, there has been a pattern shifting increases to population, to areas like you said, in Florida and Texas and other southeast coastal communities that are susceptible to both hurricanes and other factors. So as a result, data center development must consider those impacts. And again, as we layer all this data on top of each other, you’re going to see some of this fall out in terms of

MBS:

Sure.

JF:

At the optimal location.

MBS:

Yeah, I guess as we’re layering everything together, the other one that’s important to consider is the proximity to an electrical grid or transmission facility. So can you talk a little bit about that one too on I can imagine getting everything in the optimal place with all of these things on top of each other is not easy, but how does this kind of go into the overall planning of potential construction of new data centers?

JF:

So this is interesting because many economic development directors from rural areas are targeting data center development companies. So they bring jobs now, if they can enhance their position with available broadband and electrical power, they have an advantage. So many communities have already leveraged the rural broadband initiative from the Federal Communications Commission to bring these communities access to broadband. So let me give you an example. Madison County, Mississippi, Amazon Web Services is planning to invest 10 billion to construct two new data centers.

MBS:

Wow.

JF:

Meta is investing in both Kuna, Idaho and Temple, Texas to expand digital infrastructure in that region. So some will consider latency an issue in terms of serving the results from AI computations, but others not So, which makes rural locations very attractive.

ES:

It’s that time, again, grab a cup of coffee or your favorite beverage. We’re going to do the numbers in the housing market. Here’s what you need to know. In July, 2024 home prices increased year over year by 4.3% between June and July of this year. Home prices decreased by 0.1% on a monthly basis. Although July is the 150th consecutive month that the US has seen year over year home price gains, monthly home price growth is starting to slip. Annual forecasts are also showing smaller anticipated gains. Next year, prices will inch up by 2.2%. Much of this sluggishness can be attributed to the high mortgage interest rates that are continuing to challenge the housing market. As buyers remain cautious, sales remain low. However, the highly anticipated rate cuts from the Federal Reserve this fall may help improve consumer purchase sentiment for the housing market. The states with the highest increases year over year were Rhode Island, 10.6%. New Jersey, 9.7%, Connecticut, 8.3%, South Dakota, 8.1% in Illinois, 7.5%, no states posted annual home price declines, and that’s the sip. See you next time.

MBS:

Well, so now that we have all of those different components, Joe, if we layer them all together, how are they equally weighted? Does something have a higher priority than other? How do you take all of this information and determine what is an optimal location? If you were to determine a new place to build a new data center,

JF:

This is what geospatial technology and location intelligence was really built to do, right? It’s the availability to integrate location based data onto a single platform and then both visualize and analyze data to make reasonable decisions about optimal sites. So from a land acquisition standpoint, it’s a game of information arbitrage, meaning all things being equal, those with the data have a first mover advantage. It is then both an analytical tool and a marketing tool for communities and companies looking for investment opportunities.

MBS:

Okay, interesting. Now, I guess the next question I have to ask is are there actually areas in the US that are optimal or is it better to look at other places around the globe? What does the location, intelligence and geospatial information telling us?

JF:

So the couple areas that I mentioned are, well-known the DC area, the San Francisco Bay area. It’s now the chance to look at other areas where there may be those opportunities, and that’s where this power of integrating data comes into play. So remember, we’re at the early innings of ai. Sure, yeah. The top five hyperscalers spent $158 billion on data centers in 2022 led by Amazon and Google and Meta. This is only going to expand. And some of the expectations are that in five years, you’re looking at about a 30% annual growth rate. So there’s going to be competition for locations because they drive jobs, they drive economic development. But there’s also going to be this balance between the need for energy capacity, the need to bring down our carbon footprint and all of the likes of that. We’re in balance with renewable, non-renewable sources, power capacity, offtake capacity, which means even if you had the capacity, the electrical infrastructure, is it going to get to where it needs to be? What’s the distribution capacity? So I’ve got the electricity, now I need to distribute the electricity. And that’s what the communities are going to be looking at and the sites are to see where it is going to be the advantage of where to place these sites.

MBS:

I guess it kind of goes without saying Joe, but as I hear you talk about this demand, it sounds like a massive financial investment. And I think that when you look at this is a lot of money to accommodate this massive amount of data that is needed. It

JF:

Is. And it goes to the fact of how much is required to build out the infrastructure, whether you are somebody like an Amazon that is buying the chips from Nvidia and others that are being used in large language models, or whether you are a data center company that needs land to build these new data centers. And then you’re looking at the areas of electrical power. So whether that’s building new power generation facilities and the infrastructure to supply that power to the data centers, this is what’s adding up to the tremendous investment in the artificial intelligence ecosystem. And that is bringing in all of these different factors from fiber to power to chips and driving this big buzz around ai.

ES:

Before we end this episode, let’s take a break and talk about what’s happening in the world of natural disasters, CoreLogic’s hazard, HQ command, central reports on natural catastrophes and extreme weather events across the world. A link to their coverage is in the show notes. The end of August and early September brought a number of storms in Asia. Typhoon Shanshan blew over southern Japan in late August, although the storm had a similar track and intensity to previous typhoons that caused significant damage. Typhoon Shanshan had limited financial insured impact. August also brought Hurricane Ernesto to Bermuda. Then early September saw Typhoon Yagi soak the Philippines, and in the us, hurricane Francine hit Southern CoreLogic estimated the storm surge flooding put over 125,000 residential properties at risk in that insured wind in storm surge losses could be up to $1.5 billion.

MBS:

Wow. Well, this is so interesting, Joe. And I guess just to wrap today, the question that comes to my mind is, I mean, as you said, this is still really in its infancy in many ways, this Gen AI is new technology and how things have changed so much in just the last two years compared to the 30 years prior to that. Where do you think we’re going to go? Where do you think AI is going to take us in terms of the demand for the data centers as well as if you are a company trying to get ahead of the curve or ahead of the wave as this expansion, what are some things that people could keep in mind or think about?

JF:

Yeah, this investment opportunity that these hyperscalers are looking at, the metas, the Googles, they’re in the mix. There’s also some of the commercial investment areas that are looking as an opportunity for further investment as we build out new applications of whether that’s using large language models or generative ai, is that going to lead to these new applications for further productivity in the workforce? And then that in itself is a driver of demand for new AI applications. So we’re at the early innings. We don’t know exactly how far it’s going to go. This is the big question out there, right? How much more are we going to expect from the computational requirements and therefore the electrical and broadband requirements? And that’s a big unknown, but that’s the exciting

MBS:

Part. Yeah, absolutely. And I don’t know, I think we should put a little sticky note, Joe, and bring you back in three years time and see where the industry has gone just in three years too. Because I think things are evolving and changing so much that it’d be great to see kind of how we moved to accommodate these large language models and gen AI and the demand for data that they make and the demand for energy. So this has been great. Joe, thanks so much for joining me today on Core Conversations at CoreLogic podcast.

JF:

Thanks, Maiclaire, great to be here.

MBS:

Okay. And thank you for listening. I hope you’ve enjoyed our latest episode. Please remember to leave us a review and let us know your thoughts and subscribe wherever you get your podcast to be notified when new episodes are released. And thanks to the team for helping bring this podcast to life producer, Jessi Devenyns, editor and sound engineer, Romie Aromin, our facts guru, Erika Stanley and social media duo, Sarah Buck and Makaila Brooks. Tune in next time for another core conversation.

ES:

You still there? Well, thanks for sticking around. Are you curious to know a little bit more about our guest today? Joe FICA brings over 40 years of experience in location, intelligence and geospatial information technology with the purpose of providing clients the unique value proposition afforded by location-based software in data to applications in retail, insurance, banking, telecommunications in real estate. Joe has worked for large multinational companies as well as startups, and is now a principal in the location intelligence product team at CoreLogic.

The CoreLogic statements and information in this blog post may not be reproduced or used in any form without express written permission. While all the CoreLogic statements and information are believed to be accurate, CoreLogic makes no representation or warranty as to the completeness or accuracy of the statements and information and assumes no responsibility whatsoever for the information and statements or any reliance thereon. CoreLogic® is the registered trademark of CoreLogic Solutions, LLC.