20-second highlights:

Only five days after the Executive Order (EO) 14415 (that we disclosed in detail on last week’s briefing), a set of interviews with 16 executives, investors, analysts and policymakers was published by Reuters (bear with us, please), reporting that the administration is weighing whether to extend access to some Chinese rare earth material past that same January 1 cutoff.

In other words, the administration is reconsidering returning with the waiver for certain items that would be virtually impossible to secure outside of China.

The subtlety of those interviews is that the market was not well aligned with the EO’s implications, and an avalanche of lobbying must have happened. 

The reason is arithmetic, and we gave some color last week on the CW30 briefing edition.

To put into perspective, let’s zoom in on the U.S.’ demand for NdFeB permanent magnets. It’s currently estimated to be around ~40,000 to ~50,000 t a year, including permanent magnets inside assembled components.

In contrast, the U.S.’ current disclosed industrial-scale installed capacity is around ~5,600 t, and a part of it is still ramping up:

And the pipeline of future magnet-making plants in the U.S. is relevant, and if all succeed, will reach levels close to its demand:

While the capacity is being duly ramped up and built, imports are the solution. The U.S. International Trade Commission data for imports of permanent magnets for consumption, considering only permanent magnets, not assembled parts and components, shows:

The POTUS signed a Presidential Determination under Section 101 of the Defense Production Act, delegating to the Secretary of Commerce the authority to institute export restrictions on recoverable critical minerals and materials.

The Determination names the specific classes: 

Note that it does not include other critical minerals such as copper scrap, which is already covered under Proclamation 10962 of July 30, 2025 (Adjusting Imports of Copper Into the United States). 

The White House officials briefing named tungsten and black mass as the target flows and put U.S. e-waste exports near 33,000 t per month.

No implementing rule exists yet, and no timeline was given.

Now compare that class list against what the West is building. Almost every near-term heavy rare earth unit outside China comes from scrap, because no new Western heavy mine delivers before 2028.

Caremag, the Carester plant at Lacq in southwestern France, is the largest single example.

At full ramp, it is expected to produce 800 tpa of NdPr oxide, 500 tpa of dysprosium oxide and 100 tpa of terbium oxide, which the parties put at roughly 15% of current world production of the two heavy oxides, from magnet recycling and concentrate.

USA Rare Earth took about 13.6% of Carester alongside InfraVia's French-state-seeded Critical Metals Fund, as we covered in CW30, and the plant carries around €216M of French and Japanese government support.

Section 6 of EO 14415 says a ‘foreign project financed or insured by EXIM or the DFC sits outside the sourcing prohibition’. So the credential regime tells Caremag it may sell into the US perimeter. 

But now, the recent Presidential Determination targets one of the classes of material Caremag needs in order to run: end-of-life magnets. Meaning Caremag’s France facility may not have easy access to that feed (as it was hoping for).

The party most exposed to a U.S. scrap export rule is not the Chinese refiner that the instrument was (supposedly) targeting, but rather, the most exposed and harmed are the allied recyclers whose feed contract runs through a U.S. collection network.

Energy Fuels just began construction of its heavy rare earth expansion at White Mesa, Utah.

Phase 1 adds separated oxides of up to:

The Tb and Dy circuits are targeted for end-2027, the rest for end-2028, at a capital cost near US$104M funded from government loans and grants plus US$0.96bn of working capital held at March 31.

A 2029 step would take the mill to 6,294 tpa NdPr, 80 tpa Tb and 288 tpa Dy.

The schedule is tied to anticipated feed from the Donald Rare Earth and Mineral Sands project joint venture with Astron Corporation in Victoria, Australia.

Astron targets a Phase 1 final investment decision now in Q3 2026, with debt discussions running through Export Finance Australia. ASM shareholders vote on the Energy Fuels acquisition of the Korean Metals Plant on August 12, and the Vacuumschmelze close remains set for early 2027.

Now put the tonnes next to the date they are supposed to serve:

Neo Performance Materials' chief financial officer put the same point in one sentence at an investor conference, which we carried in CW30: the trigger for a commercial-scale heavy line would be new mines producing something the company could separate.

He was describing a capacity decision waiting on a feed decision.

For the heavies, the only feed that arrives before 2028 is scrap, which is precisely the class the U.S. moved to take authority over three days after White Mesa, the largest new Western circuit currently under construction, broke ground.

Nature Communications just published work by Ryan Manzuk of Princeton and Sebastien Philippe of the University of Wisconsin-Madison estimating that 2,000 to 5,000 t of natural uranium left the Democratic Republic of the Congo embedded in cobalt-hydroxide shipments between 2000 and 2024, against no reported DRC uranium production for decades.

Less than 10% appears to have been declared to the International Atomic Energy Agency (IAEA). And roughly 65% went to Chinese-owned refiners, which handle around 95% of DRC cobalt. A further 1,000 to 4,000 t is estimated to sit in mine tailings.

In this case, the method matters quite a lot, even more than the headline number.

Of 31 mining operations examined, three consistently imported enough phosphoric acid to strip uranium before export, and by 2024 roughly one tonne of exported cobalt in five showed any sign of removal. 

The authors put China's embedded 2024 imports near 40 t against a 10 t Additional Protocol reporting threshold.

What is curious (in a good way) is that The Financial Times and Lighthouse Reports published a parallel investigation the same day.

Set that beside Section 3 of EO 14415, which requires prime contractors and subcontractors at any tier to file an indentured bill of materials tracing every component, part, software element, and material back to raw-material origin, with guidance due within 180 days and implementing regulations 90 days after that.

Again, this is just a demonstration that the shareholder premium varies more depending on the ‘national instrument’, more than the mineral itself.

On July 28, Codelco chairman Bernardo Fontaine told Radio Infinita there is no possibility of reaching the 1.7 Mt target within four or five years, and that the company has missed its projections for seven years running.

Guidance for 2026 stands at 1.331 to 1.357 Mt.

And the first-quarter output fell 8.1% year on year to 272,000 t with C1 costs up 10% to 231.8 cents/lb, after a 2025 in which record capital expenditure of US$5.07bn bought a 0.5% production increase.

In short, declining production with increasing operating costs is a strong signal of potential price appreciation for the mineral (or at least a pressure to increase its price).

Adding to that, most long-run copper balances still carry a Codelco recovery toward 1.7 Mt. Removing it takes ~350,000 to ~400,000 t out of the supply side, which is roughly a full year of a top-ten mine. 

Interesting seeing this against the exchanges’ moves over the week:

Still on costs, the IEA warned this month that sulfuric acid shortages put more than a seventh of global copper output at risk. You know this already because you read it on CMJ’s CW22 edition.

Codelco choosing margin over tonnage also removes the swing supply that has capped every previous copper rally, which is a structural change to the balance rather than a quarterly miss.

The market appears to treat three things as settled:

What this framing may be underweighting:

What would change this thesis, each item dated and checkable:

Westinghouse Electric confidentially submitted a draft Form S-1 to the SEC on July 31 for a proposed IPO, disclosed by Cameco alongside its second-quarter results. Cameco holds 49% and Brookfield Renewable Partners 51%, after acquiring the AP1000 vendor at roughly US$8bn enterprise value in 2023. Share count and price range are undetermined. The June 23 conditional commitment of up to US$17.5bn for long-lead equipment on as many as ten AP1000s sits inside the asset now being marketed, so the reactor order book gets a public price before the separative work units to fuel it have been contracted, against a non-Russian enrichment pool that mostly arrives 2028 to 2032. X-Energy and Standard Nuclear listed via traditional IPOs this year and Holtec filed in July, so the nuclear build is being financed on public equity rather than utility balance sheets. Watch whether the public S-1 quantifies enrichment supply commitments, or leaves the fuel-cycle gap to the risk factors.

Anglo American and Teck named the future executive leadership team of Anglo Teck on July 30. Anglo takes three of the four top roles, with Duncan Wanblad as group chief executive, John Heasley as chief financial officer and Ruben Fernandes as chief operating officer; Teck's Jonathan Price becomes deputy chief executive and chief strategy officer. Headquarters will be Vancouver. The US$53bn all-share merger targets around US$800M of annual pre-tax synergies by year four, plus US$1.4bn of average annual Collahuasi and Quebrada Blanca integration synergies from 2030 to 2049, with final approvals expected between September 2026 and March 2027. We flagged in CW30 that Teck's up-to-CAN$400M Canada Growth Fund instrument for Trail was still at definitive-agreement stage, and that an unsigned instrument transfers awkwardly. Trail is Canada's only germanium producer and the only supplier of germanium dioxide to the United States, so Canada's flagship critical-minerals instrument will now be administered by an Anglo-led board. Watch whether the Trail definitive agreements sign before completion, and whether the negotiated state offtake rights survive the change of control in substance as well as in contract.

Lynas and LS Eco Energy signed binding agreements on July 27 for cross-subscription of zero-interest convertible instruments of about A$29M each, maturing in five years and convertible from year three, formalising the March framework under which LS Eco Energy builds a rare earth metallisation plant in Vietnam using Lynas oxide, with staged circuits and samarium metal as the stated priority. LS Eco Energy is also running a feasibility study on a US magnet plant, in discussions with Virginia. Samarium sits on both the DFARS covered list and China's control list, and almost nobody outside China makes samarium metal, so the priority ordering is a compliance decision rather than a market-size decision. Lynas is positioning as the oxide cornerstone around which partners build metal and magnets, while Energy Fuels and MP buy the downstream outright, and those two capital models will not survive the same deadline slip equally. Watch for a definitive metal supply agreement and the Virginia siting decision.

Harena Rare Earths drew a State Department framing of its Ampasindava ionic clay project in Madagascar, with Washington describing its Africa minerals strategy as targeting sectors long dominated by opaque, predatory investment. Executive chairman Ivan Murphy named MP Materials, USA Rare Earth and Solvay as potential refining partners and said an exploitation permit was expected within weeks. The project targets first production mid-2028, roughly 4,000 t/yr of rare earth oxides including about 1,700 t of magnet oxides, at a total cost near US$150M against a DFC commitment of up to US$4.84M. Under Section 6 of EO 14415, a DFC-supported foreign project sits inside the sourcing perimeter, so a US$4.84M ticket buys a Madagascan project a compliance credential worth a multiple of the check. The live question is not the funding but the routing, because whoever separates that ionic clay feed captures the node the deposit cannot. Watch the exploitation permit and the first named separation counterparty.

Prospect Lithium Zimbabwe brought Africa's first lithium sulfate plant to full operation at Goromonzi, a US$400M Chinese-owned facility, confirmed on a ministerial tour reported July 26. Harare confirmed that 13 further minerals, from cobalt to platinum group metals and rare earth elements, cannot be exported in raw form from the start of 2027. More than US$1bn of investment has been reported since the February raw-export freeze, against 1.128 Mt of spodumene concentrate exported in the year to December 2025. The beneficiation mandate moved the processing step onshore without moving the ownership, so for Western buyers the intermediate now leaves Zimbabwe already inside a Chinese balance sheet, which is a worse starting point than the raw-export status quo it replaced. Watch whether any non-Chinese operator commissions a Zimbabwean conversion plant before the January 1, 2027 extension takes effect.

Hercules Metals recruited the Arizona Sonoran Copper management team to advance the Leviathan copper discovery in Idaho, with George Ogilvie as president and chief executive from September 1 and six further executives joining across finance, legal, permitting, capital markets and resource geology. The incoming group took Arizona Sonoran from about C$125M of implied equity value in 2021 to Hudbay's roughly C$2.0bn acquisition in June, raising more than C$300M for the Cactus project along the way. This is a capital-markets signal rather than an operating one. A proven copper exit team choosing a US jurisdiction, immediately after Hudbay paid C$2.0bn and while Anglo Teck consolidates the top of the market, points at where the sector expects the next transaction premium to originate. Watch for a maiden resource or published economics at Leviathan, which is what would convert the appointment into an asset.

The Department of Energy invited nuclear onto its largest legacy enrichment site, and the winning bid chose gas.

DOE recently selected Brookfield to develop, and NextEra to power, a data center campus of more than US$100bn on portions of the 3,556-acre Paducah Site in western Kentucky, the former gaseous diffusion uranium enrichment plant. The design carries up to 4.6 GW of dedicated generation, 2 GW of gas and up to 2.6 GW of batteries, supporting 1.8 GW of utility capacity and over 1.2 GW of compute by 2032. The November 2025 Request for Offers explicitly invited proposals integrating small modular reactors. None were selected. Brookfield sits on both sides of the week, as Paducah developer and as Westinghouse's majority owner, which makes the choice the clearest available read on whether SMRs clear on a commercial timeline today.

The copper scrap carve-out isn't the end of the story. It now runs on a second, separate export-control track with its own clock.

The July 30 determination carves copper scrap out of its scope, because copper scrap already sits under a separate presidential proclamation where export licensing on high-quality scrap is scheduled alongside the 2027 and 2028 duty steps. The United States therefore has scrap export control advancing on two independent legal instruments, one for copper and one for everything else, on different timetables and under different agencies. Almost nobody is modeling either, and the copper track is the one with dates already attached.

The largest ex-China separator is being run by an interim chief executive.

The July 27 Lynas cross-subscription with LS Eco Energy was issued in the name of interim chief executive Pol Le Roux, where the March framework agreement behind it carried Amanda Lacaze's. The change lands in the same quarter the company disclosed that its Malaysian heavy expansion now costs A$294M against A$180M guided, a 63% increase it attributed partly to Chinese export curbs on processing equipment. Governance transition at the one Western operator already producing heavy oxide is a project-schedule risk before it is a management story.