Latest / Real Estate Meets AI / A Conversation with Will O'Donnell from Prologis
Transcript
- Sandeep Ahuja: Here's something wild to think about. Roughly 2.8 % of global GDP flows through one company's building. The company is ProLogis. And today I'm talking to Will O'Donnell, who's global head of corporate development and growth and founded ProLogis Ventures. They're a venture arm that deployed hundreds of millions of dollars across startups and is now helping the world's largest logistics landlord evolve from four walls a roof into what they call intelligent infrastructure platform. and they just partnered with Nvidia ⁓ build AI data centers. From warehouse to AI infrastructure, this conversation is going to be a ride. So let's dive right in. Will, welcome to Real Estate meets AI. Will O'Donnell: It's indeed pleasure. Thank you for having me here today. Sandeep Ahuja: So you have been inside more warehouses and logistical facilities than almost anyone in industry. You what's actually happening on the ground. So we'll kick it off with our AI meter. Where on a scale of one to five with A being AI averse and five being AI forward, where is logistics and industrial real estate in the world right now when it comes to AI? ⁓ Will O'Donnell: think ⁓ combination think people like most companies are at the entry stage, but really intrigued and trying to figure out what the applications would be. So AI use cases ⁓ have been in ⁓ for a long Whereas, okay, if you're doing inventory planning or ⁓ route optimization, companies using AI for those functions. Sandeep Ahuja: we're moving on to our second to the last segment called rapid fire. This is all personal. ⁓ These are one, one to two word coffee or tea. Okay. Morning person or night owl. Will O'Donnell: Alright. Coffee. Night out. Sandeep Ahuja: spreadsheets or slide decks. Will O'Donnell: god, neither? Probably spreadsheets. ⁓ I think obviously with the large language models and the massive influx come, people are trying to figure out ⁓ what the implications are. I'm really, really bullish on the opportunity for a couple of reasons. I one, it is physical movement of goods. ⁓ And one of the challenges the faces is everything is siloed. So we own a warehouse, someone else operates the warehouse. Sandeep Ahuja: Okay, I like that. AI tool you use most in the day. Will O'Donnell: chat GPT. Sandeep Ahuja: Okay. ⁓ prologes. Does it go PGT? you kind of like, know, made it your own name? Will O'Donnell: Prologis GPT actually, that's the technical brand. Oh, that's a good question. We haven't come up with a... We acronym everything else, so I'm surprised we haven't acronym that one yet. You have thousands of companies driving the trucks. You have people in the containers. You have the out there. So there's a lot of touch points by different organizations throughout a supply chain, which is why it's literally called a chain. And because of that, a lot of the data is siloed. And what the large language models allow you to happen get much better access to information in real time and connect the dots. Sandeep Ahuja: That's funny. I'm sure it's going to happen soon. I'm putting PGT out in the world. Will O'Donnell: Yeah, it probably is PGT. I've just yeah, I'm sure PGT is what it is. I just have been ignoring it. Sandeep Ahuja: Oh, that's funny. All right, well, last thing. You've spent 22 years at one of the companies watching itself reinvent itself. You have backed 53 startups. You have seen the works. What doesn't work and what's coming next? So what's the one piece of advice that you would give to a real estate logistics professional who's just trying to figure out their AI strategy right now? Will O'Donnell: You also start looking at things like computer vision on how do I take physical information and actually turn it actual data The challenge that a lot of companies are facing, is you need clean data. as a point, if your data is a mess, if your data is not consistent, if your data is in silos, it's really hard to extract the most value. Sandeep Ahuja: some are all in, some are paralyzed, and most are between. But what's the one piece of advice that they should walk away with, with someone who's definitely getting a front row seat and is part of building something so unique? Will O'Donnell: Yeah. I think one of the character traits that's going to be most important in the coming decade and beyond is intellectual curiosity. AI is new for everyone. the speed of what we're having to learn and evolve really push things forward is just different. ⁓ And the amount of that's happened. from AI that you could. lot of the work that industry is doing right now is how do we get our data in the right spot? But having more more conversations with our customers about AI, how to think about it, how to integrate it into warehouse operations, and how put the right infrastructure in place to allow people to maximize value. So being intellectually curious, a big fan of adjacent learning. So how do I go look at what other people, other companies? one my favorite books read that I recommend to people is Range. And it's really about the importance for to get a diversity of perspective and experience. And I think the ability to connect dots and see how Sandeep Ahuja: makes sense. So for all that are listening, there's a few things I pulled away. One, clean data is necessary. I think we've heard that on this show time and time again, garbage in, garbage out. So the cleaner your data, the more types of data points you're recording, ⁓ better it is. Two, I heard in some way that you're saying folks are kind of just starting to get interested. ⁓ Mm. just kind of dipping their feet. So on a scale of one to five, is that like a or a two? Maybe not a one, where is that? Will O'Donnell: different parts of an organization or different technologies can be blended to actually drive outcomes is a critical part. To answer directly your question of what companies can do, ⁓ think as a starting point, defining an outcome, like what important or where is the pain point that we want to apply AI against? And it can be completely overwhelming if I'm looking at No, I'd say probably on the industry-wide, there are two, three. have some companies, there's some very big tech-leaning companies out there that are very important logistics that are at the five level. That said, the technology itself is evolving and just the changes we've in the last three years. What was... Sandeep Ahuja: Yeah. Will O'Donnell: a prologist or a real estate company as a whole and say, what's everything that AI could do? But an easy starting point is like, what is a process that a significant number of our employees touch that they do on a very frequent basis? And where is their friction or pain within that process that we could potentially apply AI against? I think getting and building the muscle on how do you innovate? And one of the things that's challenging in real estate, like revolutionary three years ago is now antiquated. So just the leaps and bounds that it's making. excited about it because I think it's one of those continual learning processes. one really knows at this point where we'll end up or what the opportunity is. There's just a lot of excitement about it. One of the things that we had done because Sandeep Ahuja: Mmm. Will O'Donnell: Clean data was an issue that the prologists have been grappling with a while. And for about four or five years, we had metrics around data centricity and getting people to enter data correctly. Three years ago, we actually had 35 % of the company's bonus tied to six categories of data and hitting 99 % accuracy. And weekly were sent out. Everything was color coded. We're used to having a blueprint of a building and building that building. So there's a fair degree of certainty before I begin construction. In an aspect like this, like, okay, I know the outcome to some degree that I want to create efficiency around this process, how I get there and the steps that take are a learning process. So how to run pilots, how to look at the data and understand like, okay, even if it didn't do exactly what I wanted or expected, that's a win because I learned. When you put that kind of focus and light on it, we did it. And that was the year GPT's happened to be released in November. So one of the biggest changes that we had was prior to a lot of data centricity, it was a dictation that you need to do this. And the proposition or outcome wasn't being expressed necessarily in terms of the people who are using the data. Sandeep Ahuja: Yeah. Will O'Donnell: and moving forward. I think that comfort level of we're just going to try. reality is most decisions at a bottle on Amazon phrase are two door decisions. By that is I can make it and if it doesn't work, I'll go back and do what I was doing. There's very few things that are one door Like ⁓ once you're that door, like there's no going back. And especially testing and Sandeep Ahuja: Yeah. Will O'Donnell: because we didn't know how it would be used yet. But soon as GPTs came about, we have 97 % of the company using our GPT on a weekly, almost daily basis. we've had a couple hundred GPTs built by employees. have double, well actually probably 30 or 40 enterprise GPTs. So the takeoff has been dramatic and powerful. what we're spending a lot of time now is how do we really accepting the fact that a lot of these things are Two-door decisions like I'm gonna try this out if it doesn't work. Well, we'll just continue what we're doing. That's fine the reality is you'll probably learn and you'll better ⁓ and it's era process that people go through and I think the challenge that people run into also this is there there's a hundred questions that I feel like I need to answer What are like? Sandeep Ahuja: Yeah. Will O'Donnell: institutionalize it. So looking at processes that we can, rather than individuals using it to fix a component of the process, how do we build agents or how do we integrate in solutions we're finding across an entire job function. So it's really, really exciting, but ⁓ happen to be very fortunate that we have this initiative, the Cleanup our Data, and then suddenly people could recognize the value of it and the four or five binary questions that I could start with and run a small pilot because If question is like does this actually improve our process if the answer is no Then I don't need to go answer the other 99 and I'll go figure out something else But there's a tendency of time that humans want to be like well I know they're gonna ask me the tax implications of if I bill a customer this well excitement and the ability now to get people to maintain clean data is much higher because they actually can reap the benefits and understand implicitly why it's of value to them. If customers don't want to buy it, you don't ever need to figure out the tax implication. It doesn't matter, right? So I think breaking it down into bite size and then understand, we're piloting. It's completely two-way door. And what's the question we need to answer to give us confidence that we should progress and continue to move forward? Sandeep Ahuja: It doesn't matter. Yeah. Yeah, ⁓ that's fascinating. Even the way you tell the story, it sounds like it was almost the coincidence that there was this huge initiative to make the data so legible and so part of everyone's job. ⁓ And also AI started booming at a similar time. But I'm a foreign believer. Luck happens to those with vision. ⁓ So that that tells me about the leadership is about vision and forward leaning. That is sage, sage, sage advice. So for all that are listening, the two, three things that I picked up. One, be curious, stay curious, try things. And that kind of flows into the second. Try things. Don't be scared to try things. And it's OK to try them in a bite-sized way. You don't have to solve everything all at once, which makes so much sense. Well, Will O'Donnell. Will O'Donnell: Hahaha! Yeah, I think my... Yeah. Yeah, I mean, I think my high school wrestling coach always said luck is when preparation meets opportunity. The reality is, I mean, we had been doing AI work before and as a company, we realized data analytics was going to be critical to our success moving forward. It's just the democratization work. So before I had to go to data scientists and ask them to run queries for me. Now we all have extremely powerful engine. Sandeep Ahuja: Yes. founder of Prologis Ventures, global head of corporate development and Prologis. Thank you for bringing the Real Talk today. This was such a pleasure. Will O'Donnell: No, appreciate it. It was awesome. Good conversation and thank you for having me. Sandeep Ahuja: It just is easier. It just is faster. Will O'Donnell: Yeah, it's just easier. So yeah, it was fortunate, but it also our company had clearly recognized that be competitive the next decade. We needed to start with making sure that we had the right data and clean data and the infrastructure that our IT team has built is world class, just our data stack and pulling all our information of individual solutions or SaaS and do a common data lake. So we're just able to Sandeep Ahuja: Yeah. Will O'Donnell: do some really cool interesting things with ⁓ Sandeep Ahuja: love it. So let's talk a little bit about the Prologis transformation. We've talked a little bit about the last three years and the focus on data, but you've been there 22 years. You've watched it go from AMB property the huge ⁓ multi, multi, dollar merger Prologis. And now the describes itself as intelligent infrastructure. So take me inside that evolution. What does Prologis look like today that would shock someone? Will O'Donnell: Yeah. Sandeep Ahuja: who last paid attention, I don't know, five years ago. Will O'Donnell: think estate's an odd business in you spend three months negotiating a contract with a customer and then try to ignore that customer ⁓ for next five years. Every other you would go into, like think of bank. You an account, they offer all sorts of services and value and their goal is to... Sandeep Ahuja: you Will O'Donnell: capture more of your wallet, but really provide a holistic solution because for convenience wise, I don't want to have to go to one place for my bank account, another place for my mortgage, another place for my car loan, another place for... Right? So there's a convenience factor that by combining these in and bundling these solutions together, consumers find value in it. we look at real estate, historically it has been, well, here's your floor walls and a roof and each customer, you go deal with it individually. Sandeep Ahuja: Sure. Will O'Donnell: When we took a step back and we're really looking at where we have unique advantage and what's to us, customer is always kind of number we actually have a very unique advantage because of our size and scale to move beyond the premise of just sign the lease and we'll call you later to how do we really become a partner throughout the life cycle ⁓ of your business? The way we've thought about it is one of the most powerful things that we can do for our customers partner with them to extract the maximum value they can get out of their So when they own a warehouse, how do help you optimize the throughput? How do we help you retain labor? How do we help you integrate in technology that allows systems to work seamless? How can we provide a digital layer that attacks freight theft, for example? Sandeep Ahuja: Mmm. Will O'Donnell: And because we have the size and scale, we can actually invest in solutions across a much higher customer base, which allows us to create efficiencies, bring down costs and create actually market standards. But for customers, there's value because we can help simplify their process running the supply chain and bring in our best practices and our knowledge and our unique insights. to help them extract the most value they can from their fixed assets. Sandeep Ahuja: That makes sense. So really what you're saying is a pretty major transformation that has occurred over the years as, know, Prologis has continued to invest in technology and, you know, share the wealth of knowledge that it has accumulated with its customers is that you're helping your customer serve their customer so that for all parties involved. Will O'Donnell: Yeah, and it's things where, to your comment earlier at the beginning on 2.8 % of the world's GDP through our warehouse, like, that's incredible amount of data that to help drive efficiencies for our customers, help them understand where maybe friction or pain points are happening. The comment I was making earlier about how all these different silos, We happen to own a physical asset that a lot of these different silo companies flow through. So anything we can do to help enable adoption, enable better visibility, a benefit. And ⁓ an example, when you drive past industrial parks, you'll see trucks lined up outside waiting to get in. We can create a more efficient check-in process using digital cameras, et cetera, and tie that to a bunch of workflows. That's lost cost. Usually when you're eliminating a cost, it's at the detriment of someone, but the trucking companies don't want their drivers sitting idle. The shippers don't want people sitting outside just sitting idle. And the municipalities and communities don't want trucks just idling So ⁓ a lot of opportunities we feel like we're uniquely positioned to attack. And then sustainability is a big area of focus. So our goal is to really be able to provide our customers renewable energy across our platform at or below the cost of what they're paying already. Energy generation is warehouses adopt more technology. You start putting EV vehicles in, ⁓ the grid needs to be upgraded. And yes, we are starting to data centers earlier, which you mentioned. Sandeep Ahuja: Yeah. Will O'Donnell: We spent a lot of time thinking about energy generation. So how do we create on-site on-prem generation to help our achieve their business goals as the grid struggles to catch up to what's needed in the current environment? Sandeep Ahuja: Totally, super, super insightful. The last comment here and then we'll move on. But ⁓ thing that I'm hearing, which I think is such an exciting massive shift is that you're not building a box, you're running the operating system inside it is ⁓ simplest way ⁓ I understand what you just said. So moving on, elephant in the room or maybe the Nvidia chip in the room, Prologis obviously just committed $25 billion to building data centers. Will O'Donnell: Yeah. Yeah. Sandeep Ahuja: And in February, you announced this partnership Nvidia, a to study micro data centers for distributed AI. The world's largest warehouse company is now literally building the homes for AI. Walk me through the strategic logic. How did you get from we store stuff to we house artificial intelligence? Will O'Donnell: Yeah, mean, I think solar logistics company and our business has traditionally been logistics of moving atoms. Data centers are the logistics of moving electrons. from that perspective, it's just a continuation of the strategy. The reality is ⁓ we have almost buildings, we have 14000 acres within our land bank. We have an energy team who's exceptional. and knows how to energize land, has very deep relationships with utilities kind across our markets and is viewed as a fantastic partner. I mean, we have a balance sheet and we know how to execute. So utilities like working with us, there's certainty when we're going to hyperscalers and others that we know how to execute. ⁓ And have a pretty significant pipeline ⁓ of land between our land bank and existing buildings looking to convert. So it's an area that we've been very active in in the last couple of years. ⁓ Last year and a half, two years, we've built out best in class world class teams. It is important. There's so much opportunity that we really wanted to scale up and have the expert resources so we can have a incomparable opportunity just on execution. something very excited about. It's obviously a growing area. There's a ton of capital, but. Sandeep Ahuja: Mm-hmm. Will O'Donnell: I think we are very uniquely positioned because of our existing platform and our expertise and energy and our ability to build buildings that we have a very unique value proposition. Sandeep Ahuja: could see that. Most definitely I could see that. In fact, the micro data center concept, it's so fascinating. The five to 20 megawatts near utility substations for AI inference workloads and the five pilot sites, I think I read somewhere by the end of 2026. Why does that model make more sense than just building massive hyperscalers? Because it sounds like it's both and. yeah, yeah. Will O'Donnell: Yeah. We're doing both. Yeah. I think with the, the micro data centers is something that we're starting to explore. where we have been most active up until now is the hyperscalers. because again, I mean, that's where the industry has been most active and that's where the most demand is. so ⁓ it's a thesis we're testing right now and beginning to explore on, is there opportunity for these microsites? What would it look like? and how would we go about it. And then ⁓ have a very robust pipeline on the hyperscaler, much larger 300 ⁓ data centers are out there right now. Sandeep Ahuja: Yeah, and that's a pretty big impact too, and the connection to the utility and the ability to know how to build. are certainly unique skill sets. All right, so let's kind of shift to your VC hat. You founded Prologis Ventures and you've invested in, I want to say ⁓ 50, 53. ⁓ Will O'Donnell: Yeah. You 53 companies, yeah, 53 across supply chain, PropTech, energy, sustainability. Obviously, we've done a lot of stuff in data and AI. The way we've looked at it is we're a unique position because of Prologis being our LP, but then also our deep customer relations we refer to as a thesis investor, we'll spend a lot of time. mapping out workflows, out, for example, the data center industry and there's opportunity, where there's friction and where potential startups could provide a disruption. And in other cases, we're looking at very mature companies that come in and be partners. So we do a pretty extensive by-builder partner analysis, but we call ourselves stage agnostic because at the end of the day, Sandeep Ahuja: Mmm. ⁓ Will O'Donnell: Our goal is to be able to go in and find a solution for identified pain point. It allows us to be much more pragmatic than, okay, blockchain's coming out is going to revolutionize the world. And it was followed by Web 3.0 and now we're in AI. The reality is we're in the business of commerce and there are very specific reasons why the industry has developed the way that it is. And there's constraints that are there. So our goal is to find the most relevant technology. for that and be able to solve and provide solutions ⁓ Yeah Sandeep Ahuja: That makes sense, you know, kind of on a general thesis. I think I fully where you're getting at, but what's the number one pain point that you're currently investing against for all that are listening? Will O'Donnell: I mean, one that everyone's looking at right now is around energy. resilience, generation, et cetera. I that's an area we're spending a lot of time in. And combination of how do we build out business lines within Prologis that can provide solutions and how do we find partners that we can in. The two other areas we're spending a lot of time would be an obvious one, AI. So what are the implications of AI? What will today? What's the future going to look like? How do we adopt it ⁓ within And then how do we provide AI-driven solutions for our customers, especially on that framework of us being an operating system for their business? And then the group that kind of the infrastructure strategy behind everything inside and around the warehouse, what we call operations essentials. That's a big priority too on how do we find the right partners, with startups and sometimes very mature companies to be able to provide that holistic solution for our customers. So our is not to be the best in class for 8,000 different categories, but we can be the strategic partner that can help choose our partners. Sandeep Ahuja: Mm-hmm. Yeah. Will O'Donnell: what solutions work best for them. Sandeep Ahuja: that makes so much sense and why wouldn't you? So obviously you've got companies like, you know, Locust Robotics or Outrider or Gideon doing autonomous forklifts. So give me the before and after. What did a warehouse floor look like two years ago versus what your portfolio companies are enabling today? Will O'Donnell: Yeah. the time difference little more than two years, but I think the ⁓ challenge with is it's all throughput. goods come in, you have to store them and then move them out. And as you think about e-commerce, the customer expectations on delivery has changed pretty dramatically. So we all believe it's our God-given right that Sandeep Ahuja: Yeah, please. Will O'Donnell: I should get a package in a day, right? So what are the implications then on supply chain that to happen? So that's why pretty massive investments have been made or starting to be explored around automation. So how do I leverage automation to enhance the efficiency of a labor workforce? Something like Locust is a great example where in picking and packing, used to have humans have to walk seven to 11 miles a day grabbing things off shelves. Sandeep Ahuja: Mm. Will O'Donnell: With Locus, I can stay in one spot and the robot comes to me, I put the good in and it goes and goes to somewhere else. So you increase the satisfaction of the job pretty significantly, you increase the throughput, et cetera. So I think you went from very manual, like siloed operations, things were done on Excel paper, you have WMS systems that were inventory to much more complicated. so you're trying to integrate automation into it now. You're using cameras, as we were talking about earlier, computer vision to optimize operations or for security purposes. You're really, what our customers are getting very excited about is how do we capture and create data that gives them actionable insights? The supply chain will always be unpredictable. There's going to be tariffs and floods and et cetera. Sandeep Ahuja: Mm-hmm. Will O'Donnell: But the faster I can compile information and connect the dots, the faster I can respond. So what our customers and probably one of the biggest differences is where a lot of time ⁓ spent trying to compile and build and research what was happening, the faster that happens, the faster humans can be focused on decision making and the steps forward. And that's where we're seeing Sandeep Ahuja: Yes. Will O'Donnell: A lot of our customers get excited about use cases in AI. It's how do we increase our ability and efficiency of our workforce by giving them the right information at the right time. Sandeep Ahuja: Yeah, yeah, all of that sounds so natural. It sounds so obvious, obviously, you know, to spend time and energy. yes. ⁓ AI. ⁓ Automation making operations better. Absolutely. ⁓ the flip side, obviously you get hundreds of pitches every year. What's the pitch type you keep getting that you're just like, listen, this is never going to work for warehouses and data centers. Stop, stop trying to pitch this. Will O'Donnell: I think it gets back to when technology is in search of a solution versus there's a very clear defined pain point that someone could put things against. I think one of the challenges that exists in an industry like supply chain or prop tech, that unless you have a very clearly defined understanding of what the problem is, the adoption of technology or the use of technology is off and the thing which ⁓ I underestimated, even having been at ProLadis for 12 years when we started Ventures, was the amount of change management it takes to get people to behave differently. And I this is where... ⁓ Sandeep Ahuja: That's the hardest thing in the whole world is to get people to change what they do. Will O'Donnell: Yes, exactly. that is what a lot of companies struggle with that it's incrementally better or it's not really solving a real pain point, it could be the flashiest grade, like interesting technology, but people aren't going to adopt it because it didn't actually make their lives easier in a way. I mean, if you think about really great technology, you pick it up and you know how to use it and you don't have to go through Sandeep Ahuja: Yes. Will O'Donnell: 10 hours of training, you don't have to relearn, it just feels a natural progression. Technology that's incrementally better or completely makes people change the processes is tough. And we at Prologis have spent a lot of time as we're looking at technology, also looking at the processes behind it. And we have the right process in place? Are we efficient? Is there extra steps that we're taking that we shouldn't be? Sandeep Ahuja: Mm-hmm. Will O'Donnell: Because digitizing bad process just leads to digital bad process. So we've seen where it's like, okay, well, I want a new technology, but I want to do everything exactly how I did it. And all you end up with is a fancy digital screen of their own process. ⁓ And it doesn't ⁓ So it's one those things like we have a operational excellence team and they spend a lot of time with internal groups really mapping out Sandeep Ahuja: Yes. Will O'Donnell: both the outcome of where we want to get to. So defining that North Star is super important because people think they're on the same page. And then six months later, they realize like, OK, we now started going on a road trip and you got in your car and drove the San Diego and I drove the Miami and now we're pissed at each other. Yeah. But it's really clear outcome and what desire is. then mapping out the process to get there. Sandeep Ahuja: We both have great weather, but totally different great weather. Yeah. Will O'Donnell: making sure it's repeatable and then that's when we start applying technology against it the change your management is easier you get faster buy off because people under stand the outcome and they feel a part of actually designing the optimal solution. Sandeep Ahuja: Yeah, yeah, couldn't agree more. More thoughtful, even if it takes a longer, but it's a super thought out process, like taking a step backwards is sometimes saves, I don't know, two, five, 10 steps in the future. Will O'Donnell: mean, personality is one likes to move very fast. ⁓ but it's, ⁓ it's sometimes you have move to move fast. ⁓ And the analogy often uses the trip one that like, if I just announced we're on a road trip and everyone's that runs in the car in six months, it's a complete cluster. Like we're yelling at each other. Like we didn't plan, but if you actually sit down, plan it out. Sandeep Ahuja: Sure. Exactly. Will O'Donnell: define it and really good process gives people freedom because they then know the rules of engagement the lane. Bad process is bureaucratic. But if you spend the time developing here's the outcome we're going to here's the process each of our roles. Yeah that takes a little bit more time up front but six months down the road everyone's rowing in the same direction. When changes happen which they do like we're doing really hard stuff and if Sandeep Ahuja: Yes. Yep. Will O'Donnell: If we didn't make a mistake, we weren't trying hard enough. So you need to make sure people are aligned, them being data centric and looking at data and communicating, here's what we've learned and here's why we're going to pivot. But that work upfront allows it to happen in the success versus, yeah, we're going to go do this and everyone in the company needs to behave differently. Good luck. Have at it. Sandeep Ahuja: Sure. Yeah, that does not sound fun. All right. So that was super insightful. Will O'Donnell: Now.