Energy Weekly Review 2026-07-24
Week In Review
This week’s energy news continued a theme visible all summer: the technologies that were “coming soon” a decade ago are now generating measurable share of the grid and setting weekly records. On the solar side, LONGi Once Again Breaks World Record for Crystalline Silicon-Perovskite Tandem Solar Cell Efficiency at 35.5%, while a Berlin team’s Triple-junction all-perovskite solar cell based on graphene oxide bilayer achieves 27.3% efficiency — with a durability figure worth watching. On the deployment side, US trade press reported that Solar and storage account for 91% of new U.S. grid capacity in first half of 2026, and CAISO’s daily peaks documented in Solar and storage drive US grid growth in H1 with record Q1 capacity share showed batteries carrying a large share of California’s evening demand.
Advanced nuclear had one of its most concrete news weeks yet. Analysts revisited the Reactor Pilot Program in Developers, officials talk advanced reactor outlook with NRC, following through on the four microreactors that met the July 4 criticality target. Parallel to fission, the private fusion sector marked a specific hardware milestone with LLNL and Pacific Fusion achieve 3,000-shot milestone with Sirius pulsed-power prototype, and the broader industry context was laid out in In transition: Commercializing fusion power. Commonwealth Fusion Systems’ CFS named first international partner in UKAEA’s LIBRTI programme is a reminder that the fuel-cycle side of fusion — tritium breeding — has become a serious industrial workstream rather than a footnote.
Geothermal, storage, and materials rounded out the week. Fervo Energy reported that Fervo Energy Learning Curve Continues on 3rd Generation Well Design, Boosting Drilling Rates by 143% Since Its First Cape Station Well, continuing the learning-curve story that has geothermal on a cost trajectory nuclear developers have only started to match. On the storage frontier, The biggest problem with solid-state batteries may finally be solved — a mechanical-cycling problem that has kept the technology stuck in the lab. Taken together, the week’s news reads less like disparate breakthroughs and more like a portfolio maturing: solar cells getting more efficient and more stable, storage carrying real load, nuclear pilots running, and fusion systems logging shots at 95% efficiency.
Items
LONGi Sets 35.5% World Record for Silicon-Perovskite Tandem Solar Cell
At the 2026 Solar and Energy Storage Innovation Conference on July 14, LONGi announced a certified 35.5% conversion efficiency for its crystalline silicon-perovskite tandem cell — verified by the European Solar Test Installation (ESTI). Coverage of the announcement continued through the week and reached mainstream trade press by July 21.
The number matters because it comfortably exceeds the Shockley-Queisser limit of 33.7% for single-junction cells, which is the theoretical ceiling for the crystalline silicon panels dominating today’s rooftops and utility-scale plants. Tandem architectures stack a perovskite layer that harvests high-energy blue photons on top of a silicon layer optimized for the red and infrared portion of the spectrum. The theoretical ceiling for this configuration is around 43%, so 35.5% represents roughly 83% of the practical maximum.
LONGi’s trajectory is worth tracking: 33.9% in November 2023, 34.6% in June 2024, then 35.2%, and now 35.5%. Each increment reflects better perovskite composition, tighter interface engineering, and improved encapsulation — the same problems that have historically kept perovskites out of production. The cadence suggests the physics is not what’s holding back mass production so much as manufacturing yield and the notorious moisture sensitivity of perovskite films.
For the deployment story elsewhere in this review, cell efficiency compounds. Land, mounting, wiring, and permitting costs are roughly fixed per panel; higher efficiency spreads those costs over more watts. A jump from ~25% commercial single-junction to 30%+ commercial tandems is a 20% reduction in balance-of-system cost per delivered kilowatt-hour, before any manufacturing scale effects.
Source: LONGi
LLNL and Pacific Fusion Log 3,000 Shots on the Sirius Pulsed-Power Prototype
On July 16, Lawrence Livermore National Laboratory and Pacific Fusion announced that their Sirius prototype — a pulsed-power accelerator built on an “impedance-matched Marx generator” architecture — had completed 3,000 shots at 95% energy efficiency, with each shot delivering 60 gigawatts to a resistive load in a 100-nanosecond pulse. The milestone was covered by the American Nuclear Society on July 20.
Pulsed-power fusion is the less-famous cousin of the tokamak and laser approaches, but it has physical advantages: instead of confining plasma in a steady magnetic field or hitting a fuel pellet with lasers, it delivers an enormous, brief electrical pulse to compress a target. Sandia’s Z Machine is the best-known example. What made Sirius newsworthy is not just power output but repeatability — 3,000 shots is the kind of number a commercial power plant would need to hit routinely, not once.
The announcement moves the Sirius architecture from Technology Readiness Level 4 toward TRL 5, the level at which a system is demonstrated in a laboratory setting close enough to the target environment to project real-world performance. That is a specific, unglamorous checkpoint that fusion has often struggled to hit; a lot of prototypes have shown net gain in one shot and never been fired again.
Pacific Fusion’s CTO, Keith LeChien, co-invented the impedance-matched Marx generator architecture while at LLNL, and the current collaboration formalizes a technology transfer that has been informal for years. If the pulsed-power approach can scale to megajoule pulses at kilohertz repetition rates, it offers a fundamentally different economics from the giant-magnet or giant-laser paths.
Source: ANS Nuclear Newswire
Solar and Storage Account for 91% of New US Grid Capacity in H1 2026
A pv-magazine analysis published July 22 laid out an industry milestone: in the first quarter of 2026, solar and battery storage combined for 91% of all new US generating capacity added to the grid — the highest combined quarterly share ever recorded for the two technologies. May 2026 was the first month in American history when solar generation exceeded coal generation, and solar now supplies around 13% of total US electricity.
The 91% figure is not a projection or an interconnection queue; it is a report on capacity that was actually built and commissioned in a specific quarter. That number has been climbing for a decade, but crossing 90% establishes solar-plus-storage as the default technology for new US generation, not one of several competing options. Gas remains dominant in the existing fleet, but the delta each quarter is now essentially all solar and batteries.
The underlying economics are straightforward: unsubsidized solar plus four-hour batteries is now cheaper than a new gas peaker in most US grid regions, and much of what utilities and independent power producers were previously going to build as gas has been swapped out. Investment tax credit rules and interconnection queue reforms are the political inputs, but the numerical driver is that the levelized cost of storage has fallen faster than nearly anyone forecast in 2020.
The trajectory has implications elsewhere in the grid stack: as the marginal new plant becomes a solar-plus-storage hybrid rather than gas, transmission planning, ancillary service markets, and capacity mechanisms all need to be reworked for a fleet with different operating characteristics.
Source: pv magazine USA
California’s Grid Sets Twin Records: 12.99 GW Battery Discharge and 23 GW Solar
On the evening of July 9, the battery storage fleet in the California Independent System Operator (CAISO) footprint discharged a record 12.99 gigawatts, covering 36% of total regional demand at peak. The following afternoon, July 10, CAISO logged a solar generation record of 23 GW, meeting 72% of afternoon electricity demand — the third solar record for California since June 1. Coverage of the twin milestones appeared in ess-news on July 23.
These are the kind of numbers that reframe what a “renewable-heavy grid” looks like operationally. A decade ago, the concern was whether solar and wind could reliably supply even 20% of a grid’s energy without destabilizing frequency. California is now routinely operating at solar shares that would have been considered infeasible, and the batteries that used to be a footnote in dispatch reports are now a top-three source during evening ramps.
The evening ramp is where the batteries matter most: as the sun sets, solar drops off just as commercial and residential demand peaks. Historically, California relied on gas peakers to fill that gap. On July 9 the battery fleet handled more than a third of it. That is a specific, measurable displacement of natural gas by stored solar energy — and the fleet is still growing quickly.
Texas set its own solar record on July 9 at 35.4 GW, up from 34.4 GW in May. The combined California-Texas story is that solar-plus-storage is not a coastal or blue-state phenomenon; the two largest solar generation states are one Republican-run and one Democratic-run, and both are setting monthly records driven by the same underlying cost curves.
Source: ess-news
Advanced Reactor Pilot Program Delivers Four Criticalities by July 4 Deadline
Building on the Trump administration’s May 2025 executive order for a nuclear pilot program, four advanced reactors reached criticality — the point at which a fission chain reaction becomes self-sustaining — by the July 4, 2026 target date. The American Nuclear Society’s July 22 write-up covered how the program is being viewed by NRC officials and developers as it moves from criticality to power operations.
The four participants that met the deadline are Antares Nuclear, Valar Atomics, Aalo Atomics, and Deployable Energy (a participant in DOE’s Nuclear Energy Launch Pad rather than the main Reactor Pilot Program). Radiant, running the first tenant campaign at Idaho National Laboratory’s Demonstration of Microreactor Experiments (DOME) facility, is on a phased schedule that will progress through zero-power criticality, 1 MW thermal, full power, and 150 hours of unattended operation.
What is unusual about this program is that these test reactors were authorized under DOE authority rather than a full Nuclear Regulatory Commission license — a legal path that meaningfully compresses the timeline. Companies bear all costs for design, manufacturing, operating, and decommissioning. Whether the industry can carry a commercial-license path is now the pending question, and that is the transition the ANS piece describes.
The concurrent NRC dialogue matters because eventual commercial deployment will require the standard regulatory process. Developers are trying to use pilot-program data to shorten review times, and the NRC has to decide how much test-reactor experience translates to reduced uncertainty about power-reactor safety. That negotiation, over the next 12 to 24 months, will shape when and how the first commercial microreactors show up on the grid.
Source: ANS Nuclear Newswire
Fervo Energy’s Sawtooth 7: 21 Days, 19,448 Feet, 460°F
On July 8, Fervo Energy reported that its Sawtooth 7 well at Cape Station Phase II reached total depth in 21 days — a company drilling record — with a final depth of 19,448 feet, a 7,500-foot horizontal lateral, and a bottom-hole resource temperature of 460°F. That represents a 143% improvement in drilling rate since Fervo’s first Cape Station well, and a 70% reduction in drilling time compared with earlier projects.
Cape Station is being built out as a 400 MW enhanced geothermal system (EGS), scheduled to deliver power in 2028. EGS differs from conventional geothermal in that it does not require pre-existing hydrothermal reservoirs; instead, the developer drills into hot dry rock, engineers a fracture network, and circulates water through it as a working fluid. The technology has been demonstrated at small scale for two decades, and Fervo is the first company to industrialize it.
The learning-curve numbers are the important part of the story. Drilling costs dominate EGS economics, and every day of drilling saves roughly six figures. Fervo’s third-generation well design uses lessons from oilfield drilling — the same rotary steerable and downhole telemetry techniques that made shale drilling economic — applied to much hotter rock. The Cape Station target of around $5,500 per kilowatt installed is well below the $10,000–$15,000 per kilowatt of the recent Vogtle nuclear reactors.
If Fervo can maintain the drilling learning curve into Phase II, geothermal moves from a niche renewable to a baseload technology potentially competitive with new gas. Unlike solar and wind, geothermal is fully dispatchable, running 24 hours a day at capacity factors above 90%. That combination — dispatchable, carbon-free, and on a cost curve — is what has drawn investor and utility attention this year.
Source: Fervo Energy
Triple-Junction Perovskite Solar Cell Hits 27.3% with 770-Hour Stability
A team at Helmholtz-Zentrum Berlin (HZB) reported an all-perovskite triple-junction solar cell reaching 27.3% efficiency and — more importantly — retaining more than 90% of that efficiency after 770 hours of continuous operation. pv magazine covered the result on July 13.
Efficiency records for perovskites are announced regularly, but stability numbers are what determine whether the technology can ever leave the lab. Silicon solar panels routinely warrant 25-year lifetimes; perovskites, until recently, degraded visibly in days under illumination and humidity. The 770-hour result is not yet in the same league as silicon, but it is roughly two orders of magnitude better than early perovskite cells and finally credible as a starting point for commercial development.
The technical innovation is at the hole-transport layer. The team replaced PEDOT:PSS — a standard organic conductor known for degrading under UV — with a graphene oxide bilayer combined with a self-assembled monolayer (SAM). The bilayer reduces both optical losses (fewer photons wasted at interfaces) and interfacial recombination (fewer electrons lost between layers). The result is simultaneously higher efficiency and dramatically improved stability, which are usually competing objectives in perovskite research.
Read against the LONGi silicon-perovskite tandem record earlier in this review, the HZB result covers a different segment of the perovskite roadmap. Silicon-perovskite tandems can be dropped into today’s factories with modest process modifications; all-perovskite triple-junctions are further from commercial production but offer a longer efficiency runway and eventual freedom from silicon supply constraints. Both matter.
Source: pv magazine
In Transition: Commercializing Fusion Power
The American Nuclear Society’s July 23 industry survey took stock of where the private fusion sector actually stands — separating the plasma-physics milestones (Livermore’s ignition, various net-gain claims) from the engineering, regulatory, and grid-integration work that has to happen before fusion sells electrons to utilities.
The piece flags an inflection: several fusion companies have shifted from “prove the physics” to “design and build a plant.” Commonwealth Fusion Systems is constructing SPARC in Devens, Massachusetts, and its follow-on ARC plant is under design in Chesterfield, Virginia. Helion, TAE, and Realta Fusion are pursuing pulsed, beam-driven, and magnetic-mirror architectures respectively. The engineering diversity is not just a hedge; different approaches have different heat-recovery, materials, and fuel-cycle needs.
The commercialization discussion has three inflection points still unresolved. First is tritium supply — reactors that burn deuterium-tritium fuel need to breed their own tritium, and no facility has demonstrated closed-cycle breeding at plant scale (the LIBRTI collaboration described elsewhere in this review is the current attempt). Second is materials — first-wall components have to survive neutron fluxes for years, and testing that requires a fusion neutron source that does not yet exist. Third is regulation — the NRC is still developing a licensing framework for fusion, treating it as a distinct category from fission.
The tone across the fusion industry has changed over the past year. Companies that were once vague about timelines are now committing to specific first-plant dates in the early 2030s, and utilities are signing preliminary power purchase agreements. That does not mean fusion is around the corner, but the private fusion sector is now recognizably a capital-intensive industry building physical plants, not a research program hoping for a breakthrough.
Source: ANS Nuclear Newswire
Researchers Report a Fix for the Solid-State Battery Cracking Problem
ScienceDaily reported research on July 10 addressing what has been called the biggest single obstacle to commercial solid-state lithium batteries: mechanical failure of the solid electrolyte over repeated charge and discharge cycles.
Solid-state batteries replace the flammable liquid electrolyte of today’s lithium-ion cells with a solid ceramic or polymer conductor. That change unlocks higher energy density (400–500 Wh/kg targets versus 200–300 Wh/kg for conventional lithium-ion), faster charging, and better safety. The problem is that lithium metal deposits and dissolves at the solid interface with every cycle, and the resulting mechanical stress cracks the electrolyte or grows dendrites that short the cell. Cells that look perfect after 100 cycles fail catastrophically at 500.
The new work identifies a pressure-cycling protocol combined with an engineered interfacial layer that keeps mechanical stresses within an elastic regime instead of driving them into the plastic-deformation regime where cracks nucleate. In lab cells, the approach has demonstrated cycling behavior consistent with automotive lifetime targets, though commercial packs will require validation at scale and across temperature ranges.
Read in the context of the industry roadmap, the timing matters. China is expected to publish its first solid-state battery standard in July 2026, Toyota, BYD, and CATL have announced production timelines, and CES 2026 saw a claimed 400 Wh/kg cell with five-minute recharge and minimal fade over 100,000 cycles. Whether those production timelines hold depends on cracking behavior in the real world — which is the specific failure mode this line of research is targeting.
Source: ScienceDaily
Commonwealth Fusion Systems Joins UK’s LIBRTI Tritium Programme
The UK government announced on July 1 that Commonwealth Fusion Systems will be the first international partner in the UK Atomic Energy Authority’s Lithium Breeding Tritium Innovation (LIBRTI) programme — a £220 million initiative to demonstrate net tritium production for fusion power plants. Coverage continued into the following weeks as the fusion industry unpacked the implications.
Tritium is the fuel component that most fusion designs cannot buy at scale. Global tritium inventories, largely a byproduct of heavy-water fission reactors, are small and priced accordingly. A commercial fusion plant would consume roughly 100 kg of tritium per year at full output — more than the annual global supply. For fusion to work as a power source, plants have to breed their own tritium from lithium blankets surrounding the reactor.
LIBRTI is designed to prove that at engineering scale. The facility at UKAEA’s Culham campus will use a neutron source from Shine Technologies to bombard prototype breeder blankets and measure actual tritium production, extraction efficiency, and materials behavior over meaningful operating times. CFS will manufacture the first test articles and jointly design experimental campaigns with UKAEA. The results feed directly into CFS’s design of the ARC power plant.
The announcement matters as much for what it signals about industrialization as for the specific research plan. Tritium breeding has historically been treated as a research problem to be solved after net-gain plasmas were achieved. The current work is happening in parallel — companies are engineering the fuel cycle at the same time they are building the reactors that will burn the fuel. That is what commercialization looks like in practice: fewer sequential milestones, more parallel workstreams.
Source: GOV.UK