Opening 0:00
Welcome to today's China report. Today it's energy transition and critical minerals: four reports on how China came to own so much of the clean-tech stack, and whether the West can do anything about it. Two of them aren't squarely on our topic, the International Energy Agency's minerals outlook isn't from a think tank at all, and the CSIS paper is about trade law rather than energy, but the first quantifies the chokepoints the others assume, and the second examines the West's main counter-move, so both earn their place. Just don't let me stretch them past what they actually claim.
Carnegie on battery geopolitics 0:37
Start with the Carnegie Endowment for International Peace, a long paper by Milo McBride published at the end of June. The claim is that China's battery lead is structural, not a fluke. By twenty thirty Chinese cell capacity could be roughly triple that of all the OECD countries combined. And for lithium iron phosphate cells, the cheap chemistry that's taking over, China holds about ninety-eight percent of global production capacity.
But McBride's conclusion isn't autarky. It's the opposite. Here's the logic. The mechanism is cost. Chinese lithium iron phosphate cells run somewhere between twenty-four and fifty percent cheaper than European ones. Closing that gap, McBride reckons, would take at least a thirty percent tariff, or an American-style thirty-dollar-per-kilowatt-hour production tax credit. So fighting head-to-head on the cheap chemistry is a losing battle. Instead, the argument goes, the US, Europe, Korea and Japan should trade market access for technology, and concentrate their own factories on the chemistries that aren't locked up yet, sodium-ion, silicon anodes, lithium metal.
What's it resting on? Capacity projections from the Global Clean Investment Monitor, European single-market prices as a tariff-free proxy for the cost gap, and a patent database analysis, China at about forty-three percent of battery patents in twenty twenty-four, Europe twenty-one, the US eighteen. That's announced capacity, not built capacity, and to his credit McBride says so. He also flags that new battery lines run scrap rates of forty to ninety percent, which quietly undercuts everybody's headline numbers, China's included.
It's a genuinely careful paper. But the whole recommendation assumes the partners want to cooperate, and it never really tests that, Korea and Japan hold the technology, and they're being asked to share it. And the niche strategy has a timing problem the report's own data exposes: China could have over five hundred gigawatt-hours of sodium-ion capacity by twenty thirty. The empty niche may already be occupied. What would settle it is simple, whether a single Western sodium-ion or silicon-anode firm ships at gigawatt scale.
IEA on critical minerals concentration 3:12
Now the International Energy Agency's Global Critical Minerals Outlook, published mid-July. Its message is that concentration is getting worse, not better. The average share held by the top refining country rose to seventy-two percent in twenty twenty-five. For gallium, graphite, manganese and rare earths, China is over ninety percent. And the headline: full implementation of China's October rare earth export controls could put an estimated six and a half trillion dollars a year of downstream production outside China at risk.
The argument underneath is about where the bottleneck sits. Mining is diversifying. Refining isn't. By twenty thirty-five, rare earth mining capacity outside the top producer reaches nearly fifty thousand tonnes, but refining is under forty thousand, and actual magnets only about eighteen thousand. Dig all you like; the choke moves downstream. And the fix is cheap relative to the exposure: about sixty billion dollars over a decade for magnet supply chains. The IEA calls this a mineral security premium, insurance, essentially.
The strongest evidence here isn't the modelling, it's the arithmetic. Rare earths are around forty percent of a magnet's cost but under one percent of a vehicle's value, so tripling the rare earth price adds about one tenth of one percent to a car. That's checkable and it doesn't depend on any scenario.
The six and a half trillion, though, does a lot of work it can't quite carry. It's gross exposure, not expected loss, the annual value of everything downstream that touches Chinese rare earths, across autos, high-tech, defence and energy, if controls were fully enforced. It measures the size of the room, not the size of the fire. The report doesn't say what share would actually be lost, over what period, or after substitution. And here's the awkward part: the same document argues cost pass-through is trivial and substitution is affordable. Those two claims pull against each other. One more tell, the executive summary puts the average refining share at seventy-two percent, up from seventy in twenty twenty-three; the market chapter says seventy percent, up from sixty-eight in twenty twenty. Same concept, different numbers. Small thing, but it tells you how soft the aggregate is.
Rhodium on Chinese cars in Thailand 5:45
The Rhodium Group, late July, on how Chinese cars took over Thailand. Chinese brands went from near zero to close to thirty percent of the vehicle market, and about ninety percent of electric vehicle sales, in five years.
The mechanism is three things stacked. Zero tariffs on Chinese cars under the twenty-oh-three China-ASEAN free trade agreement, against twenty percent on Japanese ones. Thai subsidies worth seven and a half to ten percent of an EV's price. And Chinese overcapacity at home pushing prices down ten percent in twenty twenty-four and another thirteen six months later.
But sales ran ahead of factories, and that's the crack. Chinese brands are less than ten percent of Thai production against almost thirty percent of sales. Local content sits between forty and sixty percent, where Japanese makers are at seventy to eighty. From there the authors name three tests: fiscal sustainability, real localisation, and politics.
The evidence is the best-sourced of the four. Thai Board of Investment disbursement records, about twelve billion baht, three hundred and sixty million dollars, over three years. Registration data. Rhodium's own tracking of Chinese investment announcements. And the Thai planning agency's estimate that around a hundred and ten thousand workers are at risk, sixteen point three percent of the auto workforce. They also cite a reported fifty percent price gap, but they flag that as stakeholder talk, not data. That's the right way to do it.
My problem is with the fiscal test, and it's the report's own number that does the damage: three hundred and sixty million dollars over three years is less than half a percentage point of the Thai annual budget. That's not a scheme that collapses under its own cost. And "tests it may not pass" is a forecast, what the evidence actually shows is one lawsuit against Neta and one letter from ten industry associations. Pressure, not failure.
CSIS on Section 301 tariffs 7:52
Finally CSIS, Scott Kennedy and Claire Reade, published the tenth of August, on the July Section 301 tariffs covering sixty trading partners. Two claims: shaky legally, and not working.
On law, they argue the forced-labour determination is thin, uniform tariffs across ninety-seven to ninety-nine percent of US imports, on a record they call cursory at best. On effectiveness, the bilateral deficit with China fell from two hundred and ninety-seven billion dollars in twenty twenty-four to two hundred and three billion in twenty twenty-five, and the overall US deficit didn't move, because deficits with everyone else grew. That's diversion, not reduction. US exports to China fell thirty-five percent. The Yale Budget Lab puts the household cost around eleven hundred dollars.
The number that matters for us is buried near the end. There's a twenty percent tariff ceiling on China under the current arrangement, twelve and a half is already used, so only seven and a half points remain for the manufacturing overcapacity case. The main instrument aimed at Chinese clean-tech is nearly out of road.
The trade data is Census, solid and public. The legal case is the authors' reading of untested law, Reade was general counsel at the US Trade Representative, so it's informed, but it's a prediction, not a finding. And the real weakness: the tariffs landed in July twenty twenty-six, and the effectiveness evidence is twenty twenty-four and twenty twenty-five data. They're judging a one-month-old policy on last year's numbers.
What to watch 9:35
Three things. Whether refining capacity outside China actually catches up with the new mines, that IEA cascade from fifty thousand tonnes of mining to eighteen thousand of magnets is the number to track. Whether Chinese production share in Thailand converges on its sales share. And whether those seven and a half points of tariff headroom get spent.
And note the contradiction, because it's real. Carnegie says closing the battery cost gap needs a tariff of at least thirty percent. CSIS says the tariff track is nearly out of legal and political room. They may both be right, which would mean the remedy one report prices isn't available on the instrument the other one examines.