Hey, the gas prices are okay right now, but won't they go up next year? While looking at related data, I discovered something a bit strange. If you focus on the years around 2015, you'll notice an unusual phenomenon. The production volume skyrocketed, yet the drilling rigs, known as rigs, have actually been decreasing rapidly. To what extent have they decreased? We are producing 13.3 million barrels, but there are less than a thousand rigs. Production is increasing, but the production facilities are decreasing. So, to put it positively, production efficiency is improving, but is this really possible? 

Hello, I'm Choi Jun. Today, I have some stories related to oil and gas for you. Since you all like energy, I thought I’d prepare something. Recently, when I went to the gas station to fill up, I sometimes thought, Wow, the prices have really dropped; it's cheap! In fact, while the price of gas is important, what's even more crucial is the domestic tax situation these days.

유류세를 계속적으로 인하해 주고 있어서 기름값이 싸게 진행되는 것 같은데, 그러다 보니 항상 걱정이 되는 거죠. 야, 이거 기름값 지금은 괜찮은데 내년 되면 오르는 거 아니야? 또 많은 분들이 유가와 관련해서는 투자도 많이 하시는 거죠. 어떤 사람은 오른다, 배팅하시는 분도 있고, 내릴 거라고 생각하는 분도 계시긴 한데, 제가 그래서 관련된 자료를 보다가 조금 이상한 것을 발견했어요. 이게 왜 이상하냐면, 한번 제가 말씀드려 볼 테니까 여러분도 한번 생각해 보시죠. 전 세계적으로 하루에 생산되는 기름은 여러 번 말씀드렸습니다만, 여러분들이 기억하시기 좋게 한 1억 배럴 정도 돼요. 조금조금씩 늘어나고 있는데, 어쨌든 하루에 전 세계적으로 생산되는 기름은 1억 배럴입니다. 그러면 거기서 1등이 누구냐, 1등은...

In the United States, the production level is fluctuating around 13.35 million barrels, reaching about 14 million barrels. This places the U.S. at the top of the global rankings in terms of production. As you can see on the screen, we have looked at the daily production levels in the U.S. over the past five years. Up until 2020, before the COVID-19 pandemic, production was steadily increasing, but then it dropped sharply due to decreased demand. Now, however, it has recovered and is on the rise again. This trend is quite noticeable. The U.S. is a fortunate country where you can dig into the ground and find just about anything. You might wonder what’s so surprising about that. 

Now, let's take a look at the second graph. At first glance, it appears a bit complicated. This graph shows both production levels and production facilities together. What this means is that the black line represents...

The amount of production generated in a day has been continuously expressed here. It has been shown since the 1980s. As you can see, it decreases steadily, but then it rises sharply after 2009 and into 2010. As I have mentioned several times, this is due to the shale revolution, which allowed the United States to return to being an energy-exporting country. However, the important thing is, what I find strange is the red graph, the red line. Why is that strange? This red line originally represents production facilities, which are commonly referred to as rigs. Naturally, if production increases, the number of rigs should also increase, which is the general trend. However, if you look closely, during the shale revolution, it rose sharply, then briefly fell, and maintained a level afterward. Focusing on the right side of the screen, around the year 2015, where I am pointing, you can see...

A strange phenomenon is occurring. While production levels are skyrocketing, the drilling facilities known as rigs are rapidly decreasing. Currently, we are producing 13.3 million barrels, but there are fewer than a thousand rigs. So, while production is increasing, the production facilities are declining. This could be interpreted positively as an increase in production efficiency, but it raises the question of whether this is really possible. What do you all think? Is it some kind of magic, or is it statistical manipulation? However, it is true that oil is still being produced. So, I looked into how this could be happening. I found that it is indeed surprising how American companies are leading in this area, which is traditionally seen as stubborn and slow to adopt new technologies. They are at the forefront of utilizing artificial intelligence (AI), which has contributed to this phenomenon.

I would like to say that recently, the productivity improvement of oil drilling facilities in the United States, specifically in shale oil production, has been greatly contributed to by artificial intelligence. In fact, if we talk about finding oil underground, there are various investigation stages and a lot of work involved in the production process. First, to locate oil, you might be familiar with seismic waves and various other methods. Data is gathered through these means, and then complex graphs and experiences are used to determine where it might be best to drill. This usually takes several years. However, in recent years, particularly over the last 20 to 30 years, the analysis of such investigative data has been enhanced by the use of supercomputers, which are equipped in locations far from the area of production, where experts gather.

Let's make a decision. The so-called back office can be seen as separate from production facilities, but in reality, managing and operating these facilities in that area is a role performed by those we think of as oilmen, who are constantly moving around with pipes. However, in the case of the United States, as I mentioned regarding the shale revolution, the biggest weakness and concern with U.S. shale is that when you first extract it, it flows out quickly, but then it ends very soon after. This means there is a constant need to drill more. The issue is that going underground to perform horizontal drilling is costly, both in terms of money and time. Despite the efforts put into drilling, the actual amount extracted is only about 10% of the potential reserves that were initially estimated.

There are stories that it is not possible. So, there are some differences by region, but in any case, compared to traditional conventional oil wells, the recovery rate for shale oil is difficult to explain. Generally, we say that the recovery rate is quite low. As a result, it seems like there could be more oil to extract, but in reality, many oil wells in the U.S. are left unexploited because they are no longer producing oil. However, AI is said to be overcoming this issue. AI directs the horizontal drilling process. Various sensors collect data, and people would typically analyze it to decide whether to go left, right, or down to target the appropriate layer for extraction. Recently, AI has been taking over this task. So, robots are assisting humans.

Now, they show this visually, saying that this is good. Then, if the person agrees, they just press the OK button, and the data is sent to the drilling robot on site. The order goes through, and the robot follows the AI's instructions to target the first layer. Such scenarios are becoming increasingly common in the U.S. shale industry. So, when it comes to how much improvement this method brings through AI, the speed of horizontal drilling has improved by more than 30%. In other words, it allows for a much faster approach to targeting the first layer compared to what was done before, which increases production. This is the first point. Additionally, there are various efficiencies that can arise during the process. For example, if something goes wrong in between and they have to backtrack, it can lead to delays as they go back and forth.

You can save various factors such as time and costs. Additionally, if you drill in difficult areas incorrectly, you can reduce various risks and losses. Overall, the efficiency has been calculated to be around 36%. Now, whether 36% is high or low can depend on the context, but in this field, even single-digit efficiency is quite rare. Achieving results above 1 through AI represents a tremendous change. In the past, if international oil prices dropped to around $60 or $50, it was said that U.S. shale would lose its competitiveness. However, recently, it's been reported that U.S. shale can remain competitive even at prices in the mid-$40 range, thanks to AI. This shows that we can see significant advancements in how we drill and the overall time spent on operations, including fracking.

If the time taken seems like the past, it used to take about 11 to 12 days, nearly two weeks, but now there are sites where production from the start takes only three days. So, is this only possible in land areas? No, it's not. If you look at the entire process from exploration to production in regions like the Gulf, which have conventional oil fields and offshore Texas, in the past, it would take over nine months, but now we can reduce that to a maximum of nine days. In fact, nine days seems quite fast, but regardless, the application of AI has significantly increased efficiency compared to the past. Reports are coming in from various places, not just in shale but also in traditional production facilities. This indicates that the increase in U.S. crude oil production is greatly supported by AI. Now, another thing to note is about AI data.

To collect data, there are various costs involved, such as for sensors, which might make you think it requires a lot of money. I had similar thoughts, so I did some research. It turns out that, in fact, less data is needed. From a human perspective, if 100 data points are required to make a judgment, you might wonder where to focus: is it in this area? There are many questions to resolve. However, when it comes to AI, I found that using only about 3% of the data that a person would use, or in some extreme cases, even just 1%, can provide better solutions than humans. This means that with the existing equipment alone, we can supply the necessary data for AI. In new areas, even with a small investment, we can extract data that can be utilized for AI, which is really great.

One of the advantages of AI is truly fascinating. Most of the equipment used for drilling oil has a high probability of malfunction due to high pressure and continuous operation 24/7. However, once it breaks down, it can be quite challenging to determine where the issue occurred. First, all the valves need to be shut off, and it takes time to identify the problem. During this time, oil production is not functioning properly, leading to various cost-related issues. But AI can foresee the future. What this means is that it can predict when the equipment installed in the oil well is likely to break down. By continuously monitoring the existing equipment, it can indicate, for example, that a certain valve has a 90% chance of failing within the next week. From the perspective of the operators, this allows them to schedule maintenance ahead of the usual inspection times.

By replacing equipment or changing parts, or by conducting checks, risks can be reduced. This is something that stakeholders in the transparent ball story and the Je ball story have mentioned, as they are receiving significant help through these methods. When asked how far ahead they can predict, it varies by site, but generally, they say it's possible to look ahead about 30 days. In a way, this allows for the prediction of subtle changes that are difficult for people to notice, thereby enabling the anticipation of production equipment failures. This proactive inspection helps to increase efficiency. As a result, in the case of U.S. oil wells, compared to the past, there is a cost-saving effect of at least over $100,000 per well, although the figures can vary widely.

From now on, regarding the development of new wells, it is said that the cost can be reduced by more than 30% when viewed from the perspective of development costs. So, AI technologies, which we have been hearing about since 2023, are now being applied diligently for about two years. Data on this is now coming out. As we enter 2024, more applications are being made. Looking through the related materials from the United States, it seems that starting from the second half of next year, we will likely be able to derive more reliable average values through AI data. However, even with the data available so far, the efficiency and cost reduction aspects are likely to show at least a double-digit percentage, and it seems certain that it will be over 10%. If we estimate more generously, we can expect a reduction of about 20% to 25%.

It seems that there will be. So, in terms of productivity efficiency here, there are also cost savings involved in finding new things. Additionally, as I mentioned earlier, there are effects related to preventing breakdowns. Another important point is that AI is better at managing various factors to ensure that existing oil wells can continue to produce oil steadily. Generally, when you drill into the ground for an oil well, the oil gushes out due to the underground pressure. Controlling this is quite challenging, and over time, the pressure decreases. Consequently, the production volume gradually decreases. Therefore, as time goes by, the production volume of the oil well diminishes, and to prevent this, one might inject carbon dioxide or seawater, just like I mentioned earlier regarding Saudi Arabia.

There are cases where saltwater is introduced to increase pressure for production, and in this regard, AI has remarkable capabilities. It can adjust pressure in a surprisingly effective manner and enable production over a longer period. It also helps find better sales opportunities, reduces costs, and ensures long-lasting operation without failures. Therefore, from the perspective of the United States, it is natural to see a fascinating graph like the one I showed on the screen, indicating that the same facilities can produce more. In terms of technological advancement, especially recently, AI plays a significant role. What I find intriguing is that typically, those involved in oil exploration, whom we refer to as oilmen, are perceived as tough individuals.

There are many men, often referred to as rednecks, who are considered a very conservative group. However, in reality, these individuals are facing one of the biggest challenges in the U.S. oil and gas industry in recent years: a lack of workers. There simply aren’t enough people willing to do tough jobs that require 24-hour shifts and hard labor. As a result, finding workers has been quite difficult. Nevertheless, these individuals show a willingness to help with anything related to labor, demonstrating a proactive attitude by saying, Let’s just come and try it out. This is quite unique because, despite their conservative and seemingly negative stance on certain issues, these oilmen are very open and proactive. Whether it’s AI or any other technology, they are eager to improve productivity.

Anyone who seeks improvement tends to have a subtly open attitude, saying okay to anything. This, in a way, aligns positively with the reality of the AI industry. Initially, it was primarily small companies that utilized these technologies, but recently, major firms have also started to adopt them. A survey conducted among major companies revealed that over 92% plan to increase their investments in AI. The reasons for expanding AI investments include productivity enhancement and cost reduction, among others. Ultimately, they perceive that this is worth investing in. In the United States, there are many regions associated with shale, such as North and South Dakota, and these areas are also involved in this trend.

대표적으로 이제 텍사스, 텍사스 같은 경우는 좋은 대학들도 많지 않습니까? 그러다 보니까 최근 들어서 샌프란시스코에서 떠난 많은 IT 기업들이 이쪽으로 가고, 인력들도 가다 보니까 의도하지 않게 좋은 IT 인력 고용과 가까운 근거리에서 협업하는 구조가 많이 만들어지고 있는 거죠. 그러니까 어떻게 보면 AI 전공자라는 관련 기업들 입장에서 봤을 때는 전혀 생각지 못했던 좋은 일자리와 고객들을 발굴할 수가 있고, 그래서 실제로 AI 관련 일자리가 텍사스 지역에서만 한 30% 가까이 증가하고 있다는 이야기도 나오고 있습니다. 그래서 이러한 것들을 눈여겨보던 영국의 BP 같은 석유 메이저는 여러분들이 잘 아시는 미장에 투자 많이 하는 회사죠. 팔란티어와 이제 얼마 전에 5년짜리 계약을 맺었습니다. 원래 팔란티어와 BP 같은 경우는...

Digital twin technology, which means creating a digital replica of a physical situation and conducting various simulations, has been a collaborative effort. This time, we announced that we would work together on improving productivity through the use of AI. Of course, the specific details are confidential, but as you all know, Palantir is showing real impacts here, which I believe could lead to significant changes. As we discuss these topics, we inevitably focus on the United States. However, this isn't solely an American phenomenon; technology is such that when there's a good example, everyone tends to follow suit. Take Chinese drones, for instance; everyone talks about them, yet they are widely favored in the field because they offer the best quality and are used the most.

Because it is convenient, many people prefer good products and good technology. As a result, recently in Brazil, particularly in deep-sea oil fields, there are efforts to delegate various tasks to AI. In Brazil, for instance, oil fields are located at depths of over 3,000 meters, and they are beginning to explore these AI applications. So far, the data is very promising. It is reported that AI can reduce the time spent on tasks by about 60% compared to human efforts. Many countries and companies are looking to adopt AI, and projections suggest that by 2030, at least 15% of oil and gas production facilities worldwide will be managed by AI. Thus, utilizing AI for production brings unexpected additional advantages. One of these advantages is particularly...

When producing gas, one of the side effects is that the trapped methane is released, which increases the amount of greenhouse gases in the atmosphere. Compared to carbon dioxide, methane has a much stronger greenhouse effect. Through AI, we can precisely control this, and based on the data available so far, it is suggested that we can significantly reduce methane exposure. In Saudi Arabia, for instance, during the production process, it is quite difficult to dispose of the gases that are released, so they simply flare them off, which means they burn them. However, Saudi Aramco has been using AI to manage this process and has managed to reduce emissions by more than half. Burning these gases is not only a waste of money but also an issue that needs to be addressed in terms of reducing greenhouse gases, and AI is providing significant assistance in this regard.

We can now see that there is AI, which is quite a fascinating thing. However, when we actually try to use it, we find ourselves wondering where to apply it. We are scratching our heads. In a way, AI can be seen as a toy we play with, but rather than that, if we focus on industries where we can gather specific data and have clear goals, investing and training AI there can yield good results. The oil industry is demonstrating this, and we can say that the leading country in this regard is the United States. Consequently, there has been recent discussion about considering AI technology advancements in relation to oil price forecasts. What this means is that productivity is increasing. In the past, there were regions where production was difficult at current oil prices, but now...

There are cases, for example, in Russia and the Arctic region, as well as in Brazil, where it is known that large quantities of crude oil are buried. However, due to the current oil price levels, many regions are unable to produce. If AI technology is applied to production in these areas, costs could be significantly reduced—by about 25% to 30%. This means that even with oil prices around $60 to $80, the number of regions that can feasibly produce oil would increase considerably. As a result, production costs would decrease and production volumes would increase, leading to a downward trend in future oil prices. There are forecasts suggesting that oil prices will decline. Various institutions are evaluating these situations, but even though there is still a lack of concrete data, some numerical estimates have been made.

It is weak, but still, if we talk through the numbers, I think we can lower the price of AI by about $5 per barrel. Then, looking at various aspects, we could produce in areas where production is currently limited due to marginal costs, and also revive regions that were abandoned in the past because extraction was no longer feasible, using AI. This means we could potentially increase production in those areas. Therefore, I believe the peak period could be much further into the future than we initially thought. This connection leads me to realize that recently, among the international investments, one of the most interesting areas has been AI semiconductor stocks. Countries like Saudi Arabia and the UAE are showing significant interest and promising investments in AI semiconductors. These countries are not just vaguely thinking that they will make a lot of money by investing in this.

As they say, through technologies related to AI, when you dig into the ground, you find resources. If you can sell these resources for a longer period, then it is a country that must do anything to sustain itself, isn't it? The Middle Eastern countries, for instance. After 70 years, what will Saudi Arabia have left? Nothing, right? However, if they can make something last for 150 or 200 years more efficiently, then of course, they are in a position where they must act. This is where we can find the reasons why Middle Eastern countries are investing in AI semiconductors and various AI technologies. In fact, Middle Eastern countries have abundant energy resources, and they also have rich alternative energy sources like solar power. So, the Middle East is a region that can freely use cheap and abundant electricity, which is essential for the AI industry. Therefore, various movements in the Middle East are significant.

Why are they like that? Why are those friends acting that way? We can now see that we can concretely understand the vague thoughts we've had. When we talk about the United States and China, you might think, Hey, Dr. Choi, aren't you forgetting a country? But I understand what you all are thinking. Is China just standing still? No, they are not. China is also making various investments and efforts related to AI. They are truly sincere about combining this with visible, tangible elements that they are strongly promoting. For instance, looking at Sinopec, a representative energy company in China, we can see that they are operating various production facilities more effectively through exploration and reporting.

They are making efforts to push in that direction, and it is mentioned that they are continuously improving by deploying their own AI model developed by Sinopec. Moreover, companies like PetroChina are forming networks among Chinese companies to enhance collaboration specifically to solve our problems, rather than just developing AI technology in a general sense. It is said that PetroChina is currently collaborating with Huawei. In a way, this is different from a vague approach to technology development, where you might wonder if something will come up on a test. It is more about having to solve specific problems in order to move on to the next stage. This reflects the different goals of the Chinese government.

What we expect from AI is that it will help solve problems using data available in the present situation, allowing both parties to win and move forward in the anticipated structure. Energy, especially traditional oil and gas, is precisely the kind of target that aligns with the interests of the Chinese government. We can see that now. Chinese energy companies are involved in AI, and while they are conducting overseas exploration, they are primarily focused on domestic production and investing more in aspects related to the efficiency of distribution and other processes. 

The concerning aspect of AI related to energy in China is that the government is currently pursuing two main initiatives. The first is to create a standardized platform. If companies like PetroChina and Sinopec differ, it creates a situation where...

It's a bit difficult, so let's create a common AI platform that energy companies in China can use. Naturally, we can also establish common standards related to the data that will be involved. This means we can develop standards related to data acquisition and various procedures associated with it as we move forward. Of course, it can be beneficial to have various standards and formats competing, which might lead to a more advanced model, or standardization might be more advantageous, allowing for further progress in that direction. In any case, China is working hard to enhance its energy competitiveness through AI, leveraging what they can do best. I've discussed the United States, the Middle East, and now China. AI is now...

We need to keep in mind that it is not just a vague and interesting object. In our country, for several years now, we have been discussing smart factories and smart industrial complexes. While we have talked a lot about improving productivity through big data, it seems that clear results have not emerged. However, when we think about it, the use of big data is limited by human interpretation and action. In our country, what is most needed right now? We are facing a situation where there is a shortage of workers in the manufacturing sector due to aging and retirement. I believe that a clear goal should be set to tackle this issue by introducing AI to produce results, and that we need policies that narrow down the targets. In any case, there are various aspects related to oil.

We have talked about various stories, but conservative companies and many businesses often categorize things in this way. So, in a way, proactive companies are on the right side, while many others still express a conservative stance. However, already several companies are adopting these technologies. If you look closely, in the case of SK Innovation in our country, it is reported that they are actively embracing and working on AI-related technologies from the perspective of the oil industry. It’s quite fascinating, and I hope we can invite them to share their experiences someday. Today, we explored the mystery of why oil production facilities in the U.S. continue to decrease. Did you find it interesting? I will bring more related stories in the future to share with you.

I will do it. Thank you very much for watching until now, and I will see you again next week. Thank you.

기름값.. 더 싸질까? 인공지능으로 석유를 찾는 미국 석유산업 [미국 석유산업]

세상의 모든 일이 궁금한 호기심대마왕 최준영 박사와 함께 떠나보는 지구본 여행

광고 문의 : scape197211@daum.net
콘텐츠 및 기타 문의 : scape197211@daum.net


기름값.. 더 싸질까? 인공지능으로 석유를 찾는 미국 석유산업 [미국 석유산업]

기름값.. 더 싸질까? 인공지능으로 석유를 찾는 미국 석유산업 [미국 석유산업]

스크립트