Welcome, everyone. I'm Spencer Dale, BP's Chief Economist, and thank you all for sparing the time to join us today for the launch of this year's Energy Outlook, both here in BP's headquarters in St. James's, London, and virtually around the world, where at last count we had really quite several thousand people joining online. You're all very welcome. Thank you for sparing your time. We don't take your time for granted. We will try and make this session as informative and as interesting and as fun as possible. Thank you all very much. At this point, let me also just give a shout-out to the rest of the team that helped produce this year's Outlook. Many of them are in the room with you today. Much of the Outlook was put together over the summer.
Most normal people were either at the beach, sitting under a tree with their feet up. Thank you, the team, for all your time and effort, especially over the summer. Much appreciated. The energy system sits at the heart of modern-day society. It's critical for the everyday needs of people and businesses around the globe, adapting to changing political, technological, and environmental priorities. That central role is the very reason why many of us choose to work in or around the energy industry. It's also why the challenges and forces shaping the global energy system are forever changing. The past year has been no different. Geopolitical tensions, which came to the fore with the war in Ukraine, escalated further with the conflicts in the Middle East and the increasing use of sanctions and tariffs, refocusing attention on energy security.
The seemingly exponential growth in data centers to support the increasing use of AI. Gains in energy efficiency growing at a subpar pace, boosting energy demand. Carbon emissions continuing to rise with the risk that it becomes harder and more costly for the world to remain within a given carbon budget. There are lots of issues for this year's Energy Outlook to get its teeth into. This year's Outlook is based around two main scenarios shown here, current trajectory in green and below two degrees in blue. The scenarios are framed around different assumptions about the speed and nature of the energy transition, the single biggest uncertainty facing the energy industry. Current trajectory, as the name suggests, tries to capture the broad pathway along which the global energy system is traveling. It places weight on climate and energy policies already in place.
It places weight on government aims and pledges for future decarbonization. It also places weight on the fact or the difficulty of actually meeting those aims and pledges. There is a grid of judgment in here. As you can see on the whiteboard here, carbon emissions in current trajectories sort of stabilize through this decade, and then they gradually decline during the 2030s and 2040s. The pace of that decarbonization is slow and shallow. By 2050, carbon emissions are only around 25% lower than they are today. In contrast, below two degrees explores how different elements of the energy system might change in a pathway in which the world achieves a faster and deeper reduction in carbon emissions. Carbon emissions in below two, shown in blue here, fall by around 90% by 2050. Below two assumes a significant tightening in climate policies.
It also embodies shifts in societal behavior and preferences, which further support gains in energy efficiency and the adoption of low-carbon energy. In the Outlook, we show below two is broadly consistent with limiting the increase in global average temperatures to well below two degrees. At the risk of sounding like a complete stuck record, let me repeat the warning I always make at this point. The energy system will not exactly follow either of the paths described by the scenarios. The scenarios are not predictions. We cannot predict the future. We know we cannot predict the future. I can imagine some of you who are sitting there thinking, "Whoa, hold on. So Spencer's just told us both of these scenarios are going to be wrong." Neither of them consider those other issues he mentioned at the beginning, increasing geopolitical fragmentation, weak energy efficiency, the risk of delayed transitions.
What am I doing here? Am I wasting my time? Don't worry. I feel your pain and know you're not wasting your time. Today's presentation will address both of those concerns. Although there is only a tiny chance of the energy system turning out exactly in line with either of the scenarios, we can still use the scenarios to develop key insights about how the energy system might develop. In particular, we can identify different aspects of the energy system which are common across both scenarios.
The idea here, the way to think about this, if I can see qualitative trends in the energy system, which are apparent in a pathway which is similar to one we're on now, remember that slow, shallow decarbonization pathway, and that same trend is also apparent in that rapid decarbonization pathway shown by below two, it may give us extra confidence that those same trends will also be in pathways between those two cases. In contrast, if other elements of the energy system are very different in below two relative to current trajectory, it suggests they're more dependent on the speed of transition. The first part of today's presentation will identify some of those key common trends and some of the key differentiated trends. What about those other issues affecting the energy system over and above the speed of transition?
In this year's Outlook, we've developed a series of sensitivity analyses to consider those other issues. In the second part of the presentation, we'll take you through two of those sensitivities, considering the possible implications of increased geopolitical fragmentation and also the potential implications of weaker energy efficiency. At that point, you're pleased to know that we're going to stop bombarding you with charts and more charts and more data. We're going to stop and we'll switch to the Q&A session for the final 14 minutes or so, which is always, I think, the most fun and interesting bit. For those of you online, please feel free to submit your questions at any point during the presentation. You can also vote for the questions which you think are most interesting and put them to the top of the list.
I'm afraid, guys, for you in the room, you'll have to show a little bit more restraint. No shouting out, and you have to wait for the Q&A. Also, for those joining online, we'll invite you to take part in a real-time poll on some of the issues raised by this year's Outlook, and we'll feedback the answers from that poll before the end of the presentation. We have an action-packed session ahead of us, so please don't go anywhere. If you're watching online and thinking, "Is this the right time to go and get a cup of tea?" No, it's not. You have to stay with us. Otherwise, you're going to miss something really exciting. Please stay with us. You may have noticed from the opening slide that we're going to do something different today. We're going to have a two-handed presentation.
I'm super excited that Gareth Ramsay is going to take you through the next part of the presentation, and then I'm going to come back to outline the results of the sensitivity analysis. I've worked with Gareth, I think, for the best part now of about 20 years in a variety of different economic and energy roles. For the past few years, Gareth has been leading the production of the Energy Outlook. As some of you may know, Gareth will be taking over from me as BP's Chief Economist from the beginning of next year, where I know he will do an awesome job. Gareth, why don't you come up and I'll hand over the Chief Economist's pen.
Thank you. Thank you so much, Spencer. As he says, don't worry, everyone, that is not the last you're going to see of Spencer today. As he said, I've been leading production of this Energy Outlook. First, I'm going to talk you through some of the key elements of the Outlook, what happens to the crucial sort of moving parts of the energy system in our scenarios. Spencer is going to come back to do his even more fun bit at the end and to look at a couple of fundamental ways that things might be different to these two scenarios. Before that, what do we want you to know about what's in them?
As Spencer said, I'm first going to highlight key features which are there in both our two scenarios, things that might therefore be likely to occur across a range of other pathways as well, perhaps those in between the two scenarios as well. The first of those common trends is around oil. This chart shows the paths for oil demand in our two scenarios. You will see they are not the same. The profiles do indeed differ in terms of how long demand carries on rising for and then how much it declines. In the current trajectory scenario, that's the pathway that the world is on at the moment, oil demand keeps on rising over the rest of this decade before it then starts to gently fall back. By 2035, it's only fallen back to around its current level.
It then gradually falls over the second half of the Outlook to a little above, to around 85 million bbl a day by 2050. Whereas, as you can see in below two, the falls in demand start a bit sooner, and then they happen with much greater intensity. The oil consumption still falls all the way to around 35 million bbl a day by 2050. The common trend that I want to pull out here is that both scenarios imply that oil carries on playing a central role in the energy system for at least the next 10 -1 5 years. That matters a lot for the investment that we're going to need.
As the International Energy Agency points out last week, because of the way the output of oil fields naturally declines, we're going to need hundreds of billions of dollars of new investment every year to meet these kinds of levels of oil demand. There is a second common element in the two oil demand profiles that I want to pull out as well. That's a change in where this oil is going in our economies. This one might seem a bit more technical, but it is really important. This chart shows the growth in oil demand, the sort of average growth per year, both in the past, that's the bar on the left, and then in the future, the bars on the right in our two different scenarios. It shows where that oil is going, marked in the different colors.
For a very long time, including since 2010, as you can see on the chart here, the single biggest driver of oil demand growth in the global energy system has been rising demand for oil in road transportation, in our cars and in our trucks. We're now at an inflection point in these scenarios. You can see that all of those blue bars are negative in the future. That's the use of oil in road transportation falling in both our scenarios. In some ways, that might seem surprising. We know that in emerging economies, in particular, as people get richer, their demand for mobility, including for road transportation, rises. That's more than offset at a global level by two things happening. First, importantly, the average vehicle on the road becoming more fuel efficient over time. Second, of course, an increasing shift towards EVs.
Instead, the biggest single driver of oil demand growth in the future is not oil being combusted. It's oil being used as a feedstock in what we call the petrochemical sector, particularly for the production of plastics as well as other oil-based materials. That's the red bars on this chart. This use of oil pushes up on demand throughout the whole Outlook period in our current trajectory scenario, and even in the first half of the Outlook in below two as well. Oil for this kind of use as a feedstock is only around 15% of total oil demand now. That rises to over 30% in 2050 in current trajectory and to more than 50% in below two. That's more than half of oil being used to make things, not for energy.
Although we know this is going to be important in the future, the precise scale of it is still pretty uncertain. I won't get into the maths here, but we show in the Outlook that different but equally plausible assumptions about how much plastics we're going to want in the future can lead to really big differences in feedstock demand of as much as 10 million bbl a day oil equivalent, most of the large majority of which will be oil by 2050. This seemingly quite niche question, basically, of how this petrochemical feedstock demand is going to evolve in the future is actually going to have a really big bearing on the prospects for oil demand in the future. The next feature of the energy system that you can see in both scenarios is that the world continues to electrify.
Over the past 10 years, our demand for electricity has grown twice as fast as our demand for energy overall. That trend just continues over the Outlook, with the world's demand for power doubling or more than doubling by 2050. The vast majority of this electricity demand growth is in emerging economies, driven again by rising prosperity. It means that the share of energy that the world consumes taking the form of electricity rises from only a bit more than a fifth now to around just below 30% in current trajectory, to 30% here, about a third, and to more than 50% in below two. That's more than half of the world's energy use electrified by the end of the Outlook period. Most of this higher power demand is for growing demand where we already use a lot of our electricity. That's in industry and in our buildings.
Of course, that's being supplemented by newer uses of electricity, including, of course, those electric vehicles, but also higher data center demand to enable rising use of artificial intelligence. That higher data center demand on its own makes up around a tenth of the extra demand for electricity over the next decade or so. Important, but far from everything. Of course, data centers have a much bigger impact than that in some places globally, especially in the U.S., where it accounts for about 40% of electricity demand growth over the next decade. Of course, artificial intelligence makes its way into every conversation you have anywhere. Data centers are being discussed an awful lot at the moment. In this year's Outlook, we have a special section which is talking about how artificial intelligence might affect the global energy system, how it might impact the system.
I'm not going to talk you through all of that today. Instead, I just want to give you one takeaway from that discussion. That is that the impact of artificial intelligence on energy is unlikely to hinge on how much power data centers need. None of us know how much AI is going to improve our productivity, and so how much it's going to improve economic growth. Plausible estimates of that impact could imply increases in energy demand 20 x greater than the growth in data center power demand. AI won't just affect energy demand. It could have equally big implications for the supply of various types of energy and the efficiency of many parts of our energy system. The bottom line here is if we're going to think about AI and energy, we do need to think much wider than just data centers.
This increasing electrification of our energy system takes us to another feature of the energy system, which is common across both our two scenarios, which is that the growth in power generation is, in essence, a story about wind and solar power. Wind and solar generation, the orange bars here, account for all of the increase in power generation globally in both our two scenarios. In the case of below two, more than all of it. That means wind and solar are going to become a central foundation of our energy systems in these two scenarios. Added together, they generated about 15% of our electricity last year. By 2050, they account for more than half of global power generation in current trajectory and more than 70% of it in below two.
We often talk about wind and solar together, as I have here, but as has been the case recently, solar grows faster than wind in both our two scenarios. Its costs go down faster, it's quicker to deploy, and also it gets greater policy support as well. This increasingly central role for wind and solar power does bring with it new challenges. As we mentioned in the report, changes are going to be needed to our power systems to make them more resilient and reliable and capable of balancing demand and supply as generation comes more and more from these weather-dependent variable sources. I want to think about power generation from a slightly different angle now to help us think about the energy transition in a slightly bigger picture way. In last year's Outlook, we introduced a distinction between two different phases of the energy transition.
First, the energy addition phase. In this, low-carbon energy is growing rapidly, but it's not growing fast enough to meet the overall growth in energy demand, so use of unabated fossil fuels is growing as well. Then the energy substitution phase, in which low-carbon energy is growing fast enough to more than meet the increases in overall energy demand so that unabated fossil fuels decline. Now, one point we made in last year's Outlook was that in previous energy transitions, the world has stayed in this energy addition phase. It's not consumed less of the old energy sources. This observation has been used by others to highlight just how challenging it's going to be to actually reduce carbon emissions.
Although it's true that the world as a whole is still in this addition phase, we show in this year's Outlook that the switch from energy addition to energy substitution has already happened in a lot of places, including the EU and the United States. We calculate that around a third of the world's primary energy today is being consumed in regions which have moved from the energy addition phase to the substitution phase. That increases to around 60% of energy by 2030 in current trajectory. That's helped in part by China moving into this substitution phase. We can see the same trend for power generation. This chart shows the share of global electricity generation, which is in the power sector equivalent of energy substitution.
That's the share of the world's generation where low-carbon power generation is rising fast enough to more than meet the increases in overall power demand, so that generation from unabated fossil fuels is going down, even if power demand overall is rising. Already, as you can see, around a third of electricity used today is now in countries in this power sector substitution phase. In current trajectory, this increases to around 60% by 2035 and over 70% by 2050. This is in current trajectory, remember. This is the pathway along which the world is currently traveling. Of course, the increase in below two is even quicker than this. Those are the key common trends I wanted to highlight across the two scenarios, trends that we think might also happen in pathways that perhaps sit in between those two scenarios as well.
First, oil demand playing a major role fueling the global economy for at least the next 10- 15 years, but with that demand increasingly supported by oil for making things, not moving things, as its consumption in road transport gradually wanes. Second, the continuing electrification of our energy systems, with the huge growth in power demand met by more solar and wind power. I want to switch now to those trends which differ materially across the two scenarios, suggesting these ones are much more sensitive to how fast the energy transition turns out to be over the coming decades. The first example is the outlook for natural gas, shown in this next chart. Now, natural gas demand is actually pretty strong in both scenarios over the next decade. As you can see on the chart, it then takes two quite different paths.
That is because of two different forces that are pulling gas demand in different directions in the two scenarios. In current trajectory, so that's the pathway the world is currently on, natural gas demand keeps on rising basically over the entire outlook, or at least until the mid-2040s. By the end of the period, it's around 20% above its current level. That is mostly driven by rising demand in emerging economies as they grow and they industrialize. Gas demand is up by around a fifth by 2050. In contrast to that, in below two, natural gas demand starts to decline in the early 2030s. By 2050, it's down around a half from its current level.
A dominant force in that scenario is just the push for greater decarbonization, with gas losing even more share to wind and solar power in power generation and being displaced directly by the faster electrification of buildings and industry. In practice, both of these forces are likely to be at work in the future. The outlook for natural gas is going to depend on the relative strength of these two forces. The other feature, finally, of the energy system that I wanted to highlight as being particularly dependent on the speed of transition is the outlook for low-carbon energy technologies and low-carbon energy vectors, which are newer, less mature, more expensive. The charts here focus on just two of those. On the left, demand for low-carbon hydrogen. On the right, carbon capture, use and storage.
I should say the same big picture that I'm about to tell you about these two charts holds for others of the newer, higher-cost low-carbon energies and vectors like sustainable aviation fuel, ammonia and methanol in shipping, even direct capture of emissions from the air. The story is that in current trajectory, the growth of both low-carbon hydrogen and carbon capture, use and storage is pretty limited. Low-carbon hydrogen reaches around 75 million tons per year in 2050 in current trajectory. How do you think about that number? That is less than the current demand for hydrogen, which we get almost entirely from unabated fossil fuels, so-called gray hydrogen. Likewise, carbon capture, use and storage reaches only around 700 million tons of CO2 in current trajectory by 2050. 700 million tons means only around 2% of energy sector emissions are being captured in 2050.
The growth of low-carbon hydrogen and of carbon capture, use and storage is much stronger in below two. Low-carbon hydrogen gets to around 350 million tons per year by 2050, not just replacing gray hydrogen, but also being used, for example, in transport and in industry. Carbon capture, use and storage reaches around 6 Gt of CO2 by 2050 in below two. The reason for these sharply different profiles in our two scenarios is pretty straightforward. These types of technologies are just expensive relative to their current alternatives, whether that's coal and natural gas unabated in the case of hydrogen production, or whether it's the fossil fuels that hydrogen might be competing with, or whether it's just simply not capturing emissions in the case of carbon capture, use and storage.
These types of technologies increase materially in pathways in which we, society, are willing to bear the additional costs of reducing these harder-to-abate carbon emissions. Even then, much of the growth in these technologies you can see is concentrated in the second half of the Outlook, really after 2035, as carbon policies tighten enough and also as costs decline. Those two key features of the energy system, which are particularly dependent on how fast the transition goes, are first, whether natural gas demand rises or falls over the next 25 years, and second, the pace and the extent to which these less mature, higher-cost low-carbon energies and technologies like low-carbon hydrogen and carbon capture, use and storage develop. At which point, as promised, we're going to go back to Spencer. He's going to step back from the detail and do a bit that's even more fun.
Spencer, let me hand you back the pen.
Aha!
For now.
Thank you. I'm suddenly not sure if I want to give up that pen, actually. Thank you, Gareth. As Gareth said, in this final section, I want to shine a light on those other issues affecting the energy system over and above the speed of the energy transition. I want to focus on two, increased geopolitical fragmentation and the possible implications of weaker energy efficiency. As I said, we do this by conducting some so-called sensitivity analyses of the main scenarios. This type of analysis has the benefit that most elements of the scenarios are assumed to be unchanged, allowing the key features of each issue to be isolated and explored. It has a drawback that in reality, other elements of the energy system wouldn't be completely unchanged if these issues were to materialize.
As such, the results of the sensitivity analysis should be viewed as sort of illustrative rather than providing a complete characterization or a detailed quantification. With that caveat in mind, I want to start by considering the possible implications of a significant increase in geopolitical fragmentation. The motivation for this is, I think, unfortunately, pretty clear. There's been a significant escalation in geopolitical conflicts and tensions in recent years, including the wars in Ukraine and the Middle East and the greater use of trade sanctions and tariffs. Further escalation could lead to increasing geopolitical fragmentation. Such an increase could impact many different aspects of our economic and political world. The focus here is on how it may affect the global energy system. In particular, I want to focus on the possible implications of increasing fragmentation causing countries to reduce their exposure to international trade and becoming more self-reliant.
That's the focus here. What's sort of interesting is there's been quite a lot of discussion about this type of issue over the last year or so. Quite a few people, different people have been writing about it. What's interesting is you read those different commentaries, they often come to different conclusions. I think in part it's because different people have focused on different aspects of how this may work. One of the things we wanted to do today or in the Outlook is to try to carefully map out the various different ways, the various different channels through which a shift towards greater self-reliance may impact global energy. Let me take you through where our thinking got to.
Most directly, a shift to greater self-reliance might dampen the growth on international trade as countries move their supply chains back home or restrict them to countries or regions most politically stable or aligned with them. That weaker international trade will tend to dampen economic growth as the scope for increased specialization and competition is limited. That weaker GDP growth, weaker economic growth will then tend to feed through into lower energy demand. That's one of the channels we want to think about. Increased geopolitical fragmentation may also heighten the importance of energy security as countries seek to reduce their dependency on imported energy and energy technologies. Heightened concerns on energy security may trigger three types of reaction. First, an increased preference for domestically produced energy relative to imported energy.
If I'm worried about my energy security, I don't want to keep on importing my energy from abroad all the time. There's a similar desire that you also may want to reduce your dependency on imports of energy technologies, including low-carbon technologies, with a corresponding emphasis on developing domestic or at least diversified supply chains, even if they come at a greater cost. Third, you may place an increasing weight on energy efficiency as this reduces the need for all types of energy and so bolsters energy security. Increased geopolitical fragmentation might also lead some countries to place less weight on climate and sustainability goals. In part, this simply reflects the nature of the so-called energy trilemma. If countries place greater weight on energy security, it necessarily implies they must place less weight on the other two elements of the trilemma, either energy affordability or energy sustainability.
Moreover, that slower economic growth we just talked about stemming from weak trade may also mean that countries have less resources to devote to decarbonization, especially to those mature, higher-cost, low-carbon technologies that Gareth was just talking to us about, low-carbon hydrogen, CCUS , sustainable aviation fuel. This is sort of where we got to over the summer when you were on the beach trying to think about through all these different channels. In the Outlook, we explain how we go about calibrating these different channels. Don't worry, I'm not going to go into that today. Rather, what I want to do is try and focus on the results. How does this impact the global energy system? That's shown in this next chart here. This chart shows the impact of increased geopolitical fragmentation on the level of primary energy relative to current trajectory.
Those color blocks show you how different energies have increased or decreased. The first bar here shows you the impact of that lower GDP stemming from weaker net trade. Primary energy falls, you can see here, and it's then spread across these different fuels in terms of oil, gas, coal, and renewables. The relative weight impact on those different energies depends on the weights of those fuels in the end uses, such as the importance of the role of oil in transport, the role of coal in industry and in buildings, and how sensitive demand in those different end sectors are to movements in GDP. The second bar shows the impact of increasing concerns about energy security. The important point here is that concerns about energy security generate a mix of offsetting effects.
The increased preference for domestic rather than imported energy leads to a shift away from oil and natural gas, which are the most heavily traded fuels, towards renewables and coal, which tend to be produced and consumed more domestically. The higher cost of renewable energies as countries move away from international supply chains, together with the lower weight attached to climate goals, weighs against low-carbon energy and favors fossil fuels. This is what you end up with in this result here. What happens here is, as a result, at a global level, less oil as people move away from importing oil, less renewable energy as people place less weight on climate goals, and also the cost of renewables has gone up. More coal, because for many countries, particularly in Asia, that's a source of domestic energy. Gas is the interesting one.
Gas on its own, the increased preference for domestic energy would mean that gas would fall in a similar way to oil. As you place less weight on renewable energy and you're using less wind and solar, you need more gas to do your power sector, to your power generation, which pushes it back. The net is it sort of all evens out. That's a way to think about what's going on here. The third bottom bar shows the net impact of these two broad channels. Overall energy demand is lower, and the carbon density of the fuel mix is slightly higher, reflecting that increased share of coal. I must confess, I was a bit nonplussed when I first saw these results. We've gone to this trouble to model all these different channels. We've done all this different calibration, but the impact on the energy system is pretty limited.
Energy demand is a bit weaker, but the fuel mix isn't much changed. As we dug into the results, we realized the limited impacts at a global level are not because increased geopolitical fragmentation doesn't have important implications for energy systems. It does. Rather, increased fragmentation impacts different countries in different ways, which at a global level tend to offset each other. It's more interesting to look at fragmentation, how fragmentation impacts individual countries. That's what this chart does here, where I've compared the impact on the U.S. on the left with China on the right. The reason for picking the U.S. and China is not just because they're the two most important countries in the energy system, but also because their economic and energy structures differ in several important ways. As you know, the U.S. is a net exporter of fossil fuels, and a heightening in energy security doesn't lead it to want to reduce its demand for oil or natural gas. It does cause it to reduce its dependency on imports of low-carbon technologies. Moreover, the U.S. economy is relatively less trade-intensive and is therefore less exposed to that weaker net trade channel I was telling you about. China is close to the mirror image. Heightened energy security concerns cause it to reduce its dependency on imported oil and natural gas. Since it's a dominant producer of low-carbon technologies, its access to low-cost, low-carbon energy is largely unaffected. China is also highly exposed to international trade and is more affected by the impact of weaker net trade on economic growth. These differences in economic and energy structures lead to differences in the impact from increased fragmentation.
You can see for the U.S., the overall impact on primary energy is relatively limited. It's a relatively less trade-intensive economy. The impacts produced in oil just reflect the sheer size, the importance of oil in the U.S. economy, and also lower renewable energy as it shifts away from importing low-cost renewables. Natural gas sort of has those two cost-cutting impacts. At one level, the lower demand for energy just reduces natural gas, but it then gets crowded in because you have less renewables going on. In contrast, in China, bigger impacts on the overall energy demand are more broadly based across all four sources of energy.
The general point here is that an increase in geopolitical fragmentation would be likely to lead to more differentiated energy pathways, differentiated depending on each country's natural resources and the structures of their energy systems, accentuating some of the trends that are already in trend today. For energy importers, accelerating their transition to greater electrification powered by domestic low-carbon energy as they seek to reduce their dependency on imported fossil fuels, perhaps fostering the emergence of new electrostates. In contrast, fossil fuel producers may become more wary of increasing their dependency on imports of low-carbon technologies, preferring instead to concentrate on their comparative advantage in producing fossil fuels. Greater geopolitical fragmentation is likely to lead to greater energy differentiation. The second issue I want to consider is the potential implications of a sustained period of weak improvement in energy efficiency.
I realize at first blush, this may sound a bit dull and arcane, especially relative to all those interesting things I was just talking about, geopolitical things and all that. Stay with me. The recent weakness in energy efficiency, and importantly, its persistence, could have a major bearing on the outlook for energy demand over the next 10 years or so. This chart shows the annual growth in energy efficiency over the past 15 years, measured in terms of final energy consumption. For the first 10 years of this chart, from 2010- 2019, energy efficiency has averaged around 2% a year. That's shown by that dotted line. The way to think about this, each year, the world needed 2% less energy to produce the same level of output. Over the past five years, energy efficiency has only averaged 1.5%.
You can see we've had some particularly weak outlook turns in both 2020 and also more recently in both 2023 and in 2024. The causes of this recent weakness are not fully understood. A recent study by the IEA suggested it may reflect several factors. The increased importance of manufacturing-intensive industries in driving the post-COVID economic recovery in some key emerging economies. Also, perhaps the increasing intensity of extreme weather events and their implications for energy. Think about the increasing need for air conditioning as heat waves get more and more intense. Also, the IEA points to a slowing in investment in energy efficiency projects. I think it's fair to say that the understanding of what caused the recent period of weak efficiency gains is still quite patchy.
As such, it's hard to know if and how quickly the pace of gains in energy efficiency is likely to revert back to something closer to their historical trend. In current trajectory, the pace of efficiency gains, that weakness in efficiency gains gradually dissipates. By around 2030 or so, the trends in current trajectory are back to their normal historical rate. Suppose the recent weakness is more persistent. Suppose, for example, the weakness in energy efficiency persists for five more years and only goes back to the current trajectory profile by 2035. If we hold constant all the other aspects of the energy of current trajectory, just this weaker profile for energy efficiency leads to a materially stronger outlook for energy demand, with total final energy consumption growing by around 20% by 2035, compared with 15% in current trajectory.
Total energy demand growing by over 90 EJ in this alternative case, relative to less than 70 EJ out of 2035 in current trajectory. This stronger energy demand is important in its own right, but its impact on the fuel mix is even more pronounced. This is the key point here. The key point is that short-term cyclical fluctuations in energy demand are typically largely met by changes in fossil fuels and not by movements in renewables and non-fossil fuels. For those of you who like to think in statistics, the correlation between fluctuations in energy demand and fossil fuels is over 0.9. In contrast, for non-fossil fuels, it's around 0.1. If you think about it, this greater responsiveness of fossil fuels to demand fluctuations isn't surprising. The cost structure of renewable projects, high levels of upfront capital expenditure, low operating costs, make them less responsive to cyclical fluctuations.
Moreover, there's greater scope to vary the production and storage levels of fossil fuels over relatively short periods. This greater responsiveness of fossil fuels to demand fluctuations means that even short-lived variations in energy efficiency can have significant implications for fossil fuel demand. To illustrate this, in the sensitivity, we take the limiting case and we say that all the additional energy demand implied by the slower efficiency gains is met by fossil fuels, and we keep the growth of non-fossil fuels completely unchanged from current trajectory. Now, that's obviously a highly stylized assumption, but it captures the essence of those correlations I just talked to you about, and it highlights a potential significance that even a relatively short period of weak energy efficiency can have a big impact for fossil fuel demand.
In particular, in this alternative case, oil demand at 2035 grows by around 6 million bbl a day. That compares to almost no growth in current trajectory. You remember that story that Gareth was telling you about, the current trajectory for oil demand, broadly flat between 2023 and 2035, no growth in demand. 6 million bbl a day growth in oil demand just through that weaker profile for energy efficiency. Everything else unchanged. Likewise, for gas demand growing by over 1,000 BCM in the alternative case compared to less than 700 in current trajectory. The key takeaway here is that a weaker profile for energy efficiency, even for just five years, relative to the counterfactual, could lead to a significantly stronger outlook for oil and natural gas, indeed coal as well, with a corresponding deterioration in the outlook for carbon emissions.
Trends in energy efficiency may sound a bit dull and techy, but they really matter. Let me conclude. The energy system is currently consuming more of all types of energy. In the language that Gareth used earlier, it remains in the energy addition phase of the energy transition. How quickly it moves to energy substitution with the growth of low-carbon energy displacing unabated fuels and the intensity of that substitution is the single biggest uncertainty facing our industry. The scenarios included in this year's Energy Outlook help explore that uncertainty. As Gareth highlighted, this switch from energy addition to energy substitution has already happened in many individual countries, both for the power sector and for energy as a whole. To do so at a global level would be historically unprecedented. That's the challenge.
It's possible to use the scenarios to identify some features of the energy system which are likely to be more robust to a range of different transition pathways. The increasing electrification of the energy system, the rapid growth in wind and solar power, the central role that oil and natural gas will continue to play at least for the next 10 or 15 years. There are many uncertainties. Some of those issues are at the top of our daily newsfeeds, such as the escalation in geopolitical tensions, where, as we argue, greater geopolitical fragmentation is likely to lead to greater energy differentiation. In contrast, others struggle to even make it to the inside pages of specialist trade journals. As we show, trends in energy efficiency and indeed petrochemical feedstocks really matter.
The one thing I know for sure, fast forward a year from now, there will be plenty for the 2026 Energy Outlook to get its teeth into. Thank you very much. Okay, before we move to the Q&A, I want to briefly take you through the online poll. For those of you watching online, three multiple-choice questions will appear on your screens. With apologies for those in the room, we thought about phones and QR codes and thought this is going to be disastrous. Apologies to those in the room. For those online, I will talk you through each question, and for those in the room, I'll talk you through each question. You will have a minute or two to answer that question, and then we'll move on to the next one. We'll share the results of the poll at the end of the Q&A session. Easy.
What could possibly go wrong? A real poll, thousands of people globally, easy. No problem. Okay, question number one. What do you think will be the most important feature shaping the global energy system in the next 10 years? We've given you five possible answers. Four of those relate to the issues that we've discussed today: energy efficiency and its implications for energy demand, geopolitical fragmentation and the impact this may have on different countries' energy choices, the continuing electrification of energy systems, or the impact of AI. We've also given you a fifth option, option E, which is something else. There's something else which we haven't talked about this morning, this afternoon. The most important issue is shaping the global energy system in the next 10 years. Five possible options: energy efficiency, geopolitical fragmentation, continuing electrification, the impact of AI, or something else.
Okay, so hopefully, those of you who haven't, you've made your choice because you're about to lose your opportunity because we're going to go to question number two. Question number two is, what do you think the natural gas demand will be in 2050? Or where do you think natural gas demand will be in 2050 relative to today's level? We've given you five choices ranging from a lot higher to a lot lower. We thought this would be interesting because, as Gareth showed, the range of possible outcomes for natural gas is particularly wide. Your choice may depend on your views of the resilience of natural gas demand in different parts of the world, or it may also be influenced by your view of the slightly speed of the energy transition.
You can remember that Gareth told you in current trajectory, with a slow, shallow trajectory, the gas demand continued to rise pretty much all the way out to the mid-2040s. In contrast, in that rapid decarbonization scenario, natural gas starting to decline from the sort of early 2030s onwards. Your view of the level of natural gas demand in 2050 relative to today's level, five options from a lot higher to a lot lower. Okay, hopefully you've made your choice. Let me go to the third and final question. Following today's session, how have your views changed about the likely pace of the energy transition? We've given you five options here, ranging from a lot faster than you previously thought to a lot slower. If today's discussion hasn't really changed your views at all, you can pick option C that it hasn't really changed my views.
To be clear, this is not about your view of the likely absolute pace of the transition, but rather, was there anything in the discussion or anything that you've seen in the Outlook that has caused you to alter your view? You can pick if it's faster, A and B, slower, D and E, or if it really hasn't changed your view, option C. If I can encourage you to make your final choice, then we'll come back to show you the results at the end of the session. Here, everybody in the room, you can see whether the results are called with your own thoughts as well. Okay, the plan now is to move to the Q&A session, which is always the most fun bit, or the scary bit, depending on which seat you're sitting on.
Gareth's going to come back to the front, and we're also going to be joined by Aisha Dhaliwal. Aisha works in our team and leads all our work on thinking about the analysis of energy in industry. Aisha has kindly agreed to help moderate the Q&A session. Aisha, thank you, and over to you.
Thank you, Spencer. I have to say, as someone who studied chemical engineering, worked in power, then transport, and now industry, I'd say energy efficiency is far from dull. It's very exciting. As you showed in your charts, it's one single measure that can really move the needle when it comes to energy demand totals. We're going to take questions both from inside the room and online. With the questions online, please upvote your favorite questions. The question we ask might not be exactly what you've asked in your books, but it will be a combination of the top voted questions that we've seen. If you're in the room, please put your hand up to ask your question and introduce yourself, your name, and where you work before you ask your question. If there are any questions in the room, we can start here.
Thank you. Thank you, Spencer, Gareth, and team. Thank you for the presentation. I was really drawn into the point on energy efficiency, actually, if I may start there. I know, Spencer, you said that that drop is not well understood, and maybe therefore this is a bit of a loaded question. What is informing your perspective on the path forward, therefore? How do we think about energy efficiency? Is it because of end use? Is it what would drive a recovery back to that historical trend or potentially a path where we remain at sort of low improvement on efficiency?
When we go about producing the Outlook, we don't take a view on energy efficiency in a top-down way. We don't sort of draw a line in and then say that let's do it that way. What we do is we do it from a bottom-up perspective. You would take a view on vehicles. In terms of road transport, we take a view about how quickly vehicle efficiency is improving and, importantly, how quickly the vehicle stock is turning over. If people just hang on to their old cars, it doesn't matter if I got my new cars are really efficient if nobody's buying them. How quickly people will move to energy, to electric cars, which tend to be far more efficient. That's a story for transport and so on and so forth. That's how we built it up.
Then you look from a top-down perspective and draw comfort from the fact that it's going back to around its trend level. That's how we do build that up. I think one of the big uncertainties here is the ones that Gareth was talking about, is how might AI affect everything here in terms of energy efficiency. That's a significant one. A related question is, if we had a silver bit, what are the things that we really worried about that energy efficiency? What should we be really focusing on? Some work, I think some really nice work, I'm not sure if they published it yet, but will do soon, a group called the Energy Transitions Commission. I'm a part of that group, has done some really nice work on something called energy, what they call energy productivity, which is what we would call energy efficiency.
They say if you want to do the two most important things you can do to get improvements in energy efficiency, it is first electrify as many things as you possibly can. Electrification is just a far more efficient way of powering most goods. Secondly, it's to improve the efficiency of electrical applications. Those two things, in terms of their calculations, are by far more important than, say, installation of buildings and so on. All those other things matter, but those are the two things that the ETC highlight as being important.
Thank you very much. Our next question comes from online. The question is to do with the transition and actually goes nicely with the last question in the poll. What do you think, what's your view on the speed of the energy transition and how do you think it's changed in recent years?
Shall I grab this one?
Go for it.
How do we think what's happened in the energy transition and what's happened in the last few years? If you look at the headline numbers, the ones that really matter, clearly progress has been disappointing. Carbon emissions are still going up. Moreover, they're going up at quite significant levels. In the Statistical Review, which is now produced by the Energy Institute, they calculate, I think, that carbon emissions grew by about 1.8% in 2023, another percent in 2024. These are really significant levels. The counterpart to that is that fossil fuels, oil, natural gas, coal, have all been surprising us to the upside. We've revised our views of these different profiles. That's a counterpart to those things. The interesting question is why and the cause, which I think is the more interesting bit.
Because at the same time that we've been surprised by the growth of oil, natural gas, and coal, we've also been surprised by low-carbon energy. Wind and solar has surprised us to the upside. The IEA estimates that this year the world will invest $2.2 trillion in low-carbon energy. That's up 70% in five years. Did anybody really think it was going to go quicker than 70% in five years? At this point, you're scratching your head and going, hold on. If low-carbon energy is growing more quickly than I thought, how can fossil fuels be growing more quickly than I thought? Now you know why, because energy efficiency has surprised us on the downside. The world just needed more energy. Despite low-carbon energy growing more quickly than we expected, it's still not feeding through to the other side.
Just as a sort of back of the envelope calculation here, just to weigh us thinking about this, with all the caveats I gave, if I do a same sort of calculation as I did going forward for five years over the past five years and say, let's suppose energy efficiency had averaged 2% over the last five years, and then I allocated it across the fossil fuels in that way. Oil demand over the last five years, 2019- 2024, has grown by about 2.5 million bbl a day. That's what we've observed. Under that sort of counterfactual, oil demand would have grown by less than 1 million bbl a day over five years. A completely different sort of story. The answer to the question is, have we been disappointed by the energy transition? Yes, carbon emissions continue to rise.
When thinking about why, I think the key thing is, I think some of the narratives I see don't really pay enough attention to the why. The key thing is what's happened to energy efficiency. That's why we have labored, and I promise we're talking about energy efficiency now, that's why we've labored it so much today.
Decisions as we navigate the energy transition.
No, that's all.
Oh, I see a question in the back.
Hello, Spencer , Martin Haigh from Shell Scenarios. I very much liked and enjoyed your exposition about the fragmentation and how to assess that. I thought that was quite novel to try and break down the impacts actually from GDP concerns and security concerns. I'm wondering how you did it. I can see how you do the GDP bit by looking at the responsiveness to GDP in the past, whether short or long-term recessions or long-term cycles. With the security concern, did you look at past instance? Did you try to validate it in a way? Was it a sort of an expert judgment? You know how much fragmentation, how much security concern did you you know because there could be some countries that really perceive a really existential threat immediately, and it puts them onto a sort of a war footing that means all sort of historical parallels are out of track. Your sort of numbers don't surprise me, but they don't look like that sort of extent of security concern, for example.
Shall I do this one? You know sometimes you get those things where one of your teachers is asking you a question, you're really scared. Martin works for the Shell Scenarios team, and we're all in awe of the Shell Scenarios team. I'm now feeling very worried about answering this question. The truthful answer from Martin is with a lot of judgment and a lot of humility, and I'll talk, I won't bore everybody. We made different judgments on the different bits. What we tried to do was give enough to calibrate it in such a way that the effects would shine, would come out, but not in such a way that would be implausible. What we were trying to do here is not, and I didn't, I don't think I quoted any number during those sessions. I just quoted sort of qualitative changes.
I think in some sense, we don't want to sort of stand and fall on the precise numbers. What we want to do is highlight those differential impacts. I wanted, when the calibration was done, enough to try to bring those effects out so we could have that conversation. I hadn't thought about natural gas being caught up in that way. It was only when we did the work I saw it. I also hadn't really thought about how at a global level it would all end, sort of wash out, but you saw it in the fast dark. It was done to do that rather than try to have any degree of precision. Is that fair?
Yes, that is fair. Speaking from the nerdy side of it, from helping the person who actually did a lot of the heavy lifting of this is in the room, I'm tempted to drag him up here. I would say, Martin, it wasn't just judgment, the way Spencer describes it, although there's lots of that. We were also trying to use a sort of structure of the global energy system. Thinking about, you know, proxying in the best way we could these kind of channels for, it's as if oil demand is in some ways more expensive. Sorry, oil is a bit more expensive for you if you're an importer of it than otherwise.
It's as if the costs of electricity from renewables are going to rise a bit faster on average, just because you're no longer going for the cheapest cost producer in the case of many countries, you know, because they want to diversify their supply chains or go internally for supply. We were sort of trying to think about what the different kind of correlations between different types of energy in the energy system and thinking about how they operate. As Spencer said, putting a lot of judgment on. Maybe we can compare notes after the session.
You can tell from those answers.
Yes, sorry, everyone.
The relative close proximity to the people doing the work, can't you?
That's a great answer. I think a good conversation for after this.
No one else, just the two of us.
We've had many questions online regarding oil demand and the changes between last year's Outlook and this year's Outlook. One of the big questions is, why have we revised up our oil demand projection compared to last year?
Do I take that?
Yes.
Yeah, it is true. Someone online has been looking at last year's Energy Outlook and this year's Energy Outlook. We tend not to talk about changes year by year because we're trying to take a fresh look and sort of think about things bottom up. It is true, though, for those of you who wanted to go online or look at previous booklets. We've revised up the pathway for oil demand in our current trajectory scenario, which we also had last year, by about 6 million bbl a day by the end of the Outlook period. A sizable increase in our projection for oil demand by the end. I think all of the reasons why are interesting. There's kind of three, and I'll just mention them just to sort of talk about what's informed us over the past year.
One of those is just the data, some of the things that Spencer's been talking about. We have seen stronger oil demand than perhaps we expected in the near term, and we expect that to sort of persist for a bit. Some of it is just what we have seen. I think that kind of comes back again. Spencer promised we wouldn't talk about energy efficiency anymore. I will just mention it. It's one of the things that's happened. The second one is a sort of part of that in a way, I guess, which is every year we have to, when we do these numbers, and Martin does his in Shell, I'm sure, we're thinking about what's the sort of efficiency of the vehicle fleet in particular going to be. We kind of try and model that bottom up.
We've got a big model of the entire vehicle stock around the world. We've actually revised down the projection for improvements in vehicle efficiency gain over the next 25 years compared with what we had last year. That's not to say it's not still really important. As I said when I was talking, that's still a really important factor that actually pushes down on oil demand in road transportation, but a bit less than we had in last year. That's because, in part, it looks like vehicles are turning over less quickly than we had thought. We've done some analysis looking at that. The person who's in the room has done it. I apologize if I get numbers wrong. Average age of vehicles has risen from 12 years to about 15 years over the last few years. It's quite a big change, actually. That's
pushed down a bit on the speed with which we think those efficiency gains come in. The third one is actually something else I mentioned, which is feedstock demand. We are forever thinking about this. I talked about it in the presentation. I think it's more important than many people realize. We actually have pushed up on our estimate of feedstock demand in this projection relative to the one we did a year ago. The second part of that question, which was, was there other ways in which our demand might keep rising over the 2040.
To 2040 and beyond?
I think it's worth coming back to that feedstock thing, which I mentioned, which is... I think I promised you I wouldn't go into the maths. I will try and do this in a way that does not make you all stampede for the exit and switch off online. This all comes down, or a lot... Oh my goodness, I've lost something. This all comes down to, as we get richer, basically, how much more stuff that is made from oil do we want? A large part of this is how much more plastics does the world use as economic growth happens? That relationship has been changing. It's really interesting. What's happened over the last few years is it's been getting less responsive, the amount of plastics we buy, relative to how much our economy grows.
You need to think about how that relationship is going to change in the future. One thing that could easily push up on oil demand is if it turns out that actually that relationship goes back to something more like its historical trend. That's where you get nearly 10 million bbl a day extra oil demand by 2050. It's a huge fall in demand. That's one thing that could push up. Another one is we have to take a judgment on future mandates and policies around road transportation. As Spencer said, in our current trajectory pathway, we don't just look at policies currently in place, but also the direction in which they're traveling. We assume in many places that those stick. Who knows, if mandates change, that could be another thing that could push up on oil demand.
As we said, current trajectory is our best attempt to model the pathway we're currently on. It's not a forecast. There are many things which could push up on oil demand relative to that. I think that's it.
Oh, got a question at the front.
Hi, Paul Butterworth from CRU. You started off with some emission trajectories, current trajectory and two degrees. I didn't know whether that was a start point from your analysis or an output from your analysis, but I'm sort of assuming it's a start point. I mean, you derive your... If it's an output, the same question applies. You've mentioned petrochemicals being really important. What's your basic question? What's your assumption about the carbon in petrochemicals? Does it ultimately get emitted or is it never emitted? I could ask the same question about natural gas, I suppose. Is it just the direct emissions you're thinking about or is it the upstream emissions and everything else? That might affect, obviously, what the trajectory is.
Yeah, yeah. The answer, Paul, to the first question is, we use, that sounds like I'm going to go into defense straight away. The carbon accounting methodology we use is sort of similar to all of these frameworks, and all of these frameworks are flawed because essentially it's used at the point where the energy is used. At the point where energy is used as a feedstock, it's not combusted, and therefore there's no carbon emissions. Implicitly, it's therefore assuming that at some point at the end of that plastic's life, it's sort of stored in some sort of perfectly sealed landfill, which is clearly not right. It doesn't take account of that.
We do take account, however, of, I think we've related to your other one, if we take maybe we include estimates of methane emissions associated with the production, transportation, and distribution of fossil fuels, and that is included in our measure of emissions. That bit is there.
We also include process emissions from things like cement.
S o.
So energy and the lime.
Brilliant. I've got a question online, and then I've seen some more hands in the room, so don't worry, we'll get to you. Spencer, what surprised you most over the course of your time here at BP?
You took that one. Yeah. COVID, war in Ukraine, OPEC behavior. I think actually the thing which has surprised me most of all, I think, is just the pace of change in China is the thing which I have been most surprised about. China today accounts for about 60% of the world's purchases of EVs. It accounts for about 65% of new solar installations, about 65% of new wind installations, about 35% of new nuclear installations. We're now talking about, carbon emissions in China peaked this year or one year out or two years out. I just don't think that was a part of the conversation 11 years ago. I think that's the biggest change. It's just the unbelievable pace of change that we see in China. I must mean one of the unbelievably lovely things about this job is you get to travel and meet people.
The importance that China has placed on changing the nature of their energy mix has been a consistent narrative the whole of the time I've been going there over that 10 or 11 years. I think that is the single biggest surprise, I think, for me over this period.
I think many can agree. A question from the room.
There's a question right at the back, I can see.
Yes, curious to how the investment in the grid, your views on that play into the outlook. Clearly, a lot of renewables today are delayed because of congestion and grid. Battery technology is coming, but it's not quite there yet. Just curious as to what you assumed about grid infrastructure in the Energy Outlook.
Should I take that one?
Yeah.
Yeah. We recognize, you know, I talked about this electrification and that enormous growth in power demand, you know, 100% or 130% in below two. We recognize that grids, I think I mentioned it as an aside, that there are significant challenges associated with essentially meeting this growth in power demand. I think we recognize that is going to be among the headwinds in current trajectory in particular, that it's going to be something that is restraining the pace of power demand over the next decade at least. Fundamentally, I think we assume that there is going to need to be a significant increase in grid investment over the coming years. We don't model every bit of that bottom up in every country. I think in some ways it's helpful to look at others.
I know BNEF, Bloomberg New Energy Finance and the IEA have done some good work on this, which kind of talks about something up to a doubling in grid investment relative to the kind of $300 billion, $400 billion a year that we've been seeing recently. We have seen an increase, but I think implicit in our current trajectory pathway is a significant further increase, whether that's a doubling or something like that from the grid investment we've seen over the last few years, and therefore by implication stronger than that in below two, higher than that. I think that's the way I think about it. It's a headwind and it's a challenge, but it's one that we think and assume is met in order to meet the pathways in there.
Sticking with the power theme, we've had a question online regarding nuclear. Do you have any comments on the role of nuclear and the trends in nuclear and the energy system?
Sure. I sort of almost felt apologetic that we didn't say a word about nuclear in the presentation. I know that people are always really interested in it. It's just I would have made you sit there for even longer if we'd also talked about nuclear. Obviously, we do model nuclear and it's in there and we talk about it. Actually, there's a couple of pages in the outlook which talk about our assumptions and what we have in the two trajectories. To summarize it, basically, nuclear power has actually been falling as a share of generation over the past few years, and we do see something of a turnaround in our current trajectory pathway for nuclear power. It essentially holds steady as a share of generation over the pathway in current trajectory.
That means it has to go up quite a lot because power generation is going up quite a lot. It doesn't do more than that though, so it stays at around 8%-9% of total generation over the outlook. I would just say, I suppose the only other thing I'd mention on that is a lot of that is China. Again, coming back to Spencer's last answer, that is responsible for quite a lot of the increase in nuclear generation. It's a lot harder in developed economies given timelines, permitting processes you all know about. Also, the fact that I don't think costs are completely transparent in China sometimes, but they are managing to develop nuclear more quickly and more cheaply than in Western economies. More than half of the increase in nuclear over the future is just taking place in China, and quite a chunk in India as well, actually.
That's brilliant. Oh, we've got two questions at the back.
Hello, I'm Elena Pravettoni from the Energy Transitions Commission. Thank you so much for the presentation and also for the reference to our work. The question I had was about the rate of energy substitution in the countries where that's already happening. I wanted to ask, how much do you see data center demand slowing down that rate, if at all?
Can I take that?
Yes.
It's useful to come back to that. Some of you, the more eagle-eyed, in fact, I'll bet if you're anything like me, there'll be someone at least in this audience who was thinking of asking the question of power sector substitution. Does that sort of dip a bit in the chart I showed before it then goes up again? Part of that is the increasingly strong power demand growth we are seeing. Some countries are sort of hovering a bit between substitution and addition in their power sectors in particular, less so actually for the energy system as a whole. I suppose I'd come back to what I said, I think, in the presentation on the way, which is, you know, data center demand matters a lot for a few places.
It matters a lot for the U.S., and a chunk of that, we talk a little bit about it in the outlook, is going to be met by higher gas generation than I think would otherwise be the case. It's 40%, as I said, I think, of high U.S. power demand. As you know, and I know, that's happening in a place that has not really experienced rising power demand for quite a long time. They've been operating in a world where they haven't had to cope with that. They are now having to cope with increasing power demand. I think for them, something which is adding 40% of the increase in power demand you're facing, that is clearly affecting the energy mix. They are having to devote more natural gas to that, amongst other things. It does make at the margin a difference in some places.
I suppose I would just come back to the point, though, that we do see that substitution rate, that the share of the world that moves into that power substitution, picking up pretty substantially over the coming 10, 15 years, even with the additional boost in data centers. While no one knows, I would say our forecast is just relatively robust for how much we think data centers, that projection in current trajectory is relatively robust for how much we think data centers are going to do. It's material.
I've got another bit just to add from online to that question. Can wind and solar handle the demand required to power the infrastructure for AI? If I may just say a couple of words on that, I think that there is a chance for wind and solar to have a great impact and be very useful when it comes to data centers and powering them, especially when coupled with batteries. We've seen stories of small-scale data centers where they've been coupled with reused and recycled batteries with PV that mean that they can run 24/7 on renewable generation. How well that will be replicated across the world will be interesting to see, but there's definitely use cases where renewables will definitely be powering data centers. We might see gas turbines used to supplement that.
I guess just to add it up, we've got profiles for power demand going into data centers built into these trajectories. Those profiles are pretty much in the pack compared to everybody else's. The big picture point that Gareth said was all of the growth in power demand is met or more than met by wind and solar. It will vary across different countries and different regions. Big picture, we think all of the growth in power demand, including that needed for data centers over the long run over this period, can be met by wind and solar in both of those scenarios. Remember, we can't predict the future, but if you see the same qualitative trend in those two very different ones, it may start to give you more confidence that it may be also apparent in somewhere in between as well.
You really started, can I just jump in because I'm regretting not saying one thing which is on that, converts the question. It's funny, isn't it? I warned you in the presentation, we shouldn't just think about data centers for AI , and then we just talked about data centers for AI. I sort of mentioned in passing that AI may well be really important for how we operate our energy systems. The power system is one of those places in particular where AI could really make quite a difference. We have to keep reminding ourselves this AI is going to be used for something. Some of it may be making videos, but it will also be used commercially if it really is going to grow at these kinds of paces.
I can see some shaking of the head, but you know, it may well be used for some commercially useful applications. One of those may well be enabling us to operate grids much more efficiently, able to absorb lots and lots more smaller assets on the grid, balance demand and supply much better. It may turn out, I can perfectly well imagine that in 10 years' time, when we think about the implications of AI for our power systems, we may well not be talking nearly so much about data centers because we'll be operating them much more efficiently.
Very true. Very true, Gareth. Are there any other questions in the room?
Upstairs. My name is Krassimir Kiriakov. I'm the Head of the Commercial Section of the Bulgarian Embassy, but I have a background in project development and project management, and I dealt a lot with cogeneration. Do you factor this in into your analysis because this is an ongoing interest from investors, or it's too minor to be making impact right now of all the big numbers? Also, data centers, there is technology that uses the heat produced by the computers into cogenerating them into, you know, heating for houses or factories or all these approaches and new technologies. How do they impact the big numbers?
Yes, we do try to take account of cogeneration in different parts, and it's sort of buried within our details of some of our analysis. I don't think it's simple to describe it, but we do try to take account of that cogeneration as part of that as we're building up our picture of the energy system. It often doesn't make the headlines because it doesn't have huge impacts on the energy system, but we try to do it to the extent we can, absolutely.
Our final question. Does your analysis suggest that there's a climate change threshold that forces the world to move from CT to a more rapid transformation?
The simple answer is no, we don't try to predict when things will happen. Remember, we're not in the game of predicting. We're in the game of sort of looking at different scenarios. We do do, like the third sensitivity analysis, which I didn't talk about today, is one similar to what we included in last year's Energy Outlook, which we called a delayed and disorderly scenario. This is just recognizing that the pathway we're on at the moment, that slow shadow pathway, is gradually each year using up carbon budgets. If, for example, the world wanted to stay within a two-degree carbon budget, we show that if you stay on that pathway, you entirely exhaust that carbon budget by 2040.
It's actually a little bit worse than that because if you stay on that pathway and you've exhausted the carbon budget in 2040, you've still got carbon emissions at 30 or 35 Gt. To actually get down to something close to net zero, we do some analysis, and it's in the book, we suggest that you need to move away from the current pathway onto something which is a faster, more rapid decarbonization pathway, similar to below two, by the early 2030s. If you haven't moved off that carbon budget, that pathway onto something quicker by the early 2030s, your ability to stay within a two-degree carbon budget without undertaking a costly or disruptive transition becomes harder and harder. I think the answer is we don't know, we don't try and predict when this will happen.
What we do talk about is recognizing that often when you talk about 2050, it's a bit like my children, my children in their early 20s, and I say, "Guys, pensions." They all look at me and go, "Dad, I'm young. I'm not going to do that now. I'll do it later." That's what I'm a little bit worried about when you have these conversations about carbon. They say 2050, yes, absolutely, get to it soon. It's not 2050 or 2030 saying, "If you're still on this pathway in five years' time, your ability to stay within two degrees is getting harder and harder." That's why this matters. Thank you. Is that the last question? We're going back to the poll. Right. Okay. The poll. We had about 1,300 people apply to the poll.
The first question was, if you mind, mind you, what do you think will be the most important feature shaping the global energy system in the next 10 years? The answer shown here, wow, this is really interesting. You never know what people are going to say. It's so here, pretty much half the people look like they're saying the geopolitical fragmentation they think is going to be one of the most significant factors shaping the energy system. I had no idea what people were going to say. It's then pretty evenly spread across the other three, energy efficiency, electrification, and AI. AI coming forth, I'm quite pleased about that in some respects. I'm also quite pleased about only 3% said something else, which suggests that we haven't completely wasted everybody's time for an hour and a half. That was our answer to number one.
Number two, where do you think natural gas demand will be in 10 years, in 2050 relative to today? Quite evenly spread here, actually between, but I mean, is that not high? Is that, sorry. I've got my numbers here. About 33%, about a third of people thinking it will be a bit lower.
A bit higher.
A bit higher. Okay, the numbers I've got in front of me are wrong here. Here, the biggest number is a bit higher, suggesting a point into a world of a relatively slow energy transition with some people thinking a bit lower. That story here, a bit higher, sort of places greater weight, if you like, on the current trajectory. The third question, following today's session, how have you changed your, how has it changed your view? The majority or the biggest vote, I think it will be a bit slower than I thought before. I'm not sure if that's a good thing or a bad thing. Perhaps they're looking at the energy efficiency and worrying about it. 30 people haven't really changed my view, but the biggest one, 40%, I'm thinking that it will be perhaps a bit slower than they thought before. All right.
Sort of a bit of fun. I also think it's quite interesting as well, those results. Let me stop and thank everybody again. For those of you watching online, I'm reliably informed a pop-up box will pop up on your screens now with a very short survey just sort of saying, what bits did you find interesting today? What bits would you like to change? Please fill that in. Lots of food for thought for Gareth for next year. Far more importantly, all of this information is now available on bp.com. We've only touched a fraction of the type of analysis that's in this year's Energy Outlook. Please go online and start exploring. If you have views, if there are things you disagree with, just let us know. This is a whole part of a continuing dialogue. The same here for people in the room. Please go on bp.com.
For those of you who quite like still the hard copies of the book, the hard copies of the book will be available outside. With that, there's teas and coffees outside for people in London. Thank you very much, and thank you all for your time. Thank you.