Listen: eRobot: Energy sector embraces AI in race to innovate and transform

View Full Transcript

Hill Vaden:
All right. This is Hill Vaden, one of your EnergyCents hosts. I am here with Sam Humphreys. Sam, how are you?

Sam Humphreys:
I am great. Thank you, Hill. The sun has finally come out in the UK, and it's glorious. It's going to last all of the day. I have not been in such a good mood in ages. How are you?

Hill Vaden:
I'm all right. The humidity has finally come out in Houston. You now sweat through your shirt when you're standing in the shade.

Sam Humphreys:
Excellent. That makes for optimal running conditions there.

Hill Vaden:
Yeah, welcome to summer. We have just finished a fascinating conversation, there's a modifier for you, with Dan Bennett and Jud Jacobs on AI and its application in energy, outside of the data center. We don't get into data centers at all. We're talking about what energy sector, what energy companies are doing with AI. Can you give, I guess, first, your thoughts? And two, could you give a couple high points for people to listen to as we get into the discussion?

Sam Humphreys:
Yeah, sure. AI is obviously something that is a hot topic at the moment. You cannot look at any part of your life without AI somehow taking some shape in it. So it was great to have these guys on to just shed some light on where it fits into the energy sector.
A few things I particularly found interesting is something that I wasn't expecting. For energy exploration, for example, I kind of, maybe naively, assumed that it would be these large, super major companies that were driving innovation. Maybe that wasn't the case with some of these smaller operators that are more agile and have got more to play with, and less decision makers that are really experimenting more and leading innovation. And then where does the data ownership lie and how does that impact companies, and people as well? Just the pace at which this is all going and where it can lead to. So yeah, there's an awful lot to unpack there. We did whiz through it in only 30 minutes. But what about you, Hill? What were some takeaways for you?

Hill Vaden:
Yeah, I mean, I guess just big picture, all of the ... there's a lot of hype and sometimes that causes people, perhaps myself, to be skeptical. But I think there's a lot to be excited about within the application of artificial intelligence to the energy sector. There's all the pitfalls and all the concerns that people have raised and will continue to raise. But I feel pretty encouraged and pretty optimistic and excited to watch what's going to come out. With that, we can hand off to the rest of the podcast.
Sam Humphreys:
Enjoy.
Hill Vaden:
Welcome to EnergyCents, the show where S&P Global Commodity Insights dives deep into the captivating world where energy meets finance. Today, Sam and I have a thrilling episode lined up for you as we sit down with the remarkable Dan Bennett and Jud Jacobs to discuss the fascinating realm of artificial intelligence in the energy sector. Sam, that was written, that intro, by artificial intelligence.
Sam Humphreys:
It sure was. Hello, how are you doing?
Hill Vaden:
I am doing all right. Jud, Dan, how are both of you? Thank you for joining us today.
Jud Jacobs:
Absolutely. Thanks, Hill. Thanks, Sam.
Dan Bennett:
Yeah, great to be here.
Hill Vaden:
For listeners, I think that the modifiers, captivating, thrilling episode, that to me is what gives that away as being written by artificial intelligence that there's ... We've got more written in front of us about artificial intelligence. We're not here only to talk about large language models in artificial intelligence.
Dan, I want to start with an overview from you on artificial intelligence more broadly, and really with, I guess a bit of skepticism from me. You've seen this already, but 50 years ago we were promised flying skateboards and bipedal droids. We got Twitter and we got Facebook, which is in my book, somewhat of a letdown. If I look at Michael J. Fox on one of these skateboards and I'm sitting on Twitter, I'm not actually on Twitter, somewhat of a letdown. But how can we be sure or how confident are we that artificial intelligence really improves the energy sector versus allows us to come home to an empty house and tell somebody we love them, and get a response from nobody but a robot?
Dan Bennett:
Yeah. I think you're underselling it. We got Twitter, we got Facebook, and we got hyper-personalized adverts, that's surely a win. Flying skateboard thing, I like that because actually that's a hardware problem. Actually, part of the answer to the question is hardware is still hard and software is able to innovate at a way that far exceeds our ability to innovate with hardware.
You're right that AI goes through these boom-bust cycles for ... I did my degree at ... My undergraduate was computer science with AI back in one of the ice ages of AI in the '90s. There'd been all this hype and excitement. The Japanese in particular had made some real progress in the '80s on AI, and then it kind of collapsed under its own weight. And then what happened was big data happened, cloud happened. Moore's law continued to drive down the price of compute. All of a sudden, we get to where we are now where the cost of training these models is continually coming down, and the ability to train the models is to anyone with a credit card. That's one of the big things that's changed is 30 years ago you had to invest millions of dollars in equipment to even get started with this. Now it's anyone, like I say, with a credit card.
Now, to the energy industry, the way I think about it is that they've been using AI for a long time, but it's AI as the classic data science AI. My head of data science likes to say, "The best AI happens, best data science happens when you have that data scientist right next to the industry expert, the domain expert," because it's only by putting those two together that you can get innovation there. Things like predictive maintenance of good use cases, long history. Well, what we've had is this consumerization. We've had the iPhone moment with AI. I think that genie's out of the box. I don't think it goes back in the box. That really opens up new areas for every type of industry, including the energy industry.
Sam Humphreys:
Well, we are going to talk about energy. Jud, I'm going to bring you in here. We're talking a lot about innovation with AI and how companies are using it. From an energy perspective, what companies are kind of leading the charge in that investment? Are we seeing any particular focus from NSDs, the super majors? Or are smaller independents coming up alongside these companies and jumping in and aiding ... won't say aiding and abetting, but aiding them in this development?
Jud Jacobs:
Yeah. It might help to ... I can just give a bit of history here as well, AI and energy. Dan mentioned this, that this is not something that's new to the energy sector. We'll talk specifically a little about oil and gas now, but we'll definitely expand this into the utility and power sector as well.
Oil and gas is one of the leading managers of large data sets. If you look at some of the leading high-performance computing centers, they've long been some of the large oil and gas companies dealing with massive data volumes. So certainly we've been using AI and applying it in data-driven approaches for a pretty long period of time. What really changed probably about 10 years ago or so, during the oil price downturn at that time, was the creativity that comes out of the pressure of difficult economic times. What we began to see is that there's really this formation and centralization of AI. Specifically at that time is a lot of the unconventional operators.
They were basically looking at this and viewing this as an existential crisis. So they knew that they needed to more efficiently and more cost-effectively develop their resources. Just at the same time that they were cutting everything, they were cutting the coffee in the cafeterias, they were actually investing in AI at this time. And actually developing teams that were reaching out and saying, "How do we drill faster? What's the optimal frac program? How do we improve our supply chain and logistics?" So it was really at that time ... these smaller companies that were able to do this. The decision cycle is much smaller than at a large company. So really, it was at that period of time that we saw these types of companies really leading the charge and being effective, and they still are as well.
That's not to say, and we've certainly seen over the last five or 10 years a lot of the larger energy companies, including the IOCs and the NOCs, doing the same. They've actually developed some of their centralized groups to be able to do this. They're certainly having some successes, but I would say at this time, I think it's the smaller nimbler companies that have really effectively done this in practicing that. But we can talk more about what some of the larger companies are doing as well and how they're pushing things forward.
Hill Vaden:
Jud, you just put this in the context of cost reduction. The Onion, the newsletter, I don't know if you would call The Onion a newsletter. Published something several weeks ago or several months ago that was talking about all the fears of AI taking our jobs. The comfort was that in every scenario possible, that The Onion had tested, humans cease to exist once AI has taken over all these processes. So we don't need to worry about losing our jobs, but can you talk a little bit more about AI from a non-cost reduction standpoint within the energy sector? Is it all about cost reduction?
Dan Bennett:
I can jump in with a quick example here. Cost reduction, the other side of that coin is capital allocation, right?
Hill Vaden:
Mm-hmm.
Dan Bennett:
So there's a long history, again, oil and gas using AI to identify where you might drill a test well. That's, again, a lot of combination of AI models that might predict the production curve of that well, all of those kind of things, along with the human expert, the geophysicist or whatever it is, that can take a look at that and really make a sense. So I think it's more than just cost reduction, it is about helping with capital allocation. In capital intensive businesses, which energy tends to be, there's a strong built-in demand to use as many tools as possible, AI included, to make the right decisions there. Because you make the wrong ones, you lose big to some of your competitors.
Jud Jacobs:
I would absolutely agree. I think it is certainly beyond ... We talk about efficiency plays, whether that's cost reduction or whether that's improving productivity, or automating basic tasks to improve the productivity of your staff. The way that we looked at it, when we look at all the use cases that we've talked to our clients about, we kind of have a bit of a framework. One of it is driving greater efficiency. These are all the things that we were just talking about. But then, as well, there is this, much more, managing large and complex systems. This is to what Dan ... whether you're managing these in a capital allocation manner or whether you're managing them in an operational sense. Capital allocation, if you're not a conventional operator, what's the optimal well spacing? What's the optimal frac stages, and how do you develop that to maximize the value of your asset? You can think of it the same way in renewables as well, like how do we develop something, a green hydrogen plant, and optimally balance batteries and generation, and the others.
The third element that we look at is accelerating the innovation cycle. We've definitely seen this on the energy side. Certainly on the low carbon side where we're challenged by cost, is how do we come down that learning curve faster? AI allows you to, if you have high iteration going from one project to the next, how do you accelerate that learning from one iteration to the next and drive it down the last?
Then there's a really interesting one that you're beginning to see, especially around some of the low carbon areas, but around taking a drug discovery-like approach to new materials and advanced materials, whether those are battery materials or whether those are a more efficient carbon capture materials, but all of those things. So that's where you kind of really begin to get into a lot of the power of AI. Those are the three types of things and buckets that we see some of the applications.
Dan Bennett:
Coming back to The Onion's point, I'm super skeptical in my lifetime that we'll have this general AI that just replaces us all. To me, this is the next revolution. We can be Luddites or we can choose to harness the technology that we've got. There's a great meme out there, that, "It's not my job's replaced by AI, my job's replaced by someone using AI to do my job." That's the way I think about this. I'm a technologist. I look at its ability to write computer code. That's great to a degree, but I still need to be able to ask the right questions of the person I'm writing the code for. The AI can't do that. I still need to think more strategically about, "How am I going to fit this thing in? Is it using licenses that are appropriate for my corporation?" All those sort of things. So I don't worry about the jobs thing, I really don't.
Sam Humphreys:
Hill, I'm going to steal a question that I think you were going to ask, just because it leads on quite nicely. We're talking about the use of AI for, let's go with upstream oil and gas for example. Jud, you made a point that I found quite interesting is that unconventional players were maybe using it more than I thought. I assumed, maybe naively, that it would be ... conventional exploration of AI would be of more benefit at the moment because of the vast amounts of data that there is. We know that a lot of these models are learning off massive historical data.
So this is kind of two questions. How does, then, AI evolve and fit in with conventional exploration? Assuming that we have to counteract bias that comes within AI. We can't tell them where it's not been explored because we've got the data that has been, so where ... Is it going to look for the same thing somewhere else? And then on the flip side, how is that impacting unconventional in the new spaces? We talked about carbon capture, renewables, where is that going with this? I'll throw it to Jud. You can go first.
Jud Jacobs:
Yeah, I think the unconventional operators just benefited from AI because it was high iteration. So they were able to learn very quickly. High iteration, high speed, so they were able to learn very quickly from one to the other. That's not to say ... Certainly in more conventional developments, there's a lot of applications. We've seen some really interesting ones as well, "How do you use AI as a direct hydrocarbon indicator?" Essentially, how do you learn from past basin analogs and past discoveries to then apply it and use it as, essentially, truly the secret sauce? We've talked to some of the operators that have said that this has allowed them to improve their targeting of their developments into the tune of billions of dollars that are now being credited with some of the very high profile discoveries that have been made recently.
Certainly, if you look at things as well, in a lot of, let's say, conventional oil and gas development, it's processing and considering different scenarios of some of these large data sets. So if you're applying gen AI and traditional AI to, say, seismic processing and going through the different iterations, the different scenarios of development, you could go through many more than you would've before, which allows you to become much more accurate. So they're able to take a much more data intense approach to that. Again, they're seeing good success on that. Because I know we're veering now towards oil and gas, but there is a lot that's going on in the low carbon and the renewable space as well.
Hill Vaden:
Do you want to bring up some of that now? I mean one of the questions that Sam and I have been asking each other, Sam had it there with geology or with exploration, that if AI is using past learnings to inform itself, how do you apply that to an energy transition?
Dan Bennett:
I'll have a go at that first. AI is sometimes using past learnings. That's the classic case we always think about, but there's other ways you can use it. Something that you can do in transmission networks and power forecasting is you can build these very, very complex models of the world. Weather forecasting is like this. You build these complex models of the world, which are very numerical, high-performance compute, and you run different scenarios for you to get the answer. That's not AI. The problem with those is they can take an awful long time to run.
Well, what you can do is if you've got a model that you have faith in, you can essentially do a Monte Carlo on it and run through a whole bunch of iterations of the model with different input parameters. Give yourself a synthetic set of input parameters and output parameters and then train the AI on that. The AI will give you the inference in a second or two. What you've done is you're using the AI then as a screening tool to help you avoid a long compute run for a model that may give you an answer, that, "Ah, no, that's not the thing I wanted to think about." So yes, you're right, historical data is where everyone lands, but it doesn't have to be. That's the cool thing. It's anything where there's input and output, whether it's observed or modeled, we can use AI to short circuit some of that.
Sam Humphreys:
That's great. But where we are using historical data, I just really want to get to this, how do we overcome bias in this? Because I've seen this across other forms of AI just on the internet talking about, I don't know, who's going to be the next pop idol champion? I don't think pop idol's the right term anymore, but it's taking those things. So how are we overcoming that at the moment? Is that still down to human observation?
Dan Bennett:
Yeah, it's by having smart people look at both the input, the training data, and the output. Our rule internally is we do not use any generative AI output in any of our products without having a human review it first. Bias is one of the reasons we have for that, as well as hallucination. You're absolutely right, that problem hasn't gone away. We all have our own biases as well, so we may not even spot the bias in the training data. What you're really talking about is the need for AI governance. Governments are kind of cottoning on to some of this right now. There's this idea of responsible AI. It's the ideas of, "These are great tools, but like any great tool, we can do damage with them." A hammer can put a nail in, or it can put a hole in the wall, or it can make your thumb hurt. All three are viable uses of the tool. So, like any of these tools, we have to have intelligence and smarts and rules and policy around how we use it.
Jud Jacobs:
I think kind of gets a little bit into the progression of AI, AI in general and gen AI specifically is that a lot of this comes down to ... especially in industrial settings, it comes down to the trust of the individuals who are actually making decisions using these tools. What we've begun to see, especially in a unconstrained world, some of the things that we've seen is that what the solution that's coming up with is not a viable one. It's either not a viable one or it kind of breaks certain rules of operations.
The challenge of that is that you get individuals who are being promised these great capabilities, and their first instances they're seeing are ones that a graduate engineer would know that this is not something that you could do. This is really where it gets into, in industrial settings, that, "How do we actually derive value from this?" It's going to be being trusted by the individuals who are going to make decisions off of it. That's really been a challenge.
Dan mentioned the decades we've been at AI. Certainly, in energy and oil and gas, we've been at it for a decade plus. But we're still just beginning to see some of the benefits because, really, the engagement with the business between the technology folks who are developing the solutions, and then the engineers and the technical folks who would be using them, that's just beginning to open the door and begin to see some of those use cases. But they're still kind of sporadic within companies. The challenge is getting them at a much wider scale.
Dan Bennett:
Yeah, right. If we come back to one of the things I said earlier about ... that we've reached this tipping point of the consumerization of AI, if there's one good ... I mean, there's many good things, but if there's one good thing about that, what we're doing culturally, whether we recognize it or not, is we're training a lot more people about what AI is, and what to believe in and what to not believe in. So the more we put these tools in everyone's hands, the more they learn to use them, the more they understand what they're good at, what they're not good at, and to not just blindly trust, the better we'll all be off.
Hill Vaden:
I think in some of it, also, that it becomes less intimidating when you look backward, I guess. I mean, a thermostat in my house is, in some forms, AI. The cruise control on my car is AI. Those don't threaten me. To be honest, I agree with what you said, Dan, the person who knows AI better than I is going to take my job, not AI itself. So I think The Onion article is funny rather than scary. I think there's this whole continuum of what is ... AI doesn't have to be the robot from outer space. It can be the thermostat in the kitchen that is keeping my house whatever temperature when I'm at work.
Jud, you talked a little bit about the companies. Maybe this is Jud, maybe this Dan, I'll let each of you jump at it. We hear a lot about Microsoft. We hear a lot about Nvidia. We hear about all of these non-energy companies leading the AI revolution, if there's such a thing. Can you talk a little bit more, one of you, about where the energy companies are driving things? And do we expect to see new company formation coming out of energy company spinoffs as a result of this work within AI? Or is it all shared or leveraged by just applying Nvidia to things that Exxon does or whomever?
Dan Bennett:
I'll take a start at that. As we talked about right at the beginning, this does require a ton of compute. It does require some advanced silicon and things like that. So the hyperscalers, the Nvidias, the AMDs of the world are positioned to be able to provide those tools. Above that, provide accessible APIs and models and all the other things you need for a regular developer to be able to harness these things and take advantage of them. But what those hyperscalers tend not to be good at is really deeply understanding any particular business domain. I mean, they will tell you, and they're right, that they have focuses on energy industry, but it's not the same as being in it operationally every day or any other industry.
So I expect that you will see energy companies across the spectrum. Some of them will make breakthroughs. They'll be doing it on the back of all of this ... the stuff the hyperscalers make available to them. Some of the national companies that aren't allowed to use cloud, that I really worry about, because I think they're locked out of some of this stuff, but if you exclude them, you'll be creating these innovations that are things that the hyperscalers wouldn't be able to do on their own. And yeah, you'll see spin outs from power companies or whatever because someone created a better mousetrap and the utility or the energy company has only got so much appetite to invest in that mousetrap.
Jud Jacobs:
Yeah, it's definitely an ecosystem. People have gone into this with an uncertainty around AI and some of the other things. So initially what we've seen is a lot of partnerships. I think with the hyperscalers, there's real value in that. There's real value in understanding how the AWSs and the Microsofts of the world, how their compute infrastructure is set up and how do you partner with them to ensure that what you're developing is optimally designed to do that. There's those types of partnerships, and then there's partnerships with some of the leading AI companies. And then startups as well. You see partnerships between let's say BP and Palantir. You see Shell and SparkCognition, and some other ones.
I think the companies that are really doing this well, they mentioned that they are continuously looking outward. So they recognize they can't do this. They're continuously looking at startups. They're not so much taking the traditional corporate venture model, because this is moving so fast, but they're partnering with them where it's appropriate to do so. Absolutely, I think the most successful companies, I won't use the inverse, but the most successful companies are those that are collaborating, partnering, looking outside of the organizations, rather than those saying, "Hey, we need to do this internally because it's all IP and we don't want to share this with anyone."
The question around the spin outs, it's an interesting one. I think we will begin to see some of this develop. We've seen it a little bit initially in oil and gas, I think. They're the Cognites of the world that came out of the Ocher Group. Actually, interesting, a national company, PDO, out of Oman, Nebros or Novatech was a company that came out of it.
But I think a really interesting one is ... We've talked a lot about oil and gas. I think there's a lot going on in the low carbon and utility space, but NextEra Analytics. It's a company that ... They bought a company. They built that up. It was largely internally focused around providing those solutions. In the last two or three years, they've opened that up and now they're providing services to other utilities and power generators. It's a really interesting development that they are now saying, "Okay, well, what can we develop and what type of services can we offer into the industry?" For a group of companies that maybe haven't traditionally been innovators or new technology developers. So kind of offering that suite to those types of companies.
Sam Humphreys:
Just very quickly, I'm just looking at the clock, I've just got one more question I really wanted to ask. When it comes to these partnerships, who owns the data? Because you'll have data from these companies that are being put into these models, as they need to be, but then what does that mean long term if these companies expand and these models become open models? We've got ChatGPT at the moment. That can scour so much information publicly. I don't know where it gets everything else from. So where- how does that go? Dan, I think you were going to take this one.
Dan Bennett:
Yeah. This is one of my favorite topics, so I'll try and be brief. Look, we're seeing all this litigation against the LLM providers. OpenAI, of course, is the one with the biggest target on its back right now because a lot of people feel that they put their data on the web for search engine optimization, not to train an LLM. Actually, OpenAI and Google both responded to that with ways that you can tell them, "Hey, this is good for search, not good for training."
It's clear to me that the ownership of the data is ultimately the most important part of this. The actual answer to your question, the tongue and cheek answer to your question is whatever the contract says when you wrote it with that partnership. If you are that energy company that's going into that partnership, I would suggest you pay real attention to that because there's a long-term value stream from that data, and/or from models trained on that data, that you probably reasonably feel you should be entitled to. So the question is, does your contract permit that?
Jud Jacobs:
It's a really interesting one. We've seen this come up a lot. When you think about it, one of the group of companies that are leveraging this as well are the OEMs. They're running their piece of equipment and they're mandating that they have access to the data because they want to improve their piece of equipment for all, not just for that individual.
But you're right. Dan brought up the point of contracts. That is what we're seeing is that a lot of these contracts that have been historically written have not taken into account a lot of this kind of data sharing and data ownership. I think getting the contract right and really ... I think people are beginning to think about this a little bit differently now, but very clearly specifying the contract around all of these different terms is really important to allow, number one, to make sure that this flourishes, but also, that you have access to the information, the data that you need.
Dan Bennett:
Yeah. I mean, to put that down to a consumer, human level, if you own an EV, I suggest you bury deep into the settings menu on your EV and find the "I agree" or "I accept" that says all of your driving and everything you're doing is being shared with the manufacturer of that EV. They're doing that because by gathering that data, they can help train models for self-driving. It may well be a condition of ownership of that vehicle that you do accept that that's shared. So that OEM point is a really, really important one. It's been going on for a while with really big equipment, like engines on jet aircraft, but as the cost of cellular keeps dropping, as IOT chips keep dropping in price, it's becoming more and more that even valves and things like that can have these sensors on them, that we can be streaming the data of. So who owns the data is ... it's a really interesting question because you might feel that you've bought a piece of equipment but may not have any rights to the data coming off it.
Hill Vaden:
Yeah, so job opportunities for lawyers as we go
Dan Bennett:
Absolutely.
Hill Vaden:
... into this space. Watching the time, we try to finish with forward-looking comment, so we'll put both of you on the spot. What is a specific project or company or technology or regulation that we should watch out for over the next, say, six to 12 months that gives us some sort of indication of where things are moving directionally within the application of AI to the energy sector? Dan, you're looking more pensive right now, so I'll start with Jud because he's looking right back at me.
Jud Jacobs:
It's everybody. The conversations that we're having is that this is very much on the attention of all the senior executives, so everyone's moving forward. There are companies that are leading right now, there are company types that are leading, but I think there's not a single company that we've talked to who is not at least developing a plan or progressing that.
Until the last two years or so, it's been traditional AI. It's been much more machine learning. I think really what we're beginning to see in the last 12 months is that a lot of the use cases around generative AI are beginning to emerge. Watching that space I think will be ... I've asked a question to a lot of our clients, "Is this taking all the oxygen?" They're, "No, don't think of it that way. These are very complementary. A lot of the things that we're doing through generative AI are kind of advancing our other AI capabilities as well. So these two things are very complementary," but what I'll be watching ... We're kind of in the hard work phase right now, so what I'll be watching are the use cases that begin to come out and some real solid documentation of some of the benefits that it's delivering to the organization as well.
Dan Bennett:
Yeah. For me, we've the EU move quickly on regulation around this. Some would say too quickly. I think a lot depends on what happens in the election coming up, at the end of this year, here in the U.S. But I think we're heading down a road where there's going to be ... whether you're an energy company or any other sector, there's more disclosure coming. And a better understanding by regulators of how you're using AI, and what the risks are around it and how you're managing those risks. That is not dissimilar to what happened with Sarbanes-Oxley all those years ago. The same thing's going to happen is my sense of this.
Hill Vaden:
All right. Thank you both. I'm going to now read a AI-generated closing statement. So listen once again for the modifiers. This large language model is like a college freshman writing a paper with a word count.
Thank you for joining us on this episode of EnergyCents by S&P Global Commodity Insights. We hope you found our discussion on the fascinating intersection of energy and finance insightful and thought-provoking. To stay updated on the latest energy market trends analysis and future episodes of EnergyCents, be sure to like, follow and share our podcast. Your support is invaluable in helping us reach a wider audience and foster meaningful discussions within the industry. Remember, for more personalized solutions and in-depth market research, visit us at spglobal.com/commodityinsights. We look forward to connecting with you again soon. Until next time, keep exploring the captivating world of energy and finance.
Jud Jacobs:
That was awesome.
Hill Vaden:
All right. Thanks, guys.
Dan Bennett:
Thanks, Hill. Thanks, Sam.
Sam Humphreys:
Thank you.
Jud Jacobs:
Thank you.