Energy security: right to self-determine?

Percent of energy use provided by imports and by self-supplies for the US, Europe, and China. Also includes a breakdown of energy use by type.

The average major economy produces 70% of its own energy and imports the other 30%. This 12-page note explores energy security by country. We draw three key conclusions: into US isolationism; Europe’s survival; and the pace of EV adoption, both in China and in LNG-importing nations.


Energy self-sufficiency can be defined in different ways. Our own preferred definition is explained on page 2. Our cross-plot of energy self-sufficiency by country is also shown on page 2.

This matters because energy is the lifeblood of all economic activity, while import dependency creates vulnerabilities amidst rising geopolitical tensions, as outlined on page 3.

Our first observation is that the US is entering a new era of foreign policy. Whatever the outcome of the upcoming US election, we fear that some commentators will increasingly call for an isolationist stance, for the reasons on pages 4-5.

Our second observation is an abysmal deterioration in Europe’s energy security, falling more than any other region, despite ramping wind and solar. Failure to unlock domestic resources is an existential threat to Europe’s security, and a missed opportunity for ESG oversight, as argued on pages 6-9.

Our third observation is that the rise of electric vehicles is heavily linked to different countries’ energy security. Most notably, China is substituting oil import reliance for domestic coal and renewables. But other regions may have less motivation to ramp electric vehicles, while doing so will largely increase LNG demand, as argued on pages 10-12.

The full note contains around 20 charts tabulating energy self-sufficiency by country and over time, as a useful reference, for this increasingly important metric.


Grid capacity: a wolf at the door?

Anticipated reserve margin according to NERC forecasts from 2014 to 2033. They have been predicting falling margins but instead, they have seemingly been growing.

This 17-page note outlines how capacity markets work, in order to stabilize global power grids. We argue reserve margins in the US grid are not as healthy as they look (in the chart above). Data-centers are like wolves at the door. Capacity prices must rise. This boosts gas plants, grid-scale batteries and non-regulated utilities?


US grid resiliency is supported by c100 Reliability Standards written by NERC, enforced by FERC, and applicable to c100 balancing authorities, c325 transmission operators and c1,000 generation facilities in the US. Each violation can incur penalties up to $1M per day.

A key NERC Standard is to expect no more than 1 major grid outage per 10-year period, which in turn requires keeping reserve margins above 15%. Often quite far above. The evolution of these regulations is explained from first principles on pages 2-3.

The real challenge for resilient power grids is how to unlock sufficient long-term investment, aka the ‘missing money problem’. Different markets have capacity incentives in place, e.g., energy-only, centralized capacity markets, decentralized capacity markets. An overview of how capacity markets work is given on pages 4-5.

There might not seem to be any reason to worry about reserve margins given the title chart above. Yet we are worried (page 6) that demand will surprise to the upside (page 7), anticipated resources may underestimate renewables’ volatility (page 8), planning has become politicized (page 9) and some regions will deteriorate sharply (page 10).

What happens when reserve margins unexpectedly deteriorate? As a playbook, to inform our outlook, we provide some case studies from past data at ERCOT and CAISO on pages 11-13.

A key question in 2024 electricity markets is whether AI data centers could improve their time to market (see our AI video overview) amidst power grid bottlenecks by signing contracts to directly absorb large quantities of pre-existing power plants (e.g. per Amazon-Talen). If you think about how capacity markets work (pages 2-13), then data-centers are like a wolf at the door, for the reasons on page 14-15.

Capacity prices therefore need to increase, in order to safeguard today’s capacity chickens from the wolf at the door. We have attempted to quantify the necessary increases. This will flow through to the cash margins of gas turbines, nuclear facilities, grid-scale batteries, demand shifting contracts and non-regulated utilities, per pages 16-17.

Building energy infrastructure: constructive margin?

Distribution of capex costs of construction, engineering, equipment, materials, and misc for different types of projects. The average for construction costs are 40%.

Energy transition is the largest construction project in history, with capex costs ultimately ramping up to $9trn per year. Overall, 40% of capex costs accrue to construction firms. Hence this 10-page note evaluates energy infrastructure construction companies, their EBIT margin drivers, and who benefits from expanding power grids?


The past five years of research have led us to conclude that achieving net zero by 2050 would effectively require the largest energy infrastructure construction project in the history of human civilization, absorbing $9 trn pa of capital expenditures.

Whether or not you believe the world will hit its decarbonization goals, a construction boom increasingly seems to be imminent, linked to power grid bottlenecks and the rise of AI. Our quantifications of these rising capex costs are on pages 2-3.

Overall, 40% of capex costs accrue to construction firms, across different project types captured in our economic models. Some of our favorite examples, and different projects, are discussed on pages 4-5.

In particular, building in boom times can result in projects costing 2-3x more than building in normal times, per a note from 1Q24 that is worth re-visiting alongside this work, highlighting the value of engaging high-quality construction firms, and re-capped on page 6.

Hence we have screened 25 of the largest construction companies in the world. Generally, across our screen, we found that the average company has 100 years of operating history, employs 35,000 people, and generates $13bn pa of revenues.

Across these energy infrastructure construction companies, we evaluated which factors have impacted EBIT margins, on pages 7-8.

Specific companies include those with the most extensive history of delivering mega-projects across different sectors, those with the most specialized skill-sets, and those with particular specialization into power grid infrastructure. These companies are profiled on pages 9-10.

Back up: does ramping renewables displace gas?

Comparison of the same Australian gas plants in May 2014 and May 2024. The increasing share of renewables reduces the utilization of baseload gas plants and turns them into peaker plants.

This 12-page note studies gas power plant generation profiles, across 10 of the largest gas plants in Australia, at 5-minute intervals, comparing 2024 versus 2014, amidst the rise of wind and solar. Ramping renewables to c30% of Australia’s electricity mix has not only entrenched gas-fired back-up generation, but actually increased the need for peakers?


Australia’s electricity demand has increased at 0.9% per year in the past decade, reaching 273 TWH in 2023. 46% is still powered by coal, while 17% comes from gas.

Australia’s renewables generation has increased. Wind power generation has ramped 3x from 4% of Australia’s grid in 2013 to 12% in 2023, while solar has ramped 8x from 2% to 16%. For the data, please see our overview of each country’s grid mix.

So, what has happened to gas power plant generation profiles amidst the rise of renewables? To answer this question, we have tabulated 5-minute by 5-minute generation data, from AEMO, for ten of the largest gas power plants in Australia, in 2014 and 2024.

Gas is clearly backstopping the volatility of solar, per pages 2-4.

Hence, we have drawn five conclusions about the utilization rates of gas turbines, generation in MWH and capacity requirements in MW on pages 5-9 of the report (1-2 key charts support each conclusion).

Utilization rates have fallen at baseload gas generation facilities, which is inflationary. But utilization rates have increased at peaker plants.

It looks challenging to fully replace the flexibility of these gas turbines with grid-scale batteries, for the reasons on pages 9-10.

Conclusions, predictions and implications are summarized on page 11.

Power grids are truly amazing in their complexity, but tend to get over-simplified in energy transition roadmaps. In the words of Ludwig Wittgenstein, “don’t think, but look”.

Wittgenstein was probably not talking about gas power plant generation profiles. Yet our data show how ramping renewables have entrenched pre-existing gas turbines in the grid in Australia — similar to other analysis we have conducted in California, the UK and Europe — and arguably require c30% more peaker plants than a decade ago.

Energy transition in 1H24: 101 companies and the rise of AI?

Companies discussed in Thunder Said Energy research from 2019 to 2024.

This 13-page note summarizes the key conclusions across all of our research from 1H24, concisely, for busy decision-makers. We highlight 101 companies exposed to AI, which have come up in our recent work, to enable the rise of AI, and debottleneck its electricity supplies, out of 1,500 companies that have now crossed our screens overall.


1,500 companies have been mentioned 2,500 times in our research since 2019, and our energy transition research now includes over 1,400 research notes, data-files and models. Hence this report is part of a quarterly series summarizing the key conclusions across our work.

In 1H24, the #1 theme that has excited the entire energy world has been the rise of AI. We estimate 150GW of AI data centers will be built by 2030. 40% will be in the US. This will underlie the largest period of new generation capacity growth in history.

Companies exposed to AI have thus featured heavily in our 1H24 research, as we tried to unravel the bottlenecks across capital goods, more capital goods, energy, utilities, infrastructure and materials.

In particular, we reached 15 thematic conclusions, and specific companies stood out alongside each conclusion, as outlined on pages 3-8.

We are less worried about materials bottlenecks biting in 2024 than we have been at other points in the past, but still excited by advanced materials, for the reasons on page 9.

Our energy outlook is more balanced than at other points in the past, albeit we think LNG will surprise to the upside and there is growing value in volatility, per page 10.

The most mentioned companies in our research in 1H24, and from 2019-2024 more broadly, are profiled on pages 11-13. Fourteen companies stand out, with angles that may be interesting to explore further.

The downside of a concise, 13-page report, is that it cannot possibly do justice to the depth and complexity of these topics. A TSE subscription covers access to all of the underlying research and data.

We are also delighted to elaborate on our energy transition conclusions from 1Q24, and discuss them with TSE clients, either over email or over a call.

Solar Superpowers: ten qualities?

Solar ramps from 6% of global electricity in 2023, to 35% in 2050. But could any regions become Solar Superpowers and reach 50% solar in their grids? And which regions will deploy most solar? This 15-page note proposes ten criteria and ranks 30 countries. The biggest surprises will be due to capital costs, grid bottlenecks and pragmatic backups.


We are amending our solar growth forecasts in this research note, using ten criteria to rank the long-term competitiveness of solar in different nations, then screening 30 different nations globally.

Solar Superpowers could be capable of ramping solar to 50% of their power grids by 2050, but we think this would require meeting ten criteria…

Criteria #1-2 relate to solar insolation, and are explored on pages 3-4.

Criteria #3-4 relate to costs, impacting both the absolute levelized costs of solar, and the cost relative to other regional power prices.

Capital costs are among the most overlooked variables. Solar projects in Switzerland can be lower cost than in the sunny Sahara. See pages 5-6.

Criteria #5-6 are linked to solar siting. Further away from major demand centers, land is more available, and at lower costs. But this is counterbalanced by the cost to move power over longer distances, especially amidst grid bottlenecks. See the charts on pages 7-8.

Criteria #7-8 relate to grid integration and backups. These challenges present the greatest risks that could cause developed world countries to fall short of their solar ambitions, per pages 9-10.

Criteria #9-10 relate to incentives and competition, per pages 11-12.

Solar Superpower Scores are calculated for 30 countries, which comprised 75% of global electricity demand in 2023, and derived an average of 6% of their total power generation from solar over 2023. Results and surprises are discussed on pages 13-14.

The ultimate share of solar might reach 50% for Solar Superpowers, but we do not see any countries reaching this threshold. Australia and California come close. Different regions are discussed on page 15.

Energy trading: value in volatility?

The statistical distribution of commodity prices follows a lognormal curve. Increasing volatility will drive up mean prices and increase the value of arbitrage.

Could renewables increase hydrocarbon realizations? Or possibly even double the value in flexible LNG portfolios? Our reasoning includes rising regional arbitrages, and growing volatility amidst lognormal price distributions (i.e., prices deviate more to the upside than the downside). This 14-page note explores the upside for energy trading in the energy transition. What implications and who benefits?


Global energy markets are growing increasingly volatile, as argued in our January-2024 quantification of energy market volatility. Key reasons are the volatility of wind and solar, which are gaining share in the global energy system, as re-capped on pages 2-3.

In order to assess commodity price volatility, we have tabulated the statistical distributions of commodity prices, for a dozen major commodities, over the past 50-years. More volatile commodities generally have higher mean average prices, as shown on pages 4-5.

In order to model the impacts of rising volatility upon commodity prices, we need to fit a statistical distribution onto the commodity prices. Lognormal distributions provide a beautiful fit. Our confidence intervals for oil prices, gas prices and coal prices are outlined on pages 6-7.

How are commodity price realizations impacted by rising volatility? Mean outcomes exceed median-basis forecasts by a wider margin! The impacts of rising volatility on gas prices, coal prices and oil prices are quantified on pages 8-9.

How are the commodity price realizations available to energy trading businesses impacted by rising regional volatility? We argue that arbitrage potential will widen most in global LNG markets. Energy trading profits are often treated as one-offs. But they are one-offs that will tend to recur every quarter. We have quantified the growing value of diverting flexible LNG contracts to higher value markets on pages 10-13.

LNG portfolios are most likely to benefit from rising volatility in the global energy system, as price spikes become more frequent, and higher prices are also required to mobilize a limited number of truly flexible LNG cargoes. 25 companies’ LNG portfolios are assessed on page 14.

We think this report should be required reading for anyone with commodity-exposed interests. Commodities are volatile. But as the energy transition progresses, there is value in volatility.

Superconductors: distribution class?

Illustration of a cable made with high-temperature superconducting tape.

High-temperature superconductors (HTSs) carry 20,000x more current than copper, with almost no electrical resistance. They must be cooled to -200ºC. So costs have been high at 35 past projects. Yet this 16-page report explores whether HTS cables will now accelerate to defray power grid bottlenecks? And who benefits within the supply chain?


Superconductivity is a form of quantum magic, where particular materials show almost no resistance to electrical currents, once their temperature drops below some critical transition temperature.

Hence these materials can theoretically carry infinite quantities of current. The weird quantum effects that give rise to superconductivity are briefly described on page 2.

Hundreds of materials have been shown to demonstrate superconductivity, since the effect was first discovered in mercury in 1911. Some of the key materials, such as Nb3Sn, NbTi, BSCCO and YBCO (REBCOs), are summarized on pages 3-4.

The reason for writing this report is our growing fear over power grid bottlenecks as the biggest bottleneck in energy markets and for energy transition. We have already explored advanced conductors to debottleneck the overhead transmission network.

Yet one of the biggest unsolved challenges is expanding the distribution network in space-constrained urban environments. We outline how high-temperature superconductors could help on pages 5-7.

Superconductors have already been piloted in global power grids, with 35 past projects going back to 2000. So what costs and other details from past superconductor projects stand out on pages 8-9?

The costs of HTS cables are compared with costs of transmission and costs of distribution at conventional projects — both on a top-down and bottom-up basis — on pages 10-12.

Material implications of high-temperature superconductors are also explored, for materials such as silver, superalloys, yttrium, helium and for displacing copper on page 13.

Leading companies in superconductors include six large global producers, including leaders listed in Europe, the US and Japan, plus interesting private companies scaling up capacity. Conclusions from our superconductor company screen are reviewed in pages 14-16.

Finally, there are reasons to wonder whether higher-temperature or even room-temperature superconductors might be developed in the future, from the multi-billion member state space of possible candidates across materials science. We have predicted that AI will ultimately earn its keep by ‘figuring out’ state spaces too complex for human brains.

But in the mid-late 2020s, the most interesting angle is that we think YBCO HTSs will play an increasingly large role helping to debottleneck the distribution network, especially in space-constrained urban environments. Could project activity accelerate by 5-50x by 2030?

Low-carbon baseload: walking through fire?

This 16-page report appraises 30 different options for low-carbon, round-the-clock power generation. Their costs range from 6-60 c/kWh. We also consider true CO2 intensity, time-to-market, land use, scalability and power quality. Seven insights follow for powering new grid loads, especially AI data-centers.


Today we are increasingly receiving questions from clients looking to self-generate electricity, for large new loads, while also avoiding power grid bottlenecks. There is especially sharp demand to power new data-centers amidst the rise of AI.

Hence this report aims to compile the most extensive cross-comparison we have attempted to-date, into different sources of low-carbon baseload. We assessed 30 options across 20 different dimensions, in our LCOE database. Our methodology is described on pages 2-3.

The costs of low-carbon baseload range from 6 – 60 c/kWh. Numbers, sensitivities, capex costs, true CO2 intensities, construction times, transmission requirements, land intensity, scalability, ramp rates and reliability are all cross-plotted in the charts on pages 4-8.

Gas value chains are lowest-cost overall in the US, especially when developed directly in shale basins. Observations, discussion points and US gas market conclusions are summarized on pages 9-10.

Pure wind and solar value chains cost an order of magnitude more, when they are required to generate round-the-clock power (defined as having 3 days of battery coverage on cloudy/non-windy days). Observations and discussion points are on pages 11-13.

There are also options that use zero-hydrocarbons. They include blending wind and solar with pre-existing hydro, incubating next-generation nuclear, or housing AI data-centers alongside Iceland’s geothermal hotspots and then moving the data (!). See pages 14-15.

There is no perfect solution, however, on the quest for rapidly scalable low-carbon baseload. Hence, we close by considering whether this will delay the rise of AI, or even entrench high-carbon generation sources that would otherwise be phased out. Different options for generating low-carbon baseload reward careful consideration.

Moving targets: molecules, electrons or data ?!

New AI data-centers are facing bottlenecked power grids. Hence this 15-page note compares the costs of constructing new power lines, gas pipelines or fiber optic links for GW-scale computing. The latter is best. Latency is a non-issue. This work suggests the best locations for AI data-centers and shapes the future of US shale, midstream and fiber-optics?


One of the biggest questions in energy markets this year concerns the rise of AI. Specifically, how is the world going to electrify a possible 150GW of new AI data-centers by 2030, amidst bottlenecked power grids and bottlenecked gas grids?

This 15-page note considers the options for moving molecules, electrons and information to and from AI data-centers. Ultimately, the best locations for AI data-centers will offer the best service, at the lowest cost, and after some acceptably low lead-time.

Mostly moving electricity to an AI data-center requires constructing new AC transmission lines, or possibly even HVDCs. This turns out to be the most costly option and has the highest lead-time. Capex costs (in $M/km), total costs (in c/kWh) and logistical conclusions are on pages 2-3.

Mostly moving gas to an AI data-center requires constructing new gas pipelines, but has the advantage of alleviating power grid bottlenecks, by self-generating power on site. This is a high-cost option in $M/km terms, but a low cost in c/kWh terms, and must also overcome logistical challenges, as discussed on pages 4-5.

Mostly moving data from an AI data-center requires constructing new fiber optic links, but has the advantage of alleviating power grid bottlenecks and gas infrastructure bottlenecks, by siting the data-center within a US shale basin. This has by far the lowest costs in $M/km terms and c/kWh terms, plus other logistical advantages, per pages 6-9.

Latency myths. The pushback we are anticipating is that an AI data-center needs to be close to end-consumers, because otherwise the latencies on AI query responses will be overly high. This is simply untrue. Data and counter-arguments are outlined on pages 10-12.

Locations for AI data-centers may largely be determined by the locations of upstream gas. Further help may come from a cooler, wetter climate, lowering the energy and financial costs of data-center cooling.

There is a strange symmetry bringing the shale and AI industries together: the former suffering bottlenecks in energy demand and export infrastructure; the latter suffering bottlenecks in supply and import infrastructure. Conclusions for shale and midstream are on pages 13-14.

The fiber optic industry also sees large growth, hence we end by noting the market size, growth, material usage and leading companies on page 15.

Copyright: Thunder Said Energy, 2019-2024.