Energy Transition Outlook 2017

A global and regional forecast
of the energy transition to 2050

Highlights of our outlook

Forecasted decoupling of energy use from emissions, GDP and population trends

World energy growth has historically gone hand-in-hand with population and economic growth. Not only will energy decouple from carbon in the coming decades, but, in our view, global energy use will peak and slowly decline in the context of continued (but slowing) population and economic growth. 

This is linked to accelerating energy efficiency on a global scale, driven in the main by the growing share of electricity in the energy mix, with losses reduced through the steady uptake of efficient renewable sources.

An equal footing

DNV GL has a strong footing in the fossil and renewable energy industries. 

We are a world leading provider of risk management, assurance and technical advisory services to customers in more than 100 countries. Around 70% of our business is related to energy in one or other form. This outlook draws on DNV GL’s broad involvement across entire energy supply chains, spanning complex offshore infrastructure, onshore oil & gas installations, wind, solar and energy efficiency projects, and electricity transmission and distribution grids. 

The core model development and research for this outlook was conducted by a dedicated Energy Transition research team in our corporate R&D unit. The team relied on input from around one hundred colleagues across our organization, as well as dozens of external experts.

A strategy tool

This outlook, based on our own, independent model of the world’s energy system, was undertaken to aid analysts and decision-makers at our customers’ firms, and other stakeholders in the global energy supply chain. It will also support our own business strategy. 

Our findings suggest that there are immense challenges and opportunities in store for the industries we serve, and we explore these further in three ’industry implications’ supplements to this main publication: 

• Oil and Gas 
• Maritime 
• Renewables, Power, and Energy Use

A central case

Our intention, from the outset, has been to construct what we in DNV GL see as ’a most likely future’ for energy through to 2050. This contrasts with scenario-based approaches. Typically, scenarios are set up to contrast possible futures, for example varying the speed of the transition from the current energy mix to one dominated by renewables. As an organization with equal exposure to both the fossil and renewable energy worlds, our aim has been to produce a objective, balanced view of the future. 

DNV GL strongly supports the Paris Agreement, and the efforts of almost all the world’s countries to limit global warming from pre-industrial levels to well below 2°C. Our outlook does not see the world on track to meet the Paris Agreement climate goal.


DNV GL has independently designed a model of the world’s energy system, depicting globally interconnected demand and supply of energy, within ten regions, and the transport of energy between those regions. The core of this is a system dynamics feedback model, implemented in Stella software. The model incorporates the entire energy system — from source to end-use — and simulates how its components interact. It includes all the main consumers of energy (buildings, industry and transportation) and all the sources supplying the energy. 
By design, the level of detail throughout the model is not uniform. Sectors where DNV GL has strong expertise and large business exposure, like oil & gas and power, are reflected in more detail than the sectors where we have little exposure, like coal. However, sectors critical to the energy transition, such as road transport, are treated more thoroughly than more marginal sectors. 

Regional outlooks

We found it meaningful to produce not just a global outlook, but also to explore regional energy transitions, including inter-regional energy relationships. This provides essential insight for any company which, like our own, operates internationally. 
We have divided the world into 10 regions. Countries included in each of the regions generally share some energy characteristics, and geographical contiguity informed our selection of regions in all but one case — our ’OECD Pacific’ region, which includes Japan, South Korea, Australia and New Zealand.

Future economic growth

Future GDP is a function of population and productivity growth. 

Energy forecasts often take the number of people as a departure point and commonly rely for their projections on the World Population Prospects published biennially by the UN’s Department of Economic and Social Affairs. 

The UN has, however, been criticized for not taking country-specific education levels sufficiently into consideration, which matter for future fertility and mortality trends. For that reason, we prefer the approach used by the IIASA/Wittgenstein Centre for Demography and Global Human Capital in Austria, which specifically considers how urbanization and rising education levels are linked to declining fertility. 

Taking the IIASA models, but adjusting for a faster rate of population growth in Sub-Saharan Africa which lags other regions in terms of socio-economic development, gives us a global population in 2050 of 9.2 billion, some 6% lower than the 2017 UN median forecasts.

Energy at scale

The oil and gas industry normally presents its energy figures in millions of tonnes of oil equivalents (Mtoe), while the power industry uses terawatt-hours (TWh), to describe large amounts of electrical energy. The SI system’s main unit for energy, however, is joules, or rather exajoules (EJ) when it comes to national or global energy statistics, which is also the unit we have chosen in this outlook. The conversion factors we use are: 

1 EJ = 23.88 Mtoe
1 EJ = 277.8 TWh

Another way of understanding joules is in terms of the energy needed per person. The amount of energy used per person today averages 78 gigajoules (GJ) annually. We forecast Europe’s average primary energy use to be 76 GJ per person per year by 2050.

Learning curve effects

The premise behind the notion of ’learning curves’ is that the cost of a technology decreases by a constant fraction with every doubling of installed capacity, owing to the growth in experience, expertise and industrial efficiencies associated with market deployment and ongoing R&D. 

After a decade or two, depending on the region, we see the energy transition gaining a self-reinforcing momentum. This will be the main consequence of interacting cost and technology dynamics that enable low-carbon solutions to stand on their own feet.

Demand 1/3

We estimate total final energy demand by mid-century at 430 exajoules (EJ), up from 400 EJ in 2015. All of the increase will take place in the years prior to 2030, following which demand flattens. This, relatively modest, 7% increase contrasts with the 35% rise in global energy demand that has occurred over the last 15 years. The slow-down in demand growth is related to decelerating population and productivity growth, to faster improvements in energy efficiency, and to electrification, e.g. in heating and transport. 

At first glance, the final energy demand chart (Figure 2) looks deceptively stable across major categories of demand. Transport shows initial growth, but then declines as electrification of the road sub-sector materializes. Our analysis indicates that uptake of EVs will follow an S-shaped curve, resembling the fast transition seen, for example, with digital cameras. 

Demand 2/3

The point where half of all new cars sold are EVs will be reached just after 2025 for Europe, 2030 for North America, OECD Pacific, China and Indian Subcontinent, and 2035 for the rest of the world. 

The buildings and industry/manufacturing sectors both retain a stable share of about 30% of demand through the forecast period. 

The remaining 12% is split between agriculture, forestry, other smaller categories and the non-energy use of fossil fuels (e.g. as feedstock for lubricants, asphalt, and petrochemicals).

Demand 3/3

From the perspective of where the energy is sourced, the demand picture is more dramatic. Although total energy demand is almost flat, there are big changes in its composition. In 2015, electricity represented 18% of the world’s final energy use: by 2050, its share will be 40%, growing from 73 EJ/yr to 170 EJ/yr. Electricity replaces both coal and oil in the final energy demand mix, and the trend of electrification is clear in all regions.

We estimate total final energy demand by mid-century at 430 EJ, up from 400 EJ in 2015. All of this increase will take place in the years prior to 2030, following which demand flattens.


Our forecast shows a more dynamic transition on the supply side of the equation, with renewable energy growth leading the charge. Other rapid changes include shifts in shale gas and falling coal demand in China and several developing countries. 

But the key feature is rapid changes within the electricity mix. To satisfy the 170 EJ demand in 2050, electricity generated will amount to 206 EJ, some 18% higher than the final electricity demand owing to losses in transmission and distribution and the power used by the sector itself. 

A key result from our model is that while demand flattens, the global primary energy supply required to satisfy it will peak within our forecast period. The arbiter of this shift is rising energy efficiency.


Our forecast combines past data with our best judgment to provide expected values for variables, without quantifying uncertainties. We do, however, present sensitivity analyses, which highlight issues that are both uncertain and important. We also analyse uncertainties associated with assumptions that place our outlook at odds with other forecasts. 

The most dramatic changes in energy use come from improvements in energy efficiency. The largest changes in the energy mix come from improved cost learning rates for renewables. Behavioural changes affecting, for example, the rate of uptake of electric vehicles and the electrification of buildings, are also important and can shift the pace of transition considerably. 

None of the sensitivities discussed, however, alter the main conclusion that the world will undoubtedly experience a rapid energy transition, driven by electrification boosted by a strong growth of wind and solar power generation, and also further decarbonization of the energy system, including the decline in coal, oil, and gas, in that order.

Energy efficiency

Our outlook shows that the world’s energy system is highly sensitive to changes in energy efficiency. The world’s energy intensity (units of energy per units of GDP) has been declining on average by 1.4% per year for the last two decades. We find that this rate will almost double to an average annual 2.5% decline. 

The main reason for this is the accelerating electrification of the energy system. Simply put, electricity use is much more efficient with less heat losses than is the case for fossil fuels. This effect is accentuated by ever-more solar and wind generation capacity being added, with only inconsequential energy losses. The efficiency trend will be further boosted by the mainstreaming of EVs, which typically consume only a quarter of the energy used by comparable gasoline-powered vehicles.

There are lower efficiency improvements in aviation and maritime, due to the continued use of internal combustion engines for propulsion.

Nevertheless, we anticipate improvement rates of 2.0%/yr per passenger-trip and 0.8%/yr per tonne-mile, respectively. Our forecast ramp-up of efficiency also relies on automation and digitalization, which enables a host of efficiency enhancements, for example in manufacturing efficiencies and in the energy-efficient design and operation of buildings.

The efficiency of renewable energy

Fossil power plants convert only a portion of their input energy to electricity, as much of the input energy is lost as heat. Though combined heat and power (CHP) plants capture some of this heat for useful purposes, globally such heat losses are enormous. 

In the case of renewable power generation, electricity is generated directly from wind, solar irradiance, and from running or elevated water. Although 100% of the input is not converted into electricity, the electricity generated is considered primary energy according to the physical energy content method (as explained in section 4.2 of our main report), and the wind, sun or water not captured is never counted as part of the energy system. 

Hence, with a growing proportion of renewable power generation in the energy mix, losses to heat in the production of energy will decline. This is a major reason why the world’s primary energy supply will peak before 2030.

Achieving Sustainable Development Goal 7

"Ensuring access to affordable, reliable, sustainable and modern energy for all"

The future we forecast is one where humanity’s energy demand flattens after 2030. We foresee this happening even as the world makes steady positive progress with SDG #7, addressing the energy poverty that afflicts more than one billion people today. Energy demand flattens mainly because the energy intensity of economic activity is decelerating. Less energy is required per person. 

We forecast that the third target in UN Sustainable Development Goal No.7 — to double the rate of improvement in energy efficiency by 2030 — will be met. More specifically, we see energy efficiency doubling from an average of 1.3%/yr over the period 2000-2015 to 2.7%/yr in 2015- 2030.

Energy financing

Looking at overall energy financing needs, we calculate investment in fossil fuels by considering upstream and power-related investments for oil, gas and coal. We estimate that, globally, expenditures for fossil fuels will drop by more than half from around USD3,400 billion/yr today to USD1,500 billion/yr in 2050, while non-fossil energy expenditures show the reverse trend, increasing fivefold from around USD500 billion/yr today to USD2,700 billion/yr in 2050. 

Shifting to renewables, where capital expenditure (capex) is mostly upfront, implies a shift from an energy system with a 60/40 split between opex and capex to one with the inverse split of 40/60. In dollar terms, global opex will decline from about USD2,000 billion/yr in 2015 to USD1,500 billion/yr in 2050. Conversely, capital expenditure increases from USD1,800 billion/yr in 2015 to USD2,600 billion/yr in 2050. The energy transition can be undertaken without a significant increase in overall energy expenditures, which will stay approximately constant over time. With Gross World Product (GWP) increasing by 130% over the next 33 years, total energy expenditure is forecast to fall to less than half of its current share of GWP – from 5% to a little over 2% of GWP.

Climate implications

This outlook is one of the few we know of that predicts that humanity will collectively start using less energy in the coming decades. Even so, the emissions associated with our forecast will not bring the planet within the so-called 2 °C target — the maximum level of warming above pre-industrial levels agreed upon in Paris, 2015. 

CO2 will continue to be emitted to the atmosphere long after 2050. Simple extrapolation suggests that the first emission free year will only occur in 2090. This produces an overshoot, beyond the so-called 2 °C carbon budget, of some 700 Gt CO2. 

We nevertheless hazard an estimate that our forecast points towards 2.5 °C planetary warming by the end of the century. We also explore ways to ’close the gap’ between our forecast and the kind of future envisioned by the parties to the Paris Agreement. However, our main conclusion is that ’closing the gap’ will require a mix of extraordinary measures working in synchrony.

Carbon budget depletion

The carbon budget is an expression of how much carbon can be emitted to the atmosphere while staying within a certain temperature threshold. With climate emissions calculated from our forecast, the 2 °C carbon budget will be emptied by 2041. The 1.5°C carbon budget will be depleted in only 4 years from now, by 2021. 

It must be stressed, as we explain more fully in our outlook, that there are considerable uncertainties involved in estimating carbon budgets. Our estimated remaining carbon budget for a 2 °C warming future — 850 Gt CO2 — is an average value for which there also is significant uncertainty.

The Energy Transition Outlook 2017

At a glance

World primary energy supply is likely to peak before 2030: this will surely be a watershed moment in human history, where collectively we will need less energy to satisfy our demand.

Show presentation

Key conclusions

The future of your industry

Our forecast has major implications for the oil and gas industry, as well as for renewables, power, and energy intensive industries.

Read More