The clean machines are being built. That is where their carbon goes.

A factory week

LG Energy Solution opened one of America’s biggest battery cell plants in Lansing, Michigan on Tuesday [1]. The company put more than $2bn into the site. It employs 900 people and plans to hire about 800 more. At full capacity next year it should turn out more than 35 gigawatt-hours of battery cells a year [1].

Interior Secretary Doug Burgum spoke at the opening and called it “American energy dominance in action” [1]. That is notable. The administration has been openly hostile to electric cars and clean power. But LG has shifted much of the Lansing site away from car batteries towards grid storage, which has proved easier to sell across party lines [1].

The wider picture matches. Europe installed 36 gigawatt-hours of grid batteries in 2025, up 48% on the year before [3]. Its total working fleet passed 100 GWh for the first time. Germany led with 6.6 GWh of new capacity, Britain came second with 5.2 GWh, Italy third with 5 GWh [3]. Ukraine entered the top five with 2.9 GWh, roughly five times its previous year [3].

“Ukraine is proving that batteries are more than an energy technology; they are a resilience technology,” said Vladyslav Sokolovskyi of the Solar Energy Association of Ukraine [3]. Across most of Europe a home battery is bought to cut a bill. In Ukraine it is bought to keep the lights on.

Solar passed 3 terawatts of installed capacity worldwide this week [4]. Getting from 100 gigawatts in 2012 to the first terawatt took about a decade. The second terawatt took under three years, the third under two [4]. In 2025, 74 countries had at least a gigawatt of solar, up from 42 in 2020 [4]. In India, more than 150 GW of renewable projects were under construction as of 30 June [5].

Where a clean machine’s carbon actually sits

The same week produced an unusually direct statement about where these machines’ emissions come from. CATL, which supplied 39.2% of the world’s electric-car batteries in 2025, said all 20 of its operating plants are now certified carbon neutral [2].

Then it named the harder number. More than 80% of the emissions across a battery’s life come from its supply chain, not its factories [2]. CATL’s total value-chain emissions are more than five times those of its own operations. That is mining, refining, chemical processing, materials manufacturing and transport - the parts it does not control.

Carbon-neutral certification does not mean a factory emits nothing. The ISO standard CATL used requires a company to measure and cut its emissions first, then buy offsets to cover what is left [2]. The announcement did not disclose how much was left, or how many credits were used. It also did not split how much of its zero-carbon electricity was generated directly and how much was bought as certificates.

What CATL is doing next matters more than the certificate. It has baseline carbon data for over 100 of its main suppliers. From 2027, new suppliers must hand over product carbon-footprint data, and their emissions performance will count in supplier reviews [2]. Suppliers who do better could get priority on orders.

Cement is the same problem without the marketing. A study of 179 European cement plants, released early by Nature Communications this week, modelled how they respond to uncertainty about carbon prices, capture costs and pipelines [6]. It is peer-reviewed but still an accepted, unedited version. The finding was that confidence, not technology, decides the outcome. If plant owners believe the conditions will hold, carbon capture and biomass spread widely. The industry then becomes a net carbon sink, removing 0.6 gigatons of CO2 between 2025 and 2050. If they do not believe it, they delay, and the same plants emit 0.8 gigatons over the same period [6].

Researchers are also chipping at the material itself. A Scientific Reports paper this week tested hollow building blocks with 1.5% of the cement replaced by encapsulated beeswax [7]. In lab tests that cut the blocks’ heat conduction by 51.7% [7]. It is a single unreplicated study in an accepted, unedited manuscript, not a product.

The old machine question

A study published in Science this month put a number on the flip side. Swapping even a two-year-old petrol car for an electric one saves roughly 50% of lifetime emissions [8]. Its author, Elliot Campbell of UC Santa Cruz, set out to test the common belief that the greenest car is the one already in your driveway. The saving from driving electric caught up with the carbon of building the car faster than his own instinct suggested [8].

That is the arithmetic behind the whole build-out. It is also why supply chains are now the fight. Chinese manufacturers are wrapping factories around the world. Egypt’s Ain Sokhna alone now holds hundreds of plants across 12 industrial parks. Roughly half of the wider Suez zone’s recent investment has come from China [10].

And it is why materials carry a human cost long before a battery works. Over 40,000 tonnes of lithium are mined from African rock each year, a figure researchers expect to reach 500,000 tonnes by 2030 [9]. In the Sahel, armed and criminal groups are profiting from that trade across Mali, Niger, Burkina Faso, Chad and Nigeria, where governance is weak [9].

The accounting rules are shifting too. The Science Based Targets initiative updated its net-zero standard this month for the first time in five years [11]. Large firms must support carbon removals from 2035. But targets for emissions outside a company’s own walls, known as Scope 3, are no longer mandatory [11]. Several in the climate community read that as a loosening. Separately, a Nature Sustainability review of 183 studies looked at recycling and reuse in the Global South. It found the practice is mostly bottom-up and informal, driven by need and custom rather than environmental concern [12].

Who pays for the wires

Building the machines is one bill. Running a grid around them is another, and this week that argument got sharper.

Pennsylvania Governor Josh Shapiro signed an executive order setting requirements for data centres over 25 megawatts that want faster permitting [13]. It follows grid operator PJM’s proposal to make data centres bring their own power or face cutoffs. Utility Dive’s review of more than two dozen second-quarter earnings calls found the same theme: utilities are still selling data-centre pipelines, but facing equipment backlogs and public anger over bills [14]. “Growth remains intact, but affordability is emerging as key constraint,” analyst Shelby Tucker said [14].

In Britain, the opposition Conservatives published a report arguing that cheaper electricity - even if it means burning more gas - would speed up electrification and cut emissions. Carbon Brief’s factcheck listed ten flaws, among them that the plan would raise UK emissions and that it assumes gas plants are cheap to build [15]. Ofgem meanwhile named 16 long-duration storage projects it is minded to back under a new cap-and-floor scheme, aimed at the days-long lulls when wind and sun both fail [16].

Elsewhere in supply, Sage Geosystems began producing power from a 3 MW next-generation geothermal pilot near San Antonio. It is only the third such plant on the US grid, and has run connected for 120 days [17]. Scientific American surveyed four large bets now under way, including a Chinese compressed-air storage project in a salt cavern [18]. It also retraced solar’s fall in cost, starting from the 32 panels Jimmy Carter put on the White House in 1979 [19].

The bill already landing

While the build-out accelerates, the damage keeps accruing. Weather and climate extremes cost the European Union an estimated 822bn euros ($953bn) between 1980 and 2024, according to the European Environment Agency [20]. A quarter of that damage fell in just the last four years [20].

Only about a quarter of climate-linked catastrophe losses in the EU are insured, with cover below 5% in some countries [20]. That leaves the state as the insurer of last resort, at a time when euro-zone deficits already average around 3% of GDP. Fitch’s Federico Barriga-Salazar cited estimates that Spain’s 2024 floods imply reconstruction costs worth 0.7 percentage points of output across 2024 to 2026 [20].

One genuine piece of good news: emissions from land-use change, mostly deforestation, have fallen about 32% from their 2000s average, with a steep drop after 2015 [21]. Fossil-fuel emissions kept climbing over the same period.

Under-covered

Two studies this week point the same way: the transition works differently depending on who is doing it. A Nature Energy study across Mexico, South Africa, the USA and the UK tested a carbon tax and an information campaign on 1,589 people [23]. The two policies had divergent effects on electric-car take-up at the individual and system-wide scale [23]. And South Africa is retiring coal that still supplies 74% of its power. Its national energy plan has largely left out concentrating solar power [24]. The country has some of the best direct sunlight in the world, second only to Chile’s Atacama [24].

Meanwhile California is sitting on storage it has already paid for. A new report modelled a tenth of the state’s electric-car owners letting their cars feed power back to the grid by 2036. That alone would supply a third of California’s long-duration storage target [22].