By Michael J. Kelly

I imagine that I have been appointed the first CEO of a new agency set up by the Federal Government of the United States of America with the explicit goal of actually delivering a Net Zero CO2 Emissions Economy by 2050. My first task is to scope the project and to estimate the assets required to succeed. This is the result of that exercise, and includes a discussion of some consequences that flow from the scale and timescale for meeting the target.

Executive summary

The cost to 2050 will comfortably exceed $12T (trillion) for electrification projects and $35T for improving the energy efficiency of buildings, a work-force comparable in size to the health sector will be required for 30 years, including a doubling of the present number of electrical engineers, and the bill of specialist materials is of a size that for the USA alone is several times the global annual production of many key minerals. On the manpower front one will have to rely on the domestic workforce, as everywhere else in the world is working towards the same target. If they were not so working, the value of the USA-specific target is moot. The scale of this project suggests that a war footing and a command economy will be essential, as major cuts to other favoured forms of expenditure, such as health, education and defence, will be needed. Without a detailed roadmap, as exemplified by the International Technology Roadmap for Semiconductors that drove the electronics revolution after 1980, the target is simply unattainable.


Imagine we have a net-zero emissions economy in the USA by 2050. Three very large, interrelated, and multidisciplinary engineering projects will have been completed:

Transport will have been electrified.
Industrial and domestic heat will have been electrified.
The electricity sector – generation, transmission and distribution – will have been greatly expanded in order to cope with the first two projects.

A fourth project is to secure the buy-in of the public for what will be 30 years of social disruption, diminished living standards, and living under a command economy. The successful completion of these projects is necessary to meet the high-level target, but they are not sufficient, as I have not dealt explicitly with agriculture and other matters as described below

Current USA energy consumption

The data in Figure 1 give an indication of the energy used over the months from January 2019 to October 2021 for transport, all heat and electricity (in total and the fossil fuel contribution in the USA.   I have derived this diagram from the US Energy Information Agency data[i].

Throughout the year, the use of transport fuel is approximately constant, whereas heating energy is 75% higher in winter than summer, and much of the base load heat is of industrial origin.     Electricity use peaks in summer caused by the use of air-conditioners for cooling and has a subsidiary peak in winter from heating.

In converting transport energy and heat – currently mostly derived from fossil fuels – to electricity, we will use today’s data, assuming that the growth in demand from population growth will be offset by energy efficiency savings, both at about 10% over the next 30 years. This approximation would have to be revisited in a more detailed analysis than is given here.  Note that about 60% of electricity is provided by fossil fuels and that has to be sourced from renewables and nuclear energy.

Figure 1:  USA Data on monthly energy consumption for electricity generation, ground transportation and all heat provided by fossil fuels and that proportion of electricity generated by fossil fuels.    The average monthly values are 3044 (electricity), 2216 (transport), 2508 (fossil fuel derived heat) and 1781 (fossil fuel derived electricity) all in units of Tera-British-Thermal-units.



Transport energy is 75% of the value of the average of electricity and peaks in summer. It is nearly all provided by fossil fuels at this time.   Because an internal combustion engine converts the energy stored in its fuel into transport motion with an efficiency of about 30%, while electric motors are over 90% efficient at using energy stored in a battery, we will need to increase the electricity supply by about 25% to maintain transport in the USA at today’s level in 2050.

A small part of this transport energy is used for aviation and shipping, the electrification of which is much less advanced than the electrification of ground transport, and will, in the end, be more expensive per journey than implied using aviation fuel and bunker oil today.  The extra cost of alternatives to these fuels is not examined here in detail, and this omission allows us to insist that the estimates below are a lower bound on the total cost of delivering Net Zero. The additional electricity infrastructure required is considered in the third engineering project.


We note from Figure 1 that in summer, the USA uses 75% more energy as heat than as electricity. If this heat was provided by radiant heaters, we would need an extra grid equal to the size of today’s just to keep homes and businesses warm. If we use air-source and ground-source heat pumps, with a coefficient of performance of 3:1 – slightly optimistic given the average quality of the thermal envelope of most buildings is high (given the need to keep heat out in summer and/or heat in in winter), then the extra grid would need only to be 35% the size of the present grid, for the heat element alone. Combining this result with the figures for transport in the last section, the grid in 2050 will prime facie need to be more than 60% greater than its present size. We return to this later. However, it may be possible to reduce the amount of electricity required by further insulating buildings.

The US building stock is made up of 140M housing units, 6.9M commercial buildings and 1.3 industrial buildings, with an estimated floor space of 256B, 97B, and 14B sqft respectively[ii].  The current thermal envelope varies strongly by geographic region, and a national retrofit exercise would have to be delegated to states to cope within variation.  Such a programme could reduce the amount of green electricity needed, but for this exercise only a gross approximation of the project to bring all buildings to have the highest possible thermal envelope is possible.   For example, the cost in the UK can be estimated with reasonable accuracy as there has been a pilot retrofit programme from which the national scale cost is $1T per 15M population[iii], which would cost the US about $22T.  Using independent, but equivalent USA data on deep retrofitting, scaled up to 100% emissions reductions, comes to $20T, a remarkably consistent value[iv].  Given the extent to which cold weather climates are already well insulated, this cost may well halve, but we will need to add the improvement of insulation in hotter regions to reduce the use of air-conditioning, and this will take the cost back up again.   Also US houses are twice the size of UK houses on average which will take the costs higher to of order $35T.   It is a matter of urgency that this estimate be refined based on actual US data on retrofitting of a representative sample of US houses and other buildings.

Industrial heat for the manufacture of steel, cement and other materials has been included above.  Electric arc furnaces will accomplish some of the job of decarbonisation, but the highest temperatures still require fossil fuels.    This latter implies extra costs for reaching net-zero which will need further consideration later.

Electricity infrastructure

The grid needs to be 60% bigger in 2050 than it is currently if the USA economy as we know it now is to continue to function. Clearly, 30 years is also enough time to drive other changes in the economy that may reduce, or, indeed, add to this 1.6 factor.

Taken together, the USA grid has been called the largest machine in the world, 200,000 miles of high-voltage transmission lines and 5.5 million miles of local distribution lines.   Assuming a scaled-up grid to treble its size, we will need to add a further 120,000 miles of transmission line. This last will cost of order $0.6T based on US data on transmission line costs[v].

The 5.5M miles of local distribution lines will have to be upgraded to carry much higher currents.   Most houses in the USA have a main circuit breaker panel that allows between 100A and 200A current into the house, although some new ones are rated at 300A.    The 100A standard was set nearly a century ago when the electric kettle was the largest single appliance drawing 13A.  In a modern all-electric home, some of the new appliances typically draw rather higher currents, such as: ground-source heat pumps may draw 85A on start-up, radiant hobs when starting up draw 37A, fast chargers for electric vehicles draw 46A, and even slow ones may draw 17A, while electric showers draw 46A.   The local wiring in streets and local transformers were all sized to the 100A limit.  Most homes will need an upgraded circuit breaker panel and much local wiring and local substations will need upsizing.   The UK costs[vi] have been estimated in details at £1T, which would scale to of order $6T on a per capita basis.

The new generation of electricity must include decarbonising the 60% of the current grid that is fossil fuelled as of now.  This means that we need 400% of the current non-fossil-fuel grid capacity to be provided by new non-fossil fuel sources.   There is limited capacity for new hydroelectricity and the economics of carbon capture and sequestration is unproven.   From Figure 1, we will have to be able to deliver the peak electricity even at times in winter when local production in the north of both wind and solar electricity is low.   Using a mixture are[vii] of wind (onshore $1600/kW, offshore $6500/kW), solar $1000/kW at the utility level), nuclear ($6000/kW), the capital cost of the new capacity alone is of order $5T.   Note: there is no provision here for storage of electricity at the state level for 3-6 months which would be required.   Storage will be discussed further below.

We have identified $12T as the cost of providing the generation, transmission and distribution of electricity in a net-zero world.   Although not all borne by households, the cost is of order $100,000 per household, plus the battery costs (which would dwarf this sum.   The current hydropower storage would run the USA for a few hours and all battery storage for a few minutes.

Human resources

We now consider the human resource requirements to deliver the target economy. Atkins (A UK engineering firm, private communication) estimate that a $1 billion project in the electrical sector implies about 800 years of professional engineering time and somewhere between 2000 and 3000 years of the time of skilled tradespeople. This amounts to 24 or more engineers and 100 or more skilled tradespeople, employed fulltime for 30 years. Scaling up these figures up for the $12T electricity sector projects just described, we will need 500,000 professional electrical engineers and of order 0.8M skilled people employed full-time for the 30 years to 2050 on just this aspect of the net-zero project.  There are of order 400,000 licenced engineers at present, so we will need to more than double that number to accommodate these projects. Training this many engineers will take time and will therefore hamper progress in the coming decade during a build-up phase, meaning even more will be needed later on.

In the building retrofit sector, a range of skills – from semi-skilled to highly skilled – is required. Based on the budget, we might expect retrofit sector to need a similar workforce, of order 3M people, to deliver everything from the design of individual projects, through the materials supply chain, to the actual retrofitting work. Clearly these are both major perturbations to the national workforce. There are no prior examples of skilled workers being generated and maintained on such a scale over 30 years.

Bill of materials

The actual costs of the materials required are covered above. Here we consider the quantities required. The transition from fossil fuels to renewables is a move from a fuel-intensive energy sector to a materials-intensive energy sector. There is already considerable popular concern about the role of mining in reducing biodiversity, but this problem is about to get much worse.

As an example, a 600-MW combined-cycle gas turbine (CCGT) comprises 300 tonnes of high-performance steels. We would need 360 5-MW wind turbines, each running at a mean 33% efficiency, and a major storage facility alongside to achieve the same continuous 600-MW supply. In fact, since the life of wind turbines at 25 years is less than half that of CCGT turbines with a single life-extension refit, we would actually need 720 of them.

The mass of the nacelle (the turbine at the top of the tower) for a 5-MW wind turbine is comparable[viii] to that of a CCGT. Furthermore, the mass of concrete in the plinth of a single CCGT is comparable to the mass of concrete for the foundations of each onshore wind turbine and much smaller than the concrete and ballast for each offshore turbine. A corollary of the multiplicity of turbines or solar panels is that connecting them to the grid is more materials intensive.

A 1.8-GW nuclear power plant and turbine produce about 1000 W/kg of steel in the combined unit, compared with around 2000 W/kg for a CCGT and 2–3 W/kg from solar panels or wind turbines. These factors, of order 1000, show that the use of high-value materials (steels, silicon and long-life polymers for wind turbine blades) is much more intensive in renewables. This effect is offset somewhat by their fuel-free operation. However, the extraction of oil and gas only has a small impact on the earth’s surface compared with the opencast mining of the minerals used by wind turbines and solar farms.

If Ireland were to convert overnight to an electric vehicle fleet, the materials requirements for the batteries alone, compared with annual production today are estimated, by scaling UK estimate by the population ratio, as[ix]

1M tonnes of cobalt – almost 20 times the annual global production
1.3M tonnes of lithium carbonate (LCE) – over 7 times the annual global production
at least 36K tonnes of neodymium and dysprosium – nearly five times the annual global production of neodymium
10M tonnes of copper – nearly the annual global production in 2018

If the world is to go all-electric in 30 years, we need to convert the USA in 1.6 years, and hence we see the need for a very steep rise in the mining of these materials. Unregulated and child labour is implicated in much mining of cobalt, so there are intense research efforts to replace it without losing too much battery efficiency. Biodiversity will be under even great threat from increased mining.

Energy storage

Fossil fuels are much more effective at storing energy than any known non-nuclear alternatives (Table 1).[x]

One example was prompted by a member of Extinction Rebellion, who assured me that the back-up electricity supply for emergency wards in hospitals would be provided by batteries by 2025. The 100-MW, 128-MWh battery installed by Elon Musk near Adelaide in 2018, at a cost of $100 million, would power the emergency wards – 30% of the total – of Mt Sinai Hospital in New York for 24 hours on a single 80–20% discharge. If a storm took out the transmission lines in the New York for a week, we would need seven such batteries. The back up today is typically provided by diesel generators, which run for as long as there is fuel costing about $0.5M. This means there is a capital cost ratio of 200:1 per day or 1400:1 per week for battery versus diesel. This economic mismatch applies to all other suggested applications of batteries, for example protecting Wall St against blackouts.

There is no short-term likelihood of low-cost large-scale electricity storage. Even hydrogen is very expensive, and the fuel needed to make the hydrogen would be much more effectively used to perform the functions directly that the hydrogen would be scheduled to do.

The global context of USA actions

One can see in Figure 3 the dominant role that fossil fuels have had in energising the world economy since the 19th century. All the efforts on renewables have so far contributed only a slight divergence and fall in the fossil fuel fraction since 1980 – this has been of order 85% for a century, but has fallen to nearer 82% now. An extrapolation out to 2050 indicates a 79% contribution in 2050: there is no sign of a rapid divergence and a zeroing of the fossil fuel fraction in the next 30 years. These and many other developments, such as the quadrupling of the SUV global market in the last decade, all show the world moving away from the net-zero target.

I have made no allowances for radical technological breakthroughs in the energy sector, which might relieve the situation on the timescale of decades. Equally, however, incremental developments, such as those seen in battery technology, might be slower than anticipated, as the intrinsic limits of materials properties are approached. Any such delays would worsen the situation.

Public acceptance

The fourth project listed at the outset may be the hardest. It is clear from the public debate that the citizenry has no idea of the scale of the task of a transition to a net-zero emissions economy in 30 years. This is not only a matter of the costs, human resources and materials, but also the disturbance to everyday lifestyles as the target is approached. Opinion polls indicate that few are willing, let alone able, to pay more than very modest sums, and certainly nothing like that implied by the figure of well over $300,000 per household set out above (for electrical and retrofit actions). Worse, there will be no measurable difference in the future climate as a result of all the spending and hardship in the UK. To make a difference we would need the rest of the world, and in particular the developing world, to come on board. Poorer nations, such as India and the countries of South Asia, the Middle East and Africa, would need financial help to do so. If we assume that Europe and North America are to underwrite the rest of the world’s net-zero activities, then the costs to the UK could rise by a factor of 4.5, assuming the same per capita spend globally.  The resulting cost of getting to the global target then rises to nearly $1.5M per household, and $200T for the whole of the USA, which is a fantasy in practical terms.

By all commonly understood value-for-money measures, climate mitigation exercises simply do not add up. For homes, the $300,000 per household would be recouped almost 100 years (at today’s cost of energy), far longer than any sensible investor would tolerate. Indeed, we would require a command economy during the period to 2050 to secure the finance, skilled workforce, and the materials needed to reach the target. Further, from where we are today, it is not clear how this public acceptance can be achieved on the timescale required.

Funding for adaptation to an actual changing climate is an easier ask. Using the Thames Barrier in London as an example, extensive flooding in the 1953 storms in the East of England triggered the commissioning of various actuarial calculations.   When should a Thames Barrier be constructed such that over its lifetime the value of flood insurance claims avoided was equal to the cost of the barrier itself? The answer was ‘in the 1980s‘.   In developed countries with seismic activity, it is easy to set aside and invest multiple billions of pounds to cover future earthquakes, but that is because most people know they could be claimants during their lifetimes. For the slow-burning issue of climate change, however, this is not possible. Instead, the use of appropriate actuarial calculations could allow investment in adaptation to be attracted as and when necessary.

Spend profile and secured finance

Most of the preceding analysis assumes a constant 30-year project. In practice, however, the spend will start from near zero and ramp up. If a 40-year retrofit roll out had started in 2010, one would by now have spent of order 15–20% of the total improving housing and other buildings. In practice the spent was of order 1%. Each year of delay adds more to what must be achieved in the coming decades, requiring even greater flows of finance, human resources and materials. The training of a skilled workforce and building up the supply chain must precede mass roll-out in all sectors. The expansion of the grid must precede the mass uptake of electric heating and transport: having the cars and heat-pumps without the green electricity is the height of folly.

A project on this scale will need bespoke financing at the national level, as it is beyond the scope even of the richest companies in the world today. Even international money markets would struggle if all the world pursued net zero. Completely new economic thinking would be needed, and the Stern Report of 2006 is way out of its depth on this practical point.

A partial list of factors not yet considered

I have given no attention to agriculture, and especially methane emissions, nor forestry, which permits negative emissions while trees are growing. I have not considered aviation or shipping and specific costs there. Aviation fuel will be with us through and beyond 2050, and evolution of electric shipping is very slow beyond commuter ferries in large city harbours. The global economy depends very much on both these forms of transport, and any severe curtailment will be accompanied by falling standards of living of the middle class.   I have not considered industrial heat currently provided by fossil fuels for which electrical heating does not achieve high enough temperatures in some refining processes.

I have not included the extra costs of simultaneously running the two new infrastructure systems required to support fuelling internal combustion engines and recharging electric motor batteries. I have not considered the practical choices associated with where and how the extra electricity generation should occur, nor have I factored in the costs of any forms of electricity storage (which are very high, as seen earlier). These issues will need an early resolution, because many of the desired outcomes depend on the new infrastructure being in place. I have not examined the ever-growing costs of balancing the grid, costs which grow dramatically as more intermittent sources of electricity are used.

A major change in peoples’ lifestyles, with reductions in travel, consumption, and food variety could make a dent in the numbers above, but not reduce much the scale of the engineering projects.

A roadmap for Net Zero

The success of the IT revolution over the last 40 years is in no small part due to the existence of the International Technology Road Map for Semiconductors (ITRS). Representative engineers from every part of the sector, and all parts of the world, have gathered every two years to thrash out in great detail what needs to come out of the laboratory into development, and out of development into production, to keep Moore’s law of transistor miniaturisation on track, and with it the increase in computing power. Every player in the field knows that the other players are investing and working day-by-day to the same agreed objective.

Note the contrast between ITRS and international climate meetings. Meeting the 2050 net-zero emissions target is much more complex than semiconductor development and can therefore go wrong in many more ways. Despite this, it is being attempted without any kind of roadmap. The project is therefore more likely than not to veer in the direction of the historical Tower of Babel. No engineer would invest time or money in such a project. Investors should expect better given the scale of the enterprise.


With extra costs comfortably in excess of $35T billion, a dedicated and skilled workforce comparable to of that of the education sector, and key strategic materials demanded at many times the supply rates that prevail today, and all for no measurable attributable change in the global climate, the mitigation of climate change via a net-zero emissions USA economy in 2050 is an extremely difficult ask. Without a command economy, the target will certainly not be met.

The practical alternative

Many in the world are convinced that we face a climate catastrophe in the coming decades if this target economy is not delivered. I suggest we are certain to have an economic and societal catastrophe if we persist on the projects to deliver the net-zero economy by 2050. There is a get-out-of-gaol card, and that is the demographic transition, which started 70 years ago. The average family size in the world has halved, from 5 children in 1960 to 2.5 children now, and is continuing to fall. In developed countries, with universal primary education and more people living in cities than the countryside, the figure is below 2, and indigenous populations are in absolute decline, as it takes 2.1 children per family to maintain a population. Stable developing countries, such as Bangladesh and Lesotho, are already down to 2.5. The Chinese population will peak in the 2030s and the world population in the 2060s. A century from now, when we need copper, we will not mine it, but strip it from abandoned cities.

My analysis requires the climate change community to go back, in all humility, and ask themselves really how bad will (as opposed to might) the world’s climate become? The proposed solution seems far worse for society than the problem. Half of their analyses of the future climate are based on a CO2 emissions scenario (RCP8.5) now debunked as excessively high rather then the more likely RCP2.5 scenario. Their candour at this point would assist those making the case for funding climate adaptation, which will only be carried out when it becomes necessary. In the parlance of the Second World War, ‘Is this journey really necessary?’

Personal view

I hope this report gives the bare facts about what is implied by committing to a net-zero emissions economy for 2050. Short of a command economy, it is simply an unattainable pipe dream, and we will struggle to get 10–20% of the way to the target, even with a democratic mandate to proceed. I think that the hard facts should put a stop to urgent mitigation and lead to a focus on adaptation. Mankind has adapted to the climate over recent millennia, and is better equipped than ever to adapt in the coming decades. With respect to sea-level-rise, the Dutch have been showing us the way for centuries. Climate adaptation in the here and now is a much easier sell to the USA citizenry than mitigation. There is a very strong case to repeal the net-zero emissions legislation and replace it with a rather longer time horizon. The continued pressure towards a net-zero economy will become a crime of sedition if the public rise up violently to reject it. The silence of the National  Academies and  the professional science and engineering bodies about these big picture engineering realities is a matter of complicity.


[i] Data from the Energy Information Agency of the USA, with thanks to several members who checked my interpretation of their data to derive Figure 1: all the implications from are by me and they bear no responsibility.    Total Energy Monthly Data – U.S. Energy Information Administration (EIA)

[ii] • Number of homes in U.S. 1975-2020 | Statista   140M units commercial building 5.9M  with 97B sq ft floorspace

Industrial space in the U.S.: total space by type | Statista 10264 msqft warehouse and distribution 3472 msqft manufacturing
Size of new single-family homes in the U.S. | Statista average in 1970 1660 average since 2500.

United States Industrial Properties | Reonomy 1.3M industrial buildings:   3.472M sqft manufacturing and 10264M sqft warehouse and distribution.

Total U.S. home square footage 2015-2023 | Statista 246Bsqft

[iii]     In 2009, as Chief Scientific Advisor to the then Department for Communities and Local Government, I briefed Lord Drayson, the then Science Minister, about the challenge of retrofitting all existing buildings to reduce the energy consumption and hence emissions of carbon dioxide. I suggested a detailed pilot programme be put in train. This became a £17 million expenditure programme called 3 ‘Retrofit for the Future’, a series of projects in which over 100 social houses (i.e. smaller than the average) were subject to various measures. One group of 45 houses received complete makeovers – double and treble glazing, external cladding, extra loft and underfloor insulation, and new energy-efficient appliances. Detailed studies of emissions before and after for this group showed that for an average expenditure of £85,000, the average emissions reduction achieved was 60%, with only three dwellings achieving the 80% emissions reduction target, and another three not even reaching 30%. Linearly scaling the result to the whole housing stock and a 100% emissions reduction, produces a cost estimate of £4 trillion. See the results at:  Rajat Gupta, Matt Gregg, Stephen Passmore and Geoffrey Stevens. ‘Intent and outcomes from the Retrofit for the Future programme: key lessons’, Building Research & Information, 43(4); 435–451, 2015. See

[iv] Report: Deep Retrofits Can Halve Homes’ Energy Use and Emissions | ACEEE

[v] MISO USA: £1.6 million/km for 132kV, £2.0 million/km for 275kV and £3.3 million/km for 400kV line

[vi] The Hidden Cost of Net Zero: Rewiring the UK (

[vii] Cost of electricity by source (per Wikipedia):

gas/oil combined cycle power plant: $1000/kW (2019)
combustion turbine: $710/kW (2020)
onshore wind: $1600/kW (2019)
offshore wind: $6500/kW (2019)
solar PV (fixed): $1060/kW (utility), $1800/kW (2019)
solar PV (tracking): $1130/kW (utility), $2000/kW (2019)
battery storage power: $1380/kW (2020)
conventional hydropower: $2752/kW (2020)
geothermal: $2800/kW (2019)
coal (with SO2 and NOx controls): $3500–3800/kW
advanced nuclear: $6000/kW (2019)
fuel cells: $7200/kW (2019)

[viii] Development of 5-MW Offshore Wind Turbine and 2-MW Floating Offshore Wind Turbine Technology (