Nicolas D Villarreal has sent in a letter challenging my energy essay's argument in multiple ways. Some of his challenges seem to rest on a misreading. Briefly, I did not argue that renewable energy "will only ever be additive within capitalism and never actually displace fossil fuels." Just that the current build-out is shaped by accumulation logic rather than energy-delivery logic, and that the real question is not whether but in what form, for whom, and at whose expense.

The fixed-capital distinction

Villarreal correctly points out that there is a meaningful difference between energy production that requires fuel to work and solar panels that only require maintenance and replacement.

The distinction is real but understates what reliable, dispatchable power requires beyond solar panels: storage, inverters, transformers, HVDC converter stations, etc., each with its own lifespan (10–40 years) and replacement schedule. I discuss these system costs in §2 of my essay; let me name two components here. First, the reactive-power support, updated grid codes, and grid-forming inverters needed to replace the frequency and voltage regulation that ‘analog’ spinning generators provide physically — the absence of which collapsed the Iberian grid in 90 seconds. Second, large power transformers, which are custom-built with 1–3 year lead times, mainly in China and South Korea, with rising demand from the global RE build-out that’s happening while the existing fleet needs replacement.^[1]

Even granting the distinction in principle, Villarreal's claim that "once the capital stock stabilizes" is doing enormous work. §2 of the essay argues that steady state is not reachable on advertised timelines, and §4 develops the Lauderdale mechanism that explains why capital structurally resists low-maintenance steady states — illustrated by Combined Heat and Power, which went from 30% of US electricity in 1930 to 4% by 1975.

The electricity-energy conflation

More fundamentally, the evidence base Villarreal cites — solar prices, clean-energy investment volumes, the panel-deployment curve — is electricity-specific, but the conclusion he draws is about energy systems broadly. Those are not the same scope. Electricity is ~20% of global final energy consumption.^[2] The remaining ~80% is from direct combustion of fuels.^[3] Electrifying the latter is not yet solved technologically in multiple industries, let alone mass produceable and 'economical'.

Transportation accounts for 26–28% of global final energy, 91% from oil, down 3.5 percentage points since the 1970s.^[4] Shipping is 99%+ oil-powered.^[5]

Electrification has a genuine efficiency argument for the sectors it can already reach, e.g., electric motors at 85–96% efficiency versus 25–40% for ICE, heat pumps delivering 3–5x more heat per unit of electricity than gas furnaces. But this advantage disappears at high temperatures, where electricity usage has no efficiency advantage over direct combustion. Roughly a quarter of all industrial energy goes to process heat above 400°C, importantly for bulk goods like steel (1,700°C), cement (1,450°C), and glass (up to 1,575°C).^[6]

Even with efficiency gains, the IEA's own Net Zero scenario (which has electricity reaching ~50% of final energy by 2050) means an increase from ~31,000 TWh to 60,000–75,000 TWh.^[7] That still leaves half of the final energy unsupplied by electricity, much of it in sectors with no viable, scaling substitute. Villarreal's investment charts track financial flows (clean-energy vs fossil-fuel investment volumes), not energy delivery, and his charts omit that system costs rise nonlinearly once the RE percentage becomes nonnegligible.

On the price-decline / S-curve argument: solar module costs really have fallen ~90% over fifteen years, and may keep falling. But the rest of what a solar installation requires (racking, inverters, transformers, labor, permitting, grid connection, financing, firming) has not. Per Lazard's annual reports, total installed system costs flattened around 2017 as modules shrank to a small fraction of total project cost while the other costs show no such trend.

On surplus absorption, price pressure, and investment

Villarreal's deindustrialization argument (that US fossil-fuel investment dominance reflects industrial weakness) misreads "deindustrialization." Since this process started in the core countries, we’ve seen multiple cycles of production moving around as it is automated further. That many of the replacement industries were built outside US borders does not show ‘neoliberal capitalism’ doesn’t need heavy industry, or that there is very little happening in the US, or that the shale boom was a side show. Global manufacturing output continued to grow.^[8] As such, it is simply wrong to think that “neoliberalism [having] the character of massive deindustrialization” in the core disproves the fact that when there is a lot of capital and highly developed industry floating around, capital loves high CapEx forms of development. This on top of the fact that capitalists also try to bar competitors from entry, try to destroy or corner existing markets, and try to generate new revenue streams, such as underlies much of the datacenter buildout. This is the pattern §4 develops, of which Lauderdale, CHP, the shale treadmill, and the RE derisking architecture are instances. Asserting shale and datacenter booms are "failed rent capture" and that’s that misses the forest for the trees twice. In the former case, because energy is a special resource, and in the latter, because the datacenter buildout was and is happening to allow large players to take other organizations’ data hostage, even as this allows them to crowd the field.

On the related claim that fossil-fuel investment has been falling since the 2010s shale peak: the aggregate did fall, but the picture is misleading. What fell was investment by listed Western majors; NOCs (Aramco, ADNOC, QatarEnergy, CNPC, Petrobras) now account for roughly half of upstream spending, up from ~40% pre-2014, and private equity has absorbed much of what the majors divested.^[9] The IEA series also excludes state-strategic and military fossil infrastructure: the US Department of Defense alone consumes 100 million barrels of oil annually.^[10] And total absolute fossil investment in 2024 ($1.1 trillion) remains roughly four times the IEA's own Net Zero pathway requirement, rising since 2020.

On price pressure: the essay's argument is not a static comparison of returns (RE 5–8% vs oil/gas 15%+). It is that RE is structurally unprofitable as a market-based investment, considering no major build-out anywhere has proceeded without state guarantees. Villarreal responds that RE depressing wholesale prices will also depress fossil returns. But cannibalization is asymmetric: even in the European grid solar peaks simultaneously, driving prices down or negative precisely because production is highest, while dispatchable generation captures the price spikes that the absence of solar creates in the other hours. The same effect that erodes RE returns creates the spikes fossil generation profits from, and the effect increases as more capacity is added.

The developing world and Chinese solar

Villarreal argues that cheap Chinese solar may free the developing world from dollar-denominated energy dependence. Solar panels must still be purchased alongside inverters, batteries, and charge controllers. China produces over 80% of the world's polysilicon, manufactures over 80% of solar cells and modules, and controls 60–70% of rare earth processing.^[11] This replaces one dependency with another and calls it freedom.

Whether this new dependency is more or less exploitative depends on the configuration of power in each case, as §4 of the essay illustrates drawing on Ching Kwan Lee's ethnographic work on Chinese state capital. Rooftop solar in Pakistan or Cuba is a genuine benefit, and China's deliveries to Cuba (~1,000 MW added by 2026, with 2,000 MW committed by 2028)^[12] meaningfully improve on a collapsing Soviet-era grid under US embargo. But by itself, this does not provide firm, dispatchable power for hospitals, industry, water treatment, or transport.

Cuba is also an easy case for the broader claim: 11 million people, minimal heavy industry, tropical latitude, and a political relationship where China has geopolitical reasons to be generous. For cheap solar to free "the majority of the developing world" from fossil fuel dependence, this strategy also needs to work for Nigeria (230 million people, growing industrial demand), India (1.4 billion, with steel, cement, and fertilizer sectors), or Indonesia (275 million, nickel processing). At those scales, the numbers are entirely different — and so is the political relationship China has with each of those countries.

A deeper question lurks here. If state-directed industrial policy at Chinese scale drove the cost decline Villarreal cites, what would a worker-aligned version look like? The essay's argument is not that no transition can serve working people, but that the one we are currently getting cannot. Engaging the question seriously means confronting both the asymmetry between countries that can mobilize this scale of industrial policy and those that cannot, and what Lee documents about Chinese state capital in recipient countries: configured by local power relations, neither uniformly liberatory nor uniformly extractive, while domestic worker power certainly cannot be claimed to be highly developed.

"How fast" is not the question

Villarreal himself doubts capitalism will manage the build-out in time. He nonetheless frames the remaining disagreement around speed: I supposedly dismissed "how fast" as bourgeois ideology, while he asserts "the writing is on the wall" without clarifying what he reads there. The essay's argument is that the build-out's institutional form determines whether what gets built serves working people or serves capital.

The argument that the 2022 energy crisis served as a transition accelerant cuts both ways. REPowerEU and the US Inflation Reduction Act did boost RE buildout. But the same crisis led to a worldwide increase in coal use and replaced Russian piped gas with LNG imports, which is a ~30% less efficient delivery mode that requires liquefaction, shipping, and regasification.

The horse analogy and the direction of transitions

Lastly, on the decline of the US domesticated equine population as proof that energy transitions happen. It is a counterexample to the strongest version of Fressoz's claim, but it cuts the wrong way. Work horses (sources of low-density living exosomatic energy with minds of their own, needing rest and training) were replaced by engines powered by fossil fuels (fully dispatchable out of the factory gate). The renewable build-out runs the reverse.

What would change my mind

Investment flows ≠ energy delivery ≠ energy substitution. What would challenge the essay's argument is evidence that: (a) the system costs it identifies are being paid rather than deferred; (b) fossil fuel consumption is declining in absolute terms, not just in share or because a war ends; (c) the build-out is delivering reliable, affordable energy to working people rather than constructing financial instruments for institutional investors; (d) the materials supply chain is feasible at scale, honestly accounting for ore-grade decline, processing energy, and the qualitative difference between mining coal (which is the product) and mining low-grade ores like copper.^[13]

On criterion (a), the evidence is not encouraging. India — one of the countries where Villarreal's argument most needs to work — curtailed 300 GWh of renewable energy in Q1 2026 because the transmission grid could not carry it. The grid has delivered only 80% of its planned buildout for five years running; 20 GW of RE capacity now faces connectivity delays over four months. The causes are familiar: right-of-way disputes, environmental clearances, HVDC shortages, and an RE-vs-transmission timing mismatch (12–18 months vs 36–60).^[14] Ember calls it "a structural infrastructure challenge rather than a one-off delay."

I welcome the engagement, but my main concerns remain in what form, for whom, and at whose expense, and what this implies for building worker power.

1. US DOE (2014): the US has essentially no domestic manufacturing capacity for transformers rated above 345 kV; imports come primarily from South Korea, China, Germany, and Japan. Lead times have extended to 2–3 years for some large units by 2023–2025, driven by simultaneous demand from grid expansion, RE integration, and fleet replacement. Large power transformers require grain-oriented electrical steel (GOES), a specialty product with limited global production capacity.

   ↩

2. IEA, Electricity 2024 (Paris: IEA, 2024). Electricity's share of global final energy consumption was 20% in 2023, up from 18% in 2015.

   ↩

3. IEA, World Energy Balances (Paris: IEA, 2024). Global total final consumption (2019): 418 EJ, of which fuel 78%, electricity 22%.

   ↩

4. IEA, "Transport" sector tracking page. Oil products supplied 91% of transport final energy as of 2022.

   ↩

5. IEA, "International Shipping" sector tracking page (2022). Vessel lifetimes: 20–35 years.

   ↩

6. Steel: ~11% of global CO2; cement: ~8%. US DOE estimates ~50% of manufacturing energy goes to process heat, roughly half above 400°C.

   ↩

7. IEA, Net Zero by 2050 (Paris: IEA, 2021): electricity must reach ~50% of final energy by 2050, with ~90% from renewables. IEA, Net Zero Roadmap (Paris: IEA, 2023): renewable capacity must triple to 11,000 GW by 2030. Global electricity generation in 2024: ~30,850 TWh (Ember, Global Electricity Review 2025).

   ↩

8. UNIDO, International Yearbook of Industrial Statistics (annual). Global manufacturing value added grew from ~$5 trillion (1995) to ~$15 trillion (2022) in current USD, with substantial real growth even after inflation adjustment. China's share rose from ~5% to ~30% over the same period; the US share fell, but US manufacturing value added grew in absolute terms (~$1.5T to ~$2.5T) even as manufacturing employment fell sharply. World Bank "Manufacturing, value added (current US$)" series provides equivalent data.

   ↩

9. IEA, World Energy Investment 2024 (Paris: IEA, 2024). National oil companies' share of upstream oil and gas investment reached roughly 50% in 2023, up from ~40% pre-2014. Total fossil fuel investment in 2024: ~$1.1 trillion. Private equity expansion into upstream assets divested by listed majors is tracked by IEEFA (Institute for Energy Economics and Financial Analysis) and the Private Equity Stakeholder Project.

   ↩

10. Neta C. Crawford, Pentagon Fuel Use, Climate Change, and the Costs of War (Watson Institute, Brown University, 2019; updated 2022). The US Department of Defense consumes approximately 85–100 million barrels of fuel annually, making it the world's largest single institutional consumer of petroleum.

    ↩

11. IEA, Solar PV Global Supply Chains (Paris: IEA, 2022): China's share of manufacturing capacity: polysilicon ~80%, ingots/wafers ~97%, cells ~83%, modules ~75%. IEA, Critical Minerals in Clean Energy Transitions (Paris: IEA, 2021/2023): China processes ~58% of lithium, ~70% of cobalt, ~60–70% of rare earths.

    ↩

12. Cuba received ~100,000 bpd of Venezuelan oil under Petrocaribe, declining sharply after 2016. Domestic generation predominantly Soviet-era thermoelectric plants >40 years of service. US embargo restricts equipment, parts, and financing. China shipped $117 million in solar panels to Cuba in 2025 (up from $3 million in 2023), connected 49 new solar parks adding ~1GW by early 2026, and committed to doubling that by 2028 — nearly matching Cuba's entire current nameplate capacity. China also donated 10,000 household solar kits and 5,000 systems for critical facilities (maternity homes, nursing homes, emergency rooms). Battery imports from China rose from $7.3 million (2024) to $56 million (2025). Sources: CNN, 13 May 2026; Power Magazine; CiberCuba, 7 April 2026.

    ↩

13. Global average copper ore grade: ~1.5% (1990s) to ~0.5% (2022). At 0.5%: ~200 tonnes of ore per tonne of refined copper; 200–500 tonnes total waste per tonne of product. Energy intensity: 30–50 GJ/t and on the rise (Cochilco, 2024).

    ↩

14. Duttatreya Das, "Transmission gaps are beginning to constrain India's rapid renewables integration" (Ember, 19 May 2026). India curtailed 470 GWh of renewable energy in Q1 2026, ~300 GWh because the grid was at capacity and 170 GWh because coal plants were unable to ramp down fast enough. India has met only ~80% of its annual transmission buildout targets over the past five years; 25% of transmission projects planned for FY2026-27 are delayed by more than a year. Of 45 GW of RE projects in the pipeline, ~20 GW face connectivity delays exceeding four months. Causes: right-of-way disputes, forest/environmental clearances (including Gram Sabha consent under the Forest Rights Act), biodiversity restrictions (Great Indian Bustard undergrounding requirements), and global shortages of HVDC converters.

    ↩