Energy Weekly Review 2026-09-11
Week In Review
This was a week about who pays for firm power and how fast it can be made to appear. The US Department of Energy’s Office of Energy Dominance Financing closed a loan of up to $1.9 billion to restart the Duane Arnold Energy Center in Iowa, a 615 MW reactor shut down in 2020 that now has a named anchor customer in Google. Two days later, on the other side of the Atlantic, Google signed a 22-year power purchase agreement with Fortum covering up to half the output of Finland’s Loviisa plant. The pattern is the same in both cases and it is new: a hyperscaler’s balance sheet is being used to underwrite the economics of existing nuclear capacity, either by resurrecting a closed unit or by making a life-extension decision financeable decades in advance. Nuclear plants have always been capital problems more than physics problems; the AI buildout has supplied a counterparty willing to sign for twenty years.
Fusion, meanwhile, spent the week on hardware and paperwork rather than headline plasma shots. ITER reported a summer of assembly milestones, including preparations to push a 330-tonne toroidal field coil to 34,000 amperes — more than triple the current its cold test facility has handled so far — and the start of installation on the bellows system that will stitch the vacuum vessel to the cryostat. On the same day the Fusion Industry Association published a white paper telling US Agreement States how to prepare for fusion companies arriving in their jurisdictions. Neither item is a breakthrough, and that is rather the point: a field graduates when its news becomes logistics and licensing.
On the laboratory side, two results pushed at ceilings that have stood for a while. LONGi and Soochow University reported a 34.0%-efficient perovskite–silicon tandem cell built on a nanoscale zirconia scaffold, with an independently certified open-circuit voltage above two volts and markedly better degradation behaviour than its control. In Nature Energy, a team demonstrated a lithium–disulfur dichloride cell that coaxes three electrons out of each sulfur atom instead of two, lifting average operating voltage from 2.05 V to 2.54 V. Sulfur and silicon are the two cheapest paths to abundant energy hardware, and both got meaningfully better this week.
The deployment stories tie the rest together. In California, Google, Tesla, Sunrun and PG&E launched a 12 MW virtual power plant called SHARE out of roughly 21,000 devices already sitting in people’s homes. In Texas, FlexGen and SMT Energy commissioned a 160 MW/320 MWh battery in six weeks, three times faster than planned, into a grid setting consecutive load records. Dutch startup Ore Energy raised $43 million to manufacture iron-air cells made of nothing but iron, water and air. And Japan’s Idemitsu opened a US laboratory for space-grade CIGS cells, a reminder that the photovoltaic learning curve now extends past the atmosphere. Financing, chemistry and installation speed are all improving at once, and the binding constraint keeps moving.
Items
DOE Closes a $1.9 Billion Loan to Restart Duane Arnold
The Department of Energy’s Office of Energy Dominance Financing reached financial close on 8 September on a loan of up to $1.9 billion to NextEra Energy, financing the restart of the Duane Arnold Energy Center in Linn County, Iowa. The 615 MW boiling water reactor stopped generating in 2020, a casualty of cheap gas, flat demand and derecho damage to its cooling towers. NextEra now plans to have it back on the grid no later than the first quarter of 2029, pending regulatory approvals.
Restarting a shut reactor was, until very recently, something the US nuclear industry had never done. Duane Arnold is now the third such project to receive federal financing under this administration, following similar transactions elsewhere. The appeal is straightforward arithmetic: the reactor, its site, its licence history and much of its trained workforce already exist, so the capital required is a fraction of new construction and the schedule is measured in years rather than a decade. The unresolved questions are mechanical — the condition of long-idled components, the cooling infrastructure, and the regulatory path back to an operating licence.
What makes the deal legible is the offtake. Google will be the primary customer for the revived plant, using its output to power growing cloud and AI infrastructure in Iowa. That is the arrangement that turns a speculative restart into a financeable asset: a creditworthy buyer committing to purchase the electricity for long enough to repay the loan. The Department estimates the restart could generate more than $9 billion in economic benefit to Iowa over 25 years.
The broader significance is that the United States has quietly developed a repeatable template for recovering shut nuclear capacity — federal debt at the front, a hyperscaler offtake at the back, and an existing site in the middle. The supply of candidate reactors is finite and small, but for the handful that remain recoverable, this week established that the financing question has an answer.
Source: U.S. Department of Energy
Google Signs a 22-Year Nuclear Deal in Finland
Google has agreed a long-term power purchase agreement with Finnish utility Fortum for output from the Loviisa nuclear power plant, its first nuclear PPA outside the United States. The 22-year arrangement begins in 2028 at reduced volume and covers roughly half of Loviisa’s capacity from 2030 through 2049, running alongside Google’s expansion of data centre capacity in Finland.
Loviisa’s two VVER units have operated since the late 1970s and early 1980s, and Fortum secured licence extensions carrying them toward 2050. Those extensions are not free: continued operation of a half-century-old plant means sustained investment in components, instrumentation and possible power uprates. A committed buyer for half the output across two decades is precisely the financial certainty that makes such reinvestment rational rather than risky, which is why the agreement matters more than its megawatt figure suggests.
The structure inverts the usual complaint about data centres and the grid. Rather than simply adding load and letting the system absorb it, the buyer is funding the continued availability of existing zero-carbon generation. Finland is an unusually good setting for the experiment: a cold climate that suits data centre cooling, a grid already heavily nuclear and hydro, and a regulatory culture that recently delivered Europe’s first operating EPR and its first deep geological waste repository.
For the nuclear sector, the notable detail is geographic spread. Corporate nuclear offtake has been an American phenomenon for two years, concentrated in PJM and the Southeast. Its arrival in the Nordics suggests the model travels, and that ageing European reactors with decades of remaining licensed life may find their economics rewritten by the same demand that is reshaping the American grid.
Source: World Nuclear News
ITER Prepares to Push a 330-Tonne Magnet to 34,000 Amperes
The ITER Organization reported what it called a summer of substantial progress across assembly and installation, with the most consequential work happening at its magnet cold test facility. Engineers are preparing to test a 330-metric-tonne toroidal field coil at 34 kA — more than three times the current achieved when the facility began operations earlier this year. Further structural reinforcement is planned before testing escalates to 68 kA, which will begin with toroidal field coil #19.
The toroidal field coils are the machine’s defining components: eighteen D-shaped niobium-tin superconducting magnets, each the size of a small building, that will generate the field confining a plasma hotter than the sun’s core. Testing them at full current while cooled to a few degrees above absolute zero is not a formality. A coil that fails at 68 kA after installation inside the tokamak would be a catastrophe for the schedule, so each one is exercised at operating conditions in a dedicated facility first — an expensive, slow discipline that reflects hard lessons from earlier component problems.
Separately, installation has begun on ITER’s system of 93 bellows, the flexible metal connectors that link the vacuum vessel to the surrounding cryostat and the cryostat to the seven-storey tokamak building. Five sets of rectangular port cell bellows are now in place. Bellows are unglamorous but essential: the vessel will move as it heats and cools by hundreds of degrees, and every port, pipe and diagnostic penetration needs a joint that accommodates that motion while holding vacuum.
ITER’s public reporting has shifted noticeably in character over the past two years, from schedule revisions to hardware milestones. Magnets being tested, bellows being bolted in, and port cells being fitted out are the signature of a project in assembly rather than redesign. Whether the revised baseline holds remains to be seen, but the machine is now being built rather than argued about.
Source: ANS Nuclear Newswire
A Regulatory Roadmap for the States Where Fusion Will Actually Be Built
The Fusion Industry Association released a white paper advising current and prospective Agreement States on how to prepare for commercial fusion companies seeking to operate within their borders. It is a document about jurisdiction, and jurisdiction is about to become one of fusion’s practical bottlenecks.
The US regulatory picture for fusion has been settling along a particular line. Rather than treating fusion devices under the framework built for fission reactors, the Nuclear Regulatory Commission has proposed handling them primarily under 10 CFR Part 30 — the byproduct materials rules used for particle accelerators and similar radiation-producing equipment. The Commission closed its comment period on those proposed revisions in May. The logic is that a fusion machine cannot run away, holds a negligible inventory of fissile material, and presents hazards closer to a large accelerator than to a power reactor.
That choice has an important consequence: byproduct material regulation is largely delegated to Agreement States, the thirty-nine states that have signed agreements with the NRC to license and inspect such materials themselves. In practice, a fusion company choosing a site will often deal with a state radiation control programme, not federal regulators. Those programmes are staffed by people with deep experience in medical isotopes and industrial radiography, and considerably less in tritium handling, neutron activation of structural materials, or the specific engineering of a stellarator.
The white paper exists to close that gap before it becomes a delay. Its value is essentially anticipatory — giving state programmes a common vocabulary, a sense of what fusion licensees will ask for, and a picture of the hazards that actually matter. For a field whose commercial timelines are measured in single-digit years, the difference between a state programme that has thought about fusion and one encountering it cold could be a year of schedule.
Source: ANS Nuclear Newswire
A Nanoscale Scaffold Takes Perovskite–Silicon Tandems to 34%
Researchers at LONGi and Soochow University reported a perovskite–silicon tandem solar cell with 34.0% power conversion efficiency, achieved through what they call a dual-anchored interfacial design. The work was published in Science Bulletin.
The problem they attacked sits at one of the least glamorous places in the device: the boundary between the transparent conducting oxide and the self-assembled monolayer that the perovskite grows on top of. Self-assembled monolayers are a single molecule thick and attach to the oxide surface somewhat imperfectly, leaving gaps where charge carriers recombine and current leaks. The team inserted discrete nanoparticles of monoclinic zirconia between the two layers. The particles tune the local surface energy so the perovskite crystallises into void-free films with large grains, and they form covalent zirconium–oxygen–phosphorus bonds that anchor the monolayer far more firmly than before.
The resulting cell reported an open-circuit voltage of 1.997 V with a fill factor of 83.62% and a short-circuit current density of 20.36 mA/cm², and an independently certified open-circuit voltage of 2.014 V. Crossing two volts in a two-terminal tandem is a meaningful marker, because voltage is where tandems have historically lost the most to interfacial defects. The stability result is arguably more important: the device retained 84% of its initial efficiency after 2,000 hours of operation, while the control cell had fallen to 70% after only 1,000 hours.
Perovskite tandems have spent several years posting impressive efficiencies that faded under sustained operation, which is why commercial modules remain in the mid-to-high twenties while laboratory cells sit above 34%. An interfacial fix that improves efficiency and durability through the same mechanism — better bonding, fewer defects — is the kind of result that shortens the distance between those two numbers.
Source: pv magazine
Getting a Third Electron Out of Sulfur
A study in Nature Energy describes a lithium–sulfur battery chemistry that fundamentally changes what sulfur does during discharge. Conventional lithium–sulfur cells shuttle two electrons per sulfur atom, cycling between elemental sulfur and lithium sulfide. The new work reversibly converts lithium sulfide all the way to disulfur dichloride, accessing the S⁺¹ oxidation state and extracting three electrons per sulfur atom.
The enabling ingredient is an ionic liquid electrolyte rich in free chloride, which acts as a mediator that ferries chlorine into and out of the sulfur redox reaction while contributing little capacity itself. That chlorine-mediated pathway raises the average operating voltage from 2.05 V to 2.54 V at 25 °C and a 0.2C rate, increases sulfur-specific capacity by 58%, and delivers electrode-level specific energy above 1,700 Wh/kg, with cycling demonstrated over 100 cycles.
Sulfur has long been the most tantalising battery cathode material available: it is a byproduct of petroleum refining, effectively unlimited, and theoretically capable of energy densities several times those of lithium iron phosphate. Three problems have kept it in the laboratory — low voltage, sluggish reaction kinetics, and the polysulfide shuttle, in which intermediate species dissolve into the electrolyte and ferry charge uselessly between electrodes, consuming the cell from within. Raising the voltage attacks the first problem directly, since energy is voltage times capacity and lithium–sulfur has always been long on capacity and short on volts.
The honest caveat is the cycle count. A hundred cycles establishes that the chemistry is reversible, not that it is durable; grid and vehicle applications want thousands. What the result does establish is that sulfur’s electron budget is not fixed at two, and that the chemistry of the electrolyte can be used to reach oxidation states that plain lithium–sulfur cells never touch. That is a new design axis in a field that has been circling the same three problems for two decades.
Source: Nature Energy
Iron, Water and Air: Ore Energy Raises $43 Million
Netherlands-based Ore Energy closed a $43 million Series A to commercialise its iron-air long-duration storage system, bringing its total raised to $61 million. The funding will pay for a first manufacturing facility — sites in Germany and the Netherlands are under evaluation — with operations targeted for 2027 and gigawatt-hour-scale production by 2028.
Iron-air batteries work by rusting on purpose. Discharging oxidises iron in the presence of oxygen from ambient air; charging reduces the rust back to metallic iron. The reaction is slow, which makes the technology useless for the fast frequency response that lithium batteries provide, and wonderfully cheap, because the materials are iron, water and air rather than lithium, nickel and cobalt. Ore’s system is designed for 100-hour discharge — multi-day storage that covers the long, still, overcast stretches that defeat four-hour lithium installations.
The company has been building operational evidence rather than relying on laboratory claims. It completed a grid-connected pilot of its 100-hour system at EDF’s Lab les Renardières research site in France, following an earlier grid-connected installation in Delft, and has signed an agreement with Dutch utility Budget Thuis for large-scale commercial deployment. Two grid connections and a utility customer is a stronger position than most long-duration startups occupy at Series A.
The strategic argument is European. Because the cells require no critical minerals, the entire supply chain can sit inside Europe — a point of some weight for a continent that has watched lithium cell manufacturing consolidate in Asia and is now attempting to rebuild industrial capacity in batteries. Ore is one of several companies pursuing iron-air chemistry, with the American firm Form Energy the best-capitalised; competition on a chemistry this cheap is a good sign for the technology’s prospects.
Source: Energy-Storage.news
A 12 MW Power Plant Assembled From Equipment Already in People’s Houses
Google, Tesla, Sunrun and PG&E launched SHARE — Smart Home Assets for Reliability and Efficiency — a virtual power plant in California’s Santa Clara and Alameda counties that will supply 12 MW of capacity during peak demand periods by coordinating roughly 21,000 flexible energy devices already installed in customers’ homes. The programme runs from autumn 2026 through 2027 and is fully funded by Google.
The division of labour is instructive. PG&E operates the programme as the utility partner. Tesla, Sunrun and Renew Home enrol the existing fleet of home batteries, smart thermostats and solar-plus-storage systems that their customers already own. Carrier Global supplies battery-enabled heat pump systems as a launch partner, and Rewiring America handles deployment and coordination with local installers. Google pays for it. No one is building a power plant; they are building the software and contractual layer that turns dispersed hardware into a dispatchable resource.
Compensation is deliberately concrete. Sunrun customers receive $200 per battery enrolled. Households can take $5,000 off a Carrier heat pump system, with twenty-five receiving $10,000 off. These are the terms that determine whether a virtual power plant is real, because the resource only exists if enough people find the trade worthwhile — surrendering some control over a thermostat or a battery in exchange for money.
The 12 MW figure is modest against California’s tens of gigawatts of peak demand, but the economics are what matter. Twelve megawatts of peaking capacity built from equipment already purchased and installed for other reasons costs a fraction of the peaker plant or transmission upgrade it displaces, and it arrives in months rather than years. The inclusion of battery-enabled heat pumps is a notable extension: space conditioning is the largest single driver of residential peaks, and a heat pump that carries its own storage can shift that load without the occupants noticing.
Source: Energy-Storage.news
A 320 MWh Texas Battery Built in Six Weeks
FlexGen and SMT Energy began operations at Houston IV, a 160 MW/320 MWh battery energy storage facility in Houston, Texas. The notable figure is not the capacity but the calendar: the project was commissioned in six weeks, cutting the original deployment timeline by a factor of three.
Six weeks from construction to commercial operation for a utility-scale battery is close to the practical floor for this technology. Grid-scale storage is unusually well-suited to compressed schedules — the equipment arrives as factory-built containers, the site work is a concrete pad and a substation interconnection, and there is no boiler, turbine or fuel handling to commission. Getting to six weeks nonetheless requires the interconnection agreement, the equipment supply and the software integration to be sequenced almost perfectly, and it reflects a supply chain and an engineering practice that have both matured considerably.
The timing suits the grid it joined. According to the Energy Information Administration, ERCOT set an hourly peak load record of 91.1 GW on 22 July, 6% above the previous record of 85.5 GW set in August 2023, and weekly average peak loads over the six weeks from late July all exceeded the 2023 record. Texas load is growing through a combination of population, industrial electrification, cryptocurrency mining and data centres, and the state’s enormous solar fleet produces a steep evening ramp that batteries are specifically good at covering.
Houston IV joins more than a dozen projects that SMT and FlexGen have built for the ERCOT market. Texas has become the world’s most instructive demonstration that storage can be deployed at the speed demand grows — a contrast worth noting against generation assets that take five to fifteen years to bring online, and one reason ERCOT has absorbed record load without the reliability crises many predicted.
Source: Energy-Storage.news
Idemitsu Opens a US Lab for Space-Grade Thin-Film Solar
Japan’s Idemitsu Kosan has established a development laboratory at UC Santa Barbara’s Oasis facility to advance space-grade copper indium gallium selenide solar cells for satellites and spacecraft. The US lab complements a bench manufacturing plant in Japan aimed at establishing mass production technologies by 2027, with full-scale production to be evaluated according to market conditions.
Space photovoltaics is a market with inverted priorities. On Earth, cost per watt dominates and efficiency matters mainly because it reduces the area and mounting hardware. In orbit, mass is the binding constraint, and radiation tolerance determines how much power an array still produces after years in a high-energy environment. CIGS thin films suit those conditions well: they can be deposited on flexible, lightweight substrates and they are notably resistant to radiation damage, which means an array can be sized for the power actually needed rather than oversized to compensate for predictable degradation.
Idemitsu’s position in this chemistry is longstanding. The company took a controlling interest in Solar Frontier in 2019; Solar Frontier still holds the efficiency record for CIS thin-film cells at 23.35%, set that same year. The firm is also working with US-based Source Energy Company on CIGS modules for satellites, an arrangement announced in late 2025. Locating development work in Santa Barbara places it near a substantial share of the American space industry and its academic materials research.
The commercial logic is a straightforward consequence of launch economics. Satellite constellations now number in the thousands, and each spacecraft needs solar arrays; cheaper launch has made mass slightly less precious but constellation scale has made per-unit cost far more important. That combination favours thin films that can be manufactured in volume over the ultra-high-efficiency triple-junction cells that have dominated space power for decades — a market that is growing large enough to attract a terrestrial PV manufacturer’s attention.
Source: pv magazine