Cognite-gründer og Aker BP om AI-utviklingen, investeringene og hvordan de holder teknologien i ørene

[Musik] Godfredag og velkommen til ekonominietene, Alsa. Men 2 av denne gucciolovedagen, Shwanderer og Gustess. Nartel går ført med amerikanskisen til Abangchefen. Som i omtattet taler i Jackson Hall, var jeg om ingenting å klokke en fire norsk tin. Så skal vi et lit innblikke utrullingen av A som ser norsk industri, vi har nemlig fått besøke av poladeulig norsk. Nå er jordselig digitaliseringstirektør i AKBP, og steg en annen delsten med en grønlig i Kognite. Men først en titt på nøyetene som præger dag en ellers, for et dag går ikke eller. Så vil man gøy en først rettes mot wargomming. Og derret er vårsinkt enn for vi sier for etter at Jon Powell skal ta det å klokke en seisen norsk tid. Så har Jon Powell, som har en norsk tid, så har Jon Powell, som har en norsk tid. Han kommer på banen til å rette med dagen og kvøkring, så vi kan få venten om kommentarer. Om den central bank tansdelere føler vi med på både dollaren for vi sier. Og også de amerikanskere endte nivåene for vi sier vi har en tjoring i USA som ligger at vi ikke berett over 4-3 i dag. Det er ikke bare at han har de tjøp med at rentekomitene medlemmingsakuk, begynn trekker seg fra Fed Robert Kerv den nukkel, men han har vært missfølgnet med mange på rentekuts generelt. Og var Jon Powell sier om inflationen, var han eventuellt videre sier om sjansen for et rentekut i september. Og arbeidsmarkede generelt ville jo bli fullt med argusøyne for vi sier. Pfei, noe eller så kan du lesse om hvorfor en landets ledende skatte av vakater, menen i en stått mørk skatte, lotto, forsøker i sekkenk av det. Med et arbeidsfrødrag kan være grundlåvsstride, eller så er det vært de merke seg dagens industri i dag. Og det får en chef Nikolaj Tangens i nemlig til den sønska visensittat, at det er min värste krisen noen sinne om det har den stående batten rundt fondets indesteringer i Israel. Vi må titte litt til gran på markedele, så vi er en hove nækse i dag som ligger opp rundt kvartbrotenten med noldepris som snuset, bynner oss noen på 68-då i fattet 67-76. Vi ligger på noe Europa har stort sett løftet opp i et positivt linne börsmøsjøsset startet litt ned fra starten i dag. Grønt oss i noiden er det også pentagrent. Vi har også noen 2020 på åldstrikt som ligger i plus i dag et et rattvis så teknologi sektoren. Vi har sett ordene i litt underpresta, noen gang i en godel. Der verket om merke seg, har du lønnes ligger pent på topp som vi kan se på skjermen 5,2 % i plus på sin kvartalstal, eller så domineres listene fra en kvindor og doffnorskittanium. Også ikke helt overraskenet med på mest omstattlisten i dag, hente du då i en nesten 200 miljoner i en ny ambition for vi tar med Elshas rek. Elshas rek aktolænders ute med tal i dag. Også var vi sida på Eiffinø, analitikerne har længere vært imponerter over hög og utbyte politikk. Der i Alshas gofret alt av overskudtillaktionerne med lavinvesteringer i ny flotte. Og mange har det omtalt hög og utbyte kvartolænders som best i en klas som de heter. Det er forst å slenge på at denne karnegen negraderer i et elskjeden kyd for å kjøpe det hållet til Gorsdagens kvartalsrapport kutter også kursmålet litt i grand der med 5 kroner. Så vi lasker i forstig farvel til Avans Gassidag som mange har risket, for i stannkallerklæver på peker, så blir avsjønnerne der. Neste største er i det bevel pegget til flotte delen i går over veldig pegget og godt som en kule, men avviklingen av avans servertfall i dag. Vi skal få en dagens jesterretter i korte klampässe, vi er tilbake rett og rett. Vi lager med er fra oss e-fra TV. Og så handler om det sommer. Det er pengene går til. Vilke bruken skikke på skrivelse, når er utgivt, når man også rasker en intekt, når fra 20-tradve? Vi sender å dønne rundt til din Apple TV, der du kan føle Börsmarne, økonomien i etterne, rende, möte, pressekonfranser, bil, tester og du får stopp fra verden rundt, inkludert de klømst siste utspil. Jeg har sett å lide deg samt med det, hvis noen av det er noe som pristisjon. Søk opp finansavisen og klik på EFA direkte. Få med deg i finansforum, et helt nytt program fra finansavisen som du får både på podcast og videover yke. Her går jeg i penediktet Storren Bamvik og aksekommentator Karl Johan Malnäs, in i dydden på en rekke tema inn for finansmarkedene som dere lurer på. Hva skjer noen i med store kjorte her akse? Hvorfor reagerer Börsen så stett på rent endringer? Hvordan påvirker er oljeprisen alt ant i marketene? I finansforum bryter vi ned mytter, diskuterer og får klarer mekanisme i market. Ikke minst svara på spørsmålenen fra vår elyttere å se dere. Har du spørsmål eller er det noen tema du gruble rover? Ikke nøl med å sende din til TV-tips et finansavisen.nom kan netto of det bli tema for neste episode. Da skal jeg si velkommen til dagens tvester oss for at vi totinger, så kommer vi å rige går. Vi var taget i jesterdag til fallet og i dag. Kif digitalisation officer i NKBP, enstant enda i Danielsen, en av de funder av kognite. Lange til marketene, velkommen. Velkommen til. Lige time, vi har å checker de arkæris. I 23 januar, Alkabepi, er en av de første company i Norway, som skjer i Microsoften. Fære å gjøre system ko-pilot fra AI. Ja, vi har å få en industriale av AI. Vi har å ta seg det. Vi var å ta seg om transkriving-medtingsen av AI, så vi gikk av en sommer. Jeg har en lille konfus på, hva de har med å ta medtingsmedtingsmedt. Hva har investet i de systemen av de første? Hva har du? Det har vært så fort til å være i et tjernet, hvor det ikke er en av det som skjer i klonke. Vi er å ta det, og vi er å ta det, og vi ser det før i alle. Vi har å ta all denne employer og Alkabepi, og vi har en av de bedre av at vi har et tjernet i seg. Vi er så fint, at vi er så fint. Det er en av de coolt som jeg har. Jeg har noen av det, men jeg har noen interpretasjon i mye mesning. Så hva jeg har set i en team-smeeting og så møttelig i regionet, har jeg en av det AI-trendelige for meg i engal. Jeg har en av de voys. Jeg har en av de flere av de flere av de flere av de engal. Det er faktisk en av de flere av de engal. Det er en av de flere av de engal. domainsen. 或者 en Gmail corianderseuf-competitor - rollenovers�ler, ogviolene også mobler det - einer dem en større överstående henne. Ellber mener å melse ve metadata. Dette syns skulle en forleik av det og hjelper det. Jeg er ikke så her i dag, og jeg er forstået prioritiser meg i en bak sommer og e-mail, - - draftet respanser og gett sommer i dokumentet og transkriptjonen. I en av de flere sommer som jeg har sett, - - jeg kan ikke trypple og kjører meg til de flere sommer sommer. Jeg kan ikke. Jeg kan sende meg koppilet. Jeg er i lising, og jeg er all den info. Det er godt. Men at du er på en måte, - - er det vi ikke kan ta. Men det er det i data. Så når det er at man har et fikkstata-set, - - eller at man har spokert eller at man har sett, - - så er det mer akkurat. - Det er forkontained. Det er just forkontained, og det er forkontaktet for råbysk. Er det du løp, for hvis vi revalder for noe kognite, - - er det en av de founder som. Jeg er trynet til å build systemen som skjer data til å optimise - - at ølfield, par planen, - - du har en gang med åle companyen, industrial companyen og customen. Absolut. Hva er det som du har lyst til å optimise i systemen? Hva er AI-kondet for å optimise - - så vi har kunstret for noen som er optimise i productionen. Jeg tror vi har ganske hvor vi har drivnet vi kan gett. Vi har ganske, og vi har ganske, - - vi har en AI-spertise i industry, - - og vi har ganske, vi har ganske, - - så vi har ganske, og vi har ganske, - - men vi har ganske til å gette all of the data først. Vi har ganske til å gette all of the current state of industrial facilities - - og ganske til å gette all of that. Og så ganske ganske til å gette all of the current state of industrial facilities - - og vi har ganske til å gette all of the current state of industrial facilities. Så det er en match, som vi har ganske til å gette all of the current state - - har ganske, veldig kjøret, veldig kjøret, veldig kjøret, veldig kjøret, veldig kjøret, - - er en fantastisk match. Så jeg tror vi har ganske til å gette all of the current state of industrial facilities. Og det er en større roll, og det er en av de regner som vi har ganske til å gette all of the current state of industrial facilities. compressors at platforms and maintenance issues and - Absolutely, so Paul can detail these cases, but we just say, "Do my job for me." He'll like find the root cause of this, help me set up an inspection program. You just tell it to do your job, and then it comes up with these plans to do the job. And it has access to all this information, I can do all these kinds of analysis, even run simulators and you can do everything. So, and like I said, we work with now more than 130 different largest companies in the world. We have always been very heavy in oil and gas, and we have every single supermaider almost on our customer list, but we're also expanding into other verticals as well. So, now we are also powering the renewables, also the largest companies there with chemicals, cell and yeast and lots of other companies. And also we roll that to, you know, manufacturing, discrete and process manufacturing, and our very large companies, and this year we roll out to farm as well. And now we have almost every single large farm company on our customer list as well. And many thanks to AI for these companies. We just, it just applies everywhere. - So, not only will we hit our homes with the help of AI, but get our OCEMPIC as well maybe. But Paul, talk about as a company cause you're rolling out now, the Cognite Atlas AI, they call it. But do you use these systems? One thing is optimizing existing production, which I mean these are advanced facilities with lots of moving parts, but do you use it also in exploration, looking at geological data? - Yeah, we're using AI across the board. So there's not a single business unit in OCEMP that isn't using AI. And I think what's really, really interesting, and I'd like to talk about it a little bit cause you just touched on it. So in the recalls analysis case, so recalls analysis is something we do very frequently in industry, so a piece of equipment fails, and then you try to figure out what exactly caused that failure. And this is where we started with Atlas AI, and it's just like an incredible story because when we tried it first, we were like, okay, we'll see you, we'll test it out, see how it goes, and what we actually saw was that we could get efficiency in some of the branches of 96%. Like genuinely, 96%. So it's kind of, for me, I've worked in technology for decades now. It's one of, this one just keeps pleasantly surprising me. So now we're rolling this out across the board. We've had the subject matter experts involved. They're super engaged. Everyone's really excited about it. So what we can do now with our industrial data with Atlas AI on top, it's just phenomenal. So this is very much a part of our AI first strategy. - But these models, they have a built-in confidence in them, right? They very rarely acknowledge any mistakes or uncertainties. Like I said, obviously, if you search the whole worldwide web, you get a lot of answers. But even if you box in the data like you do with one facility or a set of facilities, like you have an Ocarby P, how do you make sure they don't suggest the solution that could potentially trigger a whole system malfunction or something, an explosion or something at one of these facilities? - So there's lots of different ways to work around that. So first of, we only have, it's work on the data that we know is facts. So it can't really come up with anything outside of that. Second, there is always like a domain expert in the loop. Like this is a tool for him or her. So they can be much more efficient than they don't have to do the legwork. Like they don't have to capture all this information from all these different systems. That's already in going on. And they could just ask natural language questions. So this is a key thing for like domain experts because they don't know programming, they don't know AI, but they could just ask this question. And then whenever you get the results, it has references always. You say you can increase the pressure because the max pressure of this system is that? Go here, you can see this in the data sheets. Where exactly where that is specified. So there's nothing that you say, which you can't also verify. - Well, here AI, often also here about large, large investments, right? Especially from the big hyperscalers, but talk about how our KBP has to sort of invest in the infrastructure. One thing is buying the software. But you have new fields like Eucydrosia, which is coming up and you just made a big discovery this week, but you also have older fields like WALHAL, SCOI, which were obviously conceived way, way, way before we even had the internet, I think. - Yeah, yeah, right. - True. - Do you have to roll out 5G, a lot of sort of ground infrastructure? - So we, and connect all these different sensors or exchange them for ones that can be connected to the outside world? - I think we're really fortunate in this sense. So we have connectivity on all of our platforms, but we've also been working with Cognite for quite a while. So in the same way that it was Christmas Eve for Cognite when this kind of blew up, it also was for us because we had made the investments on infrastructure on the data foundation, in Cognite, of Fusion. So we were AI ready. So it was just a real box of treats for us to be able to do. - You've had to do a lot of upgrades and infrastructure investments before. - But it's already been done as part of other things, like as part of having control rooms on shore, on parts of like getting just getting data to shore for analysis. So it was already there. So it wasn't done specifically for AI. Now we're just able to build on top of what we've already had in place, which is great. And of course our fields are very highly sensorized as well. So we have a lot, we have a lot of data. - When you, if you get a customer whether it's our curse or someone else, what are the sort of prerequisites that need to be in place to be able to use this? So he will now enter a contract with Telenoord to have a private 5G networks at a lot of their facilities in Norway. I saw something that building for your competitor or their competitor, Voj and Nagy, 5G private network offshore at Jules Tun, I was just renovated. - Yep. - Do you need to have fiber and sort of server farms and all these things very close to these facilities in order to make this work efficiently? - So of course we would prefer that. But it's not necessarily required. So I think if the only thing that's required is the willingness to change and try something else. So a willingness to start to try to optimize processes and this goes from the top to the bottom. - Have you been, I guess, I don't know if you've been following the discussion that we see sort of globally now with are we getting returns on these big investments? Is that a big discussion going on internally? - Yeah, it is. But I think for us, a challenge kind of when you just look at P&L, like in a short-term period, you're not actually capturing the value of the organizational knowledge. Like we're all in, we firmly believe that this is going to transform the way that we work and also transform who does work. But to get there, it's the organization that we need to move or the terminology moves much quicker than kind of the organization does. So in some ways, I think we don't put the right value on that. So just learning, even making mistakes and stuff that is also much valuable and that means that the next one you do is quicker, it's better than one after that again again and you start this flywheel. But yeah, I know absolutely we have to deliver value on it though. - But how do you go about that internally in an organization when you, I guess you have to go and ask them one for, can I spend X dollars on this, right? And often you get the question back, okay, what do we get in return for it? Is it difficult to quantify? - Well, I think the challenge with sort of expense payback. - I think the challenge with being very precise on, so it's different areas. So some of the cases are actually quite easy to quantify. Others are trickier because we don't know what we don't know. So of course you can put a business case together, but that, like there's so many unknowns in it that it's difficult to tell. So we just have to gauge it along the way as we run. What we're doing though is we're accelerating these projects. So we're really kind of boosting behind so that we can run through all the steps to actually understand where is it we fail. And typically it's not the technology. It's other things, right? Because you have this incredibly fast moving technology and huge potential coming up against a known way of doing things, right? So that's kind of the, that's the fun part actually. Like you said, this willingness to change. And we're looking at our MVP, we have a lot of willingness to change. So there's a lot of support for what we're doing here. And we're really getting into the core of the business. - But is it easier to look at a sort of facilities optimization program and say if we deploy the system, we'll increase up times from 96 to 98. - One, yes. - And then we're going to work to all your colleagues getting a little more productive because they have a new system for their email and their needs, right? - Yeah, that's very, so there's kind of proxies. - Because that's even less sort of-- - Proxies for that value, but it's very difficult to see it on the bottom line. But things that we're seeing, it's project compression time. So we can do project execution faster. Like early stage concept select compressing that by using AI production optimizations, using the throughput by using AI, these things are much more quantifiable, like increasing tool time, reducing maintenance costs, reducing shutdowns. Of course, all of these things are much more tangible. And that's unvaluable and that's why it's so great to move it from the office into the field. - Yes. - But of course to do that, we needed what you guys have done, which is a grounded fact base that we can build the models on. - But are you noticing a change in mentality amongst top management companies when everyone talks about AI in society in general, and then it's easier to get management to be able to look at these metrics. - I think it trickles down from the board saying, you have to do something, you have to do something, you have to do something, and then you have to do something. So we see this, we have deployed AI solutions now across the world, like you have customers everywhere. So we can see also that we always want to measure the value returns. So we see something like 25 to 40% benefits of running these systems towards whatever your goal was production, optimization or something like that. So this is very much in contrast to let's say reports that we've seen that a lot of AI projects fail because we see that a lot of the things that we do are not failing. But I think they always sort of working with the value and always is good but also you can't underestimate the value of the change. And also like if you have a comparison like most children will fall when they try to walk. But you don't say like oh no I don't think you should try to walk. It's not for you. You know you feel that it's good right? So you try that and you don't give up. You look maybe at somebody that makes it and you try to copy. We see a lot of cost cutting or these efficiency programs at big companies like Statskittoff or Stead. But also in the oil industry I mean 60 some odd dollars of barrel is not like a full party. It's okay but it's not a full party like 80 or 90 dollars a barrel. Are you noticing that giving a pushback on these investments as well or not only investments but definitely in the companies. But what we see is that higher resources contractors they are the ones that are pushed out under pressure. So not the regular staff isn't they're just more efficient and they can do much more of the work. That was pretty was the and contractors. I want to finish off with what's happening sort of around you in Norway because Occurian and then scale our building and data are now in light week. Where OpenAI looks like it's going to be a customer. Are you affected by this sort of in your cooperation or is this a sort of project that's happening on the side and then not really sort of observed. It's super super exciting. I think like for for our care for Enskil for OpenAI and VDN and also for Norway. I think it's really really cool. So guess the question is how can we really kind of maximize this this partnership like in Norway but also ourselves. So I think for us it's just really about exploring those ways to kind of bridge partnerships with like OpenAI and VDF for example. Yeah I think I was a to me I also like you were with AI all the time like a hundred percent of the time and when you do that you get a fear of losing it and so by having a regional sovereign data center with the same technology stack this sort of gives me a breathing room. You know we can run a BPE even though somebody would decide the US is not to export any AI compute anymore. So having a large sovereign data center is fantastic for the safety of over compute resources. Second it is opening a lot of doors for us. So I think you know we're proud of being in Norway region building a software company in AI and this also opens up for ways for us to build AI models that are industrial models. So again you know we can be AI much more. But do you either you feel like constrained by the availability of compute power? No. So far? No. Not us anyway. Not at the moment but you see that if let's say you needed to use it for for instance the defense or those type of things then you could potentially come up in a squeeze where you get and I think also maybe now 82% of all companies we know will start using AI and haven't most of them haven't yet so this will be a big demand. Interesting time stand-in for the oil in the ACO BPE in Kogne. We're going to have a short-term clamp so I'll be back in the next one. Will you be updated on the most important thing that has been done in financial life? That's it. Follow me for every year. I'm Marie Sluorensen and also commentator Carlton Mannes on Plastic Studio because I've been on the market for a long time and every week we also meet in all the trends in the economy that you get a lot of interesting guests. Click on FI Nose Grås Direct TV or search for the market and the economy that you are on the podcast. Now we come to the financial point with FF macro a podcast about the economy. What do the Norway Bank of the world want? When the next decade comes, the economy is in front of the economy. Will the crown be raised in time? If not, I want to have a journalist in the financial point. Meet one of Norway's top economists to see back in the first few months and try to clarify what we have waiting for. See the future of the meeting with Jan Dudik Andreasen, Olavsen, Haran Mangelis Andreasen and many, many more. FF macro is a podcast about the financial point. You hear the podcast. We are going to be very happy to see you and Jackson Holtal until John Powell and many many of you. Here is my colleague and commentator Carl Jan Mannes. He has been on the first meeting with the world. I am interested in all the things that you have to do. This is Paul. He has been on the agenda for eight months. He will be on the agenda and he will be on the agenda. He wants to see the inflation in three weeks. He will have a new inflation table and he will have a decision for the future. He will have the decision to make the decision. The inflation is going up and Like when you're at 10, 분들은ht. UNESCOURR, 1% leverage when it is luckily 10,7% If we're at 3%, but then it is going to be less than 10,5% So the exhibition could be getting higher than 30% as we could. In question is starting at 2018/20 And it is anxiety behind the press conference At the moment that the power could signalise present 20%. It was not going to be the end of the year, but the desire to do it is to know the Federal Reserve's cancer. The head of the head of the head was not going to be the end of the year. But it is not something that is on the influence of the country. He has not seen the estimate yet. So I think it is the same kind of possibility, but almost 100% is not good. Except that it is only 90-80% in the future. And in the future, it is only 33% in the future. And that could be 20% of the numbers, because numbers are important. And it is already the number of employees. So it was the same in the previous year as the previous year. That is Chris Waller. He is one of the two members who have been invited to the event. He is a significant member of the event. He has been working for the event for almost a few years now. But it is not a shock. It is a shock. The impact of the event has been the same. So for the viewers and the statistics, you should have thought about the numbers. So it is like 20% of the numbers are still the same. But we could have entered the third round. But it was a good day, because I have come to the event the most important days of the event. I have been working for a few days now. And not two who have been invited to the event. I have not seen the prognosis there for a long time. And it is a good thing that the press has been there. And Trump has been a part of the. Trump has been a part of the press for almost a few years now. He has given big speech. So he has not seen that Trump has been a part of the press. He has been a part of the press for almost a few years now. And he knows that one and a half hours later, the power has gone down. So he has been a part of the press for almost a few years now. He could have been a part of the press. But he could have been a part of the press. He is not going to be able to see the whole of Trump's presentation. We have a Norwegian press, which is a part of the press. He has missed a little. All the energy has been up to him. But he has not yet reached 2%. And the press has now reached 68%. For example, he goes up on the top of the screen. In the middle of the screen, he sees a group of almost half a percent. And the group of almost half a percent. Helgist of test of Mercedes NECOS 20. - og prisk ut blir det også. Det var det vi aldri for deg på denne fredagun tusen takk for at du så eller hørte på. Jamal islorsen har tilbake enn amme börspåen du låter 50 på fredagda for i med oss både betyr- -impact i en kassosilskapet og at her nu får vart er vi gavre sørenne. I mye lontiden fra alle oss her i Finanaksavisen har en strående hele galt sammen. Vi ses. Eikåne minietene er en podcast fra Finanaksavisen. Programledere er Mario Storrensen. Stein overhøygen åbenediktet storm Bamvik. Prodocenter er Lars Brentens Gram og Barsarjo Har. Vånsvarlede laktør er tryggve hegnar.