Latest / Energy Voice – Out Loud / Road to Adipec 1: AI’s promise and pressure points
Transcript
- 0:43I'm Ed Reed from Energy Voice, and I'm delighted to welcome
- 0:46Dennis Joel, CEO of AIQ, and Francois Labarie, EVPAI at
- 0:51Cognac. Today we're going to be talking
- 0:55about how AI is reshaping energy and power.
- 0:59And I think there are a number of extraordinary projections I'm
- 1:02sure we've all seen around sort of new power demand from AI.
- 1:05But I think there's also this kind of question around
- 1:07potential. Isn't that there are there are
- 1:09new ways of looking at the world that are triggered by AI and
- 1:13there's a real tension there around, you know, so the energy
- 1:15demand and the potential for transformation and, and how
- 1:18these can come together to sort of, you know, deliver success.
- 1:20I think it's going to be one of those kind of big questions that
- 1:23we are obviously seeing a lot of now.
- 1:25And I'm sure we're playing out at Adipac too.
- 1:28So listen, chaps, let's let's, let's, let's dig into how to
- 1:31unpack the sort of both that sort of transformational sort of
- 1:34sense of, of, of AI and, and I suppose also how it's being
- 1:37transformed and therefore the impact on the future.
- 1:41Dennis, what do you think that AI needs to how?
- 1:45How does AI deliver change to energy systems?
- 1:48I, I believe AI needs to deliver value to change energy systems.
- 1:52I think that's the real question that everybody asks today,
- 1:55right? How, how do you create value
- 1:57with AI? And I think with the right
- 1:59values, we can change policies, infrastructure and upscale
- 2:03workers with whatever's needed to create that value.
- 2:06And that's really why AIQ was foreign.
- 2:09It was really to bring technology and industry experts
- 2:12together and really figure out how do we transform the energy
- 2:15sector, how do we take the current workflows that we've got
- 2:18and how do we make it more productive, efficient and more
- 2:23sustainable and safer for for the world.
- 2:25Francois, there's a kind of a obviously comes to the heart of
- 2:28it value. How do you define sort of value
- 2:30for AI? Yeah.
- 2:32And, and I think that's a, that's a great lens to look at
- 2:36the AI opportunities or the promise, the value, what value
- 2:42is he going to provide? And then, as Dennis says, then
- 2:44the rest can be, can really be worked out.
- 2:48The, the, if you think about what AI has been delivering so
- 2:52far, because AI has been there for many decades.
- 2:55And, and the last AI wave brought us technologies like
- 2:59machine learning, which is today at the heart of any search, any
- 3:03social, but also a lot of industry already.
- 3:05And then, and then it's an AIQ use it for drilling
- 3:08optimization. It's been used in prediction of
- 3:11failures in, in, in the lot of the industry.
- 3:14So, so it's already part AI is already part of a lot of work
- 3:18flows in the industry and it's already bringing value in
- 3:21industry. What we see happening now is
- 3:24with generative AI is the promise to radically lower the
- 3:28barrier between humans and the machine.
- 3:32Because all of a sudden the AI that we have available can
- 3:36interact in the natural language, but also can make
- 3:39inferences, can start thinking and looking at vast volume of
- 3:42data and finding patterns that were not available to previous
- 3:45technology waves. And so, so the type of value
- 3:47that we've been unlocking for now has been very powerful, but
- 3:52in a way very, very specific and very technical.
- 3:55What we see with a new wave of AI with Gen.
- 3:57AI is that it, it has, it holds the promise of really being
- 4:01all-encompassing and, and lowering the barrier to access
- 4:06very complex calculations for all workers in any company.
- 4:10So, so by saying that going back to the, the question of value
- 4:14from them is effectively the value is, could be tapped in any
- 4:17parts of the way we operate in industry.
- 4:20So the value potential is massive.
- 4:24I mean, I think I think that there is a there is a kind of a
- 4:26really interesting question. So to sort that, but but maybe
- 4:28just to kind of come back to that kind of question in machine
- 4:30learning. I remember, you know, we we
- 4:32we've had discussion of machine learning in the past.
- 4:34I've been a you know, out of X in the past when you know,
- 4:36people have been talking very exciting about machine learning.
- 4:39What is the difference between machine learning and and and AII
- 4:42feel? I feel like you know, there
- 4:43there is the the discussion has moved on, but I feel like I've
- 4:46I've I've maybe got stuck in the past.
- 4:49That's a that's a fair question. So, so machine learning has been
- 4:53in a, in a way, very advanced statistics, looking at large
- 4:57volumes of, of data, again, large volumes of time series and
- 5:00training the models on the patterns it's seen to predict
- 5:04what will be coming next. You could think of what we're
- 5:06seeing now with the the generative AI as being the same,
- 5:11if you want the same continuity of thinking, except that it's
- 5:13now able to handle a lot more variability, like for example,
- 5:18language and a lot more context. So the the amount of variability
- 5:22and the amount of context the AI is now capable of handling is
- 5:27many, many folds, hundreds of thousands is not million folds
- 5:31more complex. So it can handle not just
- 5:33sequences of time series and numbers like it used to, but now
- 5:36we can, it can handle things like language, which is
- 5:39extremely complex because there's so much context in the
- 5:42way we build sentences, in the way we structure any discussion.
- 5:46That's the models are now capable of understanding this
- 5:50high level of variability and making much and helping you
- 5:54build and predict what's going to happen on this high highly
- 5:58viable scenario. So, so now the fact that you can
- 6:01talk to a machine and the machine will be answering with
- 6:04coherent thoughts based on the context you've been providing to
- 6:07it was completely impossible just a few years ago.
- 6:10So, so the algorithms have become a lot more complex.
- 6:12The way we've been able to train them, the volume of training
- 6:15we've been able to give them is a lot more complex.
- 6:17So that now you can, for example, ask a machine to look
- 6:21at hundreds of thousands, millions of files and say, hey,
- 6:25find context in it. Tell me all, all the work
- 6:28orders. I, I just give you the haystack
- 6:30on all these work orders. Find me what what are currently
- 6:34the most critical equipments that will be that will be
- 6:37failing And so the machine will be able to understand, hey, this
- 6:40is this is about a machine, this is about failures.
- 6:43And the documents may not even talk about failures, but they
- 6:46will be the machine will be able to infer a failure.
- 6:49They will be able to infer is this critical or not and give
- 6:52you an answer over millions of of equipments.
- 6:54This was completely impossible just a few years ago.
- 6:57Dennis yeah, I mean, I suppose, you know, it's kind of building
- 6:58on that kind of thing. Give me an example of of how
- 7:01you're using using AI, right. You know, obviously this kind of
- 7:03progression from from machine learning.
- 7:05What does what does AI do for a IQ?
- 7:07Well, like I think to continue on what Francois is saying,
- 7:09right, I mean, I think trying to solve the traditional problems
- 7:12with the traditional ways, whether it was machine learning,
- 7:15it's been around for quite some time.
- 7:16What AI is able to do is take that larger context.
- 7:19We have, you know, and Q has taken petabytes of, of our
- 7:22customers data and been able to take the subject matter experts
- 7:27and, and really redefine not only the roles within the
- 7:30industry, but the roles within our company to actually take,
- 7:34you know, 2 subject matter experts that used to be working
- 7:37in silos and be able to pull those, those things together.
- 7:41So whether that's, you know, the HSC, you know, inputs together
- 7:46with the operational folks to make sure that we keep people
- 7:49safe or whether it's to predict equipment failure so that we can
- 7:53connect it to the supply chain, be able to, to reduce costs
- 7:57around it. It's really to take, you know,
- 7:59different disciplines and different roles within our
- 8:02customers, organizations and roles within our industry and
- 8:05really reshape on how we tackle those work flows and those
- 8:09problems with the larger context and be able to take data from
- 8:13all those different sources and be able to change that.
- 8:19This year at Adipek, there were due to be more than 205,000
- 8:22attendees, with 2250 exhibiting companies.
- 8:26They're going to gather in Abu Dhabi to exchange ideas, forge
- 8:29partnerships and mobilize investment, showcasing solutions
- 8:33that deliver energy to more people, more affordably and with
- 8:36lower carbon intensity. Across the Engine 80 sessions,
- 8:39over 1800 expert speakers will explore how the convergence of
- 8:42energy, AI, investment and emerging economies is reshaping
- 8:46markets and driving opportunity worldwide.
- 8:51So Dennis, when sort of thinking about that kind of idea and kind
- 8:54of delivering change around delivering value, what do you
- 8:58find it to be the most challenging area, right?
- 9:00I think obviously kind of looking kind of globally, there
- 9:03are various, you know, difficulties, you know,
- 9:05policies, infrastructure workers, what for you is keeps
- 9:09you up at night? Well, I think, I think again,
- 9:11going back to that value creation, right?
- 9:13I mean, that really is, I mean, whether it's, you know, the
- 9:15comment, the question you asked, what's, how's it different than
- 9:18machine learning or how's it different than what we today or,
- 9:22you know, a individual saying, well, I already do that.
- 9:27I think that, you know, changing the mindset and really the
- 9:30change management and culture and understanding that we have
- 9:34to attack these problems a different way is, is probably
- 9:38the most challenging part around getting our AI solutions within,
- 9:43within any sector. And once we can embrace the idea
- 9:48of that we have to tackle problems differently, I think is
- 9:51really when you get those aha moments and people start to
- 9:54embrace AI, the the mistrust of it or, or what not.
- 9:58So I think it really the the change management and just how
- 10:01we attack problems today and how we can do that differently with
- 10:05this new, you know, technology and innovation is really the the
- 10:09hardest part to to overcome. First of all, what, what, what
- 10:13do you say in terms of the challenge for AI?
- 10:16I think the threat of value is, is certainly where we need to
- 10:19focus, but we have seen a challenge in, in the adoption
- 10:23overall, because ultimately there's been a fear of AI as
- 10:29much as there's been, there's been an excitement over the
- 10:33promise of what he could deliver.
- 10:35So, so the, the discussion with the subject matter experts, with
- 10:38the people on the floor that will be ultimately the ones
- 10:41benefiting from AI has been a very interesting conversation.
- 10:47So you could think about it in terms of change management, but
- 10:50also a lot about it is as we are bringing a technology that is
- 10:56moving, I like to say that you're always too late and too
- 10:59early with Gen. AI.
- 11:01You're always too late because it moves.
- 11:03The technology has been moving incredibly fast.
- 11:05One could argue the pace is now slowing down a little bit, but
- 11:08it's still, we're seeing new models coming up, new
- 11:10applications coming up every day.
- 11:12And at the same time, you're too early because you ultimately
- 11:16applying this technology will take infrastructures and
- 11:19policies and change management and, and, and so, so we are at
- 11:23that, at that absolute frontier when it comes to taking these
- 11:27technologies and putting it in operation in energy, for
- 11:31example. So in large industrial context,
- 11:34how we look at overcoming things, there's, there's a,
- 11:36there's a couple of approaches. The first one is when you think
- 11:39about, OK, what is this technology going to be very good
- 11:42at ultimately. And the technology today, if you
- 11:45talk about Jenny I, it's very good at just like the example I
- 11:48took at looking at very large volumes of data and provided a
- 11:53specific context to help you find some needles in a haystack.
- 11:56So searching for information on something that would be
- 11:59otherwise extremely tedious or summarizing something like on
- 12:03the large volume of data, you can think about shift handovers,
- 12:06right? Every operation has shifts and
- 12:08operations and then the new shift comes in and they need to
- 12:12know what has happened. What should I know about the
- 12:13previous shift so that I can perform my current operations
- 12:16well, right. And then AI would be very good
- 12:18at that on something that people tend to not really like doing,
- 12:22writing summaries and taking all the information.
- 12:24So, so if you find this, these, these elements in the workflows
- 12:29where the AI is going to be a very low hanging fruits and very
- 12:32effective. This is where you're going to
- 12:34start to change the minds and hearts of the, of the, of the
- 12:38people on the ground. And, and we've seen that again
- 12:41and again on root cause analysis, troubleshooting, shift
- 12:44handovers. So things that are really part
- 12:46of the current workflow, which have been so far impossible to
- 12:49tackle with AI. The main challenge we see is an
- 12:52infrastructure problem, but it's not the, the, the way you would
- 12:55think about it. It's more of a data
- 12:56infrastructure problem. How do I avoid that?
- 12:59We provide hallucination because now, so I'm asking people to
- 13:02change the way they work. I need to make sure that they
- 13:05will make operational decisions that these are based in grounds
- 13:07on facts and actual data, right. So this is this is more the
- 13:11challenge we're seeing. Yeah.
- 13:12And I think to continue on to what Francois is saying, it's
- 13:14also, I mean, an AIQ, you know, having seen those issues and
- 13:19having not only developed, scaled and deployed some of our
- 13:22products. I mean, some of the things that
- 13:24we figured out also is that, you know, this whole, we know, we,
- 13:27we focus on AI LED digitalization as opposed to
- 13:29digitalization LED AI. The idea of having too much data
- 13:34is also overwhelming for, for people to get over.
- 13:37So what we, what we've done is, is gone out there and actually
- 13:42have AI legislation where we actually only digitalize the,
- 13:46you know, the, the parts of the industry that we can actually
- 13:50change through AI very quickly, while visualization, you know,
- 13:54from a broader scale will catch up.
- 13:56So we we try to make an impact as quick as we can with AI to be
- 14:00able to change the minds and and parts open.
- 14:05Looking at that kind of question around value and I think looking
- 14:08around that kind of question around people, I feel like
- 14:10there's a kind of the challenge is kind of about trust, isn't
- 14:12it? It's about how you put your
- 14:13trust in something. And I think, you know, Francois
- 14:15mentioned sort of hallucinations and I, I mean, we, we've all
- 14:18sort of seen the instances of, of, you know, the sort of things
- 14:21on the Internet of, you know, people with the pictures with,
- 14:24of people with too many fingers and things like that.
- 14:27That's how, how do you, how do you solve that problem of trust?
- 14:29How do you how do you ensure that the answers that the AI is
- 14:34giving you are trustworthy? Well, so again, I think we're
- 14:41where we're, where we're, where we were, you know, where our
- 14:43approach was, was different is that we were able to not only
- 14:46use physics models and, and to prove our models, but we were
- 14:49able to use petabytes of our customers data to actually train
- 14:52our models. And by training our models with
- 14:54not only physical models, but but also real time data, we're
- 14:59we're able to, to prove that our models worked.
- 15:03And not only that, even then, we still had 12 months worth of
- 15:08pilots and testing and, you know, verification on that data
- 15:12and that those models. And we were now, and we've been
- 15:15able to now deploy, you know, our AI solutions across our
- 15:20customers field. I mean, we're in 500 of their
- 15:22wells today. We're on 20% of their reservoirs
- 15:25and we're on 100% of their drilling rigs with our vision
- 15:28analytics. So, you know, we've been able
- 15:30to, to, to, to continuously fine tune these models and, and
- 15:34really not only talk about AI, but, but also have proven, you
- 15:38know, results through our, our products.
- 15:40And first of all, what, what, what do you think about that
- 15:42sort of question around sort of trustworthiness?
- 15:45Yes. So trust comes at many, many
- 15:48levels. You have the, the trust that the
- 15:53information that I see is from a source that I can trace so that
- 15:57I understand how the decision has been made.
- 15:59It's also the trust. That's the what the answer.
- 16:02I see something I'm allowed to see and that's it will stay, it
- 16:06will stay proprietary to me. And then of course, there's,
- 16:09there's the, the, the fact that I understand how whatever answer
- 16:13or proposal I'm receiving is something that I can understand
- 16:17like I can, I can ask what has the rationale that is that has
- 16:20LED you to get there? So, so there's many levels of
- 16:22trust implied in, in an interaction with a machine or an
- 16:26algorithm. And, and we know how to build
- 16:30now agents out of this. So we we, we take that AI and we
- 16:34encapsulated with all the logic, which may be physics models, We
- 16:37encapsulated with prompts, we encapsulated with guardrails and
- 16:41and, and we form agents that are able give it for a specific task
- 16:45to be trustworthy, meaning that's through an interaction
- 16:49with with one of these AI agents, not I can also ask where
- 16:53does the information come from? And I can have a full
- 16:54traceability. I can ask, how did you get to
- 16:57that result? And it would give me the
- 16:58reasoning and, and the way we do that technically is that we
- 17:01anchoring these into the customer, the, the customer
- 17:05truth, if you want. So we have rebuilt a digital
- 17:09representation of the way the customer operates, the way the
- 17:11data is linked, the way that the data comes from.
- 17:14And so we use these agents on top of that, so that we are
- 17:17anchoring the agents in the reality of the customer.
- 17:22So, so that's, that's the way you can, when you hear people
- 17:24talking about agents, this is one of the reasons why people
- 17:28get excited about agents in the context of AI, because you can
- 17:31then anchor them with guard rails in the reality of the
- 17:34customer. And, and that's what we see very
- 17:37successful, being very successful and not just in
- 17:39energy. I mean we've been, we've been
- 17:41having one new customer a week since April, right, deploying
- 17:45this type of technology in industries from pharma to power,
- 17:48energy, process manufacturing. But that's the pattern that we
- 17:52see emerging. This is this is the way you will
- 17:55get trust. And I suppose you know, and sort
- 17:57of, you know, you talk about some new customers and, and I
- 18:00think, you know, when I speak to to companies, quite often they
- 18:03feel like there is the sense of excitement and there is a sense
- 18:06of sort of opportunity and potential.
- 18:08But then there's also sort of a sense of people aren't quite
- 18:10sure what the opportunities are, right?
- 18:12They're not sure. They didn't see like, how do we
- 18:15get from where we are? How do we change what we're
- 18:17doing to to, you know, where we could be right?
- 18:19What do you see? What?
- 18:21What do you see the opportunities as being?
- 18:23A lot has been known in the back office today with the with the
- 18:26deployment of the first AIS. The real opportunity for me lays
- 18:30in the operation. This is where the heart of
- 18:32industry and energy is being produced, consumed.
- 18:35So this is these optimization and, and, and there's, if you
- 18:38think about it, there's the opportunity is multiple, but a
- 18:42lot of it lies in the fact that we as humans can only digest
- 18:46certain amount of information at the time.
- 18:48We use quite a bit of time to look for information.
- 18:51We are stopped by our own abilities, for example, to use
- 18:54complex simulators that Dennis was mentioning this physics
- 18:56simulator extremely powerful. Very few people are effectively
- 18:59able to use them. And what AI does is that it
- 19:02completely opens the window, right?
- 19:04It's, it's, it's allowing somebody to have access to a lot
- 19:08more data and to draw inferences from a lot more data so that you
- 19:12all of a sudden you can be aware of what supply chain is doing or
- 19:16what type of material you have available or equipment you have
- 19:19spare parts you have available. Or you can say, hey, why don't
- 19:22you run a few simulations? If I'm optimizing today, I'm
- 19:25asked to optimize for quality of outputs.
- 19:28What should I do? What are my options today?
- 19:30And you don't need to know that behind the scene.
- 19:32It's going to have to run a couple of physics simulators,
- 19:34run some neural networks and to give you an answer, right?
- 19:37All of a sudden, all of these things get possible for anybody
- 19:41in the operation floor. And that's why people get
- 19:43excited. It's it's not about one single
- 19:46AI agent is going to run the factory.
- 19:48It's more like all of a sudden, all everybody can have
- 19:51superpowers, right? And and that's this is where you
- 19:54make a difference. Dennis, do do Do you recognize
- 19:56that about giving everyone superpowers?
- 19:59Yep. Absolutely.
- 20:00And not only that, you know, with collaboration with the
- 20:03folks like Cognite that help us do what we, we want to do.
- 20:06And that's, you know, to, you know, when we look out for
- 20:09solution, we look at solutions that's going to make, you know,
- 20:12which either going to help our, our customers produce more,
- 20:16produce more efficiently, sustainable or safer, right?
- 20:19I mean, we focus up to what Francois says to the, the
- 20:22operation itself. And that's where the heart of
- 20:25transformation is going to be. And we use the data that's
- 20:28available or make data available to actually make decisions and
- 20:34what AI produce a a more effective energy system.
- 20:39I suppose the, the, the other kind of question that always
- 20:42comes up about AI isn't, it is about power demand.
- 20:44And I feel like any kind of question, you know, and, and,
- 20:47and it always, it's always a slightly strange kind of
- 20:50conversation because it always feels a little bit like people
- 20:52who produce power are suddenly faced with the prospect of
- 20:55selling more power. And this sometimes feels like a
- 20:58terrible thing for them to have to think about.
- 21:00But how serious do you think those kind of new power demands
- 21:03are from from, from AI, Dennis? Is this something we should be
- 21:06worrying about? Well, it's very serious, right?
- 21:09I mean, especially for building foundational models.
- 21:12And today, the challenge to meet those demands are real.
- 21:16And we'll take the collaboration between the energy sector and
- 21:18the technology sector to solve them.
- 21:20At IQ, we see AI as the key enabler to transform the
- 21:23traditional approach of delivering energy systems for
- 21:26not only AI, but to our everyday life.
- 21:29Some fun facts around this is that, you know, I heard in 2035,
- 21:33there'll be almost 90 billion people on this, on this planet
- 21:36in line with the growth. I mean, oil demand will increase
- 21:40from 103 to about 109 million barrels per day.
- 21:44And that's going to, you know, LNG and chemicals will expand by
- 21:46over 40% until electricity will actually demand will surge from
- 21:519000 gigawatts to about 15,000 gigawatts, which is staggering,
- 21:54which is a 70% increase. So yes, I mean, at the end of
- 21:57the day, I think we're going to need more LNG, we're going to
- 21:59need more low carbon oil, we're going to need nuclear and you
- 22:03know all you know, and even more commercially viable renewables
- 22:06to meet all of this demand for that AI is, is, is driving.
- 22:10And, and first of all, I suppose, you know, I feel like
- 22:12there's this kind of push and pull, isn't there?
- 22:14There's a kind of question around, you know, how much
- 22:16energy does AI need? But also this question around
- 22:19does does AI actually help deliver that transition?
- 22:22Does it does it provide some of those efficiencies that you've
- 22:25been talking about and actually make operations better?
- 22:28I mean, AI has already been proven to help grid resilience
- 22:32to help accelerating these interconnects because there is a
- 22:35lot of gas and renewable that has been coming online around
- 22:39the globe. And, and we have we have seen
- 22:43our customers using AI to accelerate.
- 22:47I mean, we're talking so, so one of the challenges is we're
- 22:50talking about construction of all of these, all of these
- 22:52energy that will be required, right.
- 22:55But one of the, there are two big challenges that that most of
- 22:58the countries are facing. One is the speed at which the
- 23:02grid operators can manage the request.
- 23:04OK. There's a new the new asset
- 23:06coming online with production of a couple of gigawatts.
- 23:09How am I going to connect it to the rest of my grid?
- 23:11What's going to be my balance and how do I need to reshuffle?
- 23:14And producing this authorization can take several months.
- 23:17It actually very often delays projects.
- 23:19It's not the actual construction that delays project.
- 23:21It's getting these authorizations and what we're
- 23:23seeing is running all of these predictive models, something
- 23:27that technology can do, right? So, so we've seen in in Norway
- 23:30here and, and, and the, the, the, this has been publicly
- 23:33released, right? We've seen a radical short
- 23:35shortening of the time from a request to connect a new source
- 23:40of energy to the grids to the acceptance from, from several
- 23:43months to years to actually a few weeks, right?
- 23:45So, so very, the ability for us to put more of that power
- 23:51generation available to the grids has been dramatic.
- 23:54The other one is the resilience of the grids.
- 23:56So we have an aging infrastructure in a lot of
- 23:58countries, not everywhere, but in a lot of countries.
- 24:00How do you make sure that the grids will be able to actually
- 24:04rise up to this challenge of increased demand?
- 24:06And a lot of it is actually being able to predict where the
- 24:10demand is going to be, where the consumption is going to be
- 24:12unable to manage your assets, your physical assets and your,
- 24:16your transformation stations, but also managing the grid
- 24:19efficiently. And, and this is something that
- 24:21more and more we see an AI layer coming and, and, and so that you
- 24:26can direct a lot more efficiency, your production and
- 24:29your consumption patterns. And, and this is, this is also
- 24:32helping quite a bit flattening a little bit that, that curve and
- 24:37that demand on the infrastructure as we are
- 24:39building. I mean, there's no way around
- 24:40it. We will need to build more on
- 24:41the generation and on the, on the grid itself.
- 24:45So yes, AI has to be part of the solution as well if we want to
- 24:48meet the demand that we see, right?
- 24:50I think that's a, that's a, that's a great note to end on.
- 24:52So listen, I'd just like to say thank you to Dennis and
- 24:54Francois. Some really interesting ideas
- 24:56there around around AI and that sort of transformation
- 24:58potential, but also around, around that kind of idea of, of
- 25:02change management. How do we, how do we deliver
- 25:03change? How do we talk about it?
- 25:05How do we make it happen? I'm sure these topics will be
- 25:07coming up at effect and I'm looking forward to talking about
- 25:09it there. Thank you very much, gentlemen.
- 25:12Albin, Ed. Reed.