Was this email forwarded to you? Sign up here for Telemetry.

The gallium nitride (GaN) chip in an F-35's radar module, the neodymium magnets in an electric vehicle (EV) drivetrain, the germanium lens in a night-vision scope: these components share a single point of failure. Refining and processing capacity for almost all of them is concentrated among a small number of firms in China, a dependency that has survived decades of reports, hearings, and executive orders because no policy package addressed enough failure modes at once.

But then, between January 15 and February 26 of this year, four coördinated federal moves landed in sequence. A Section 232 proclamation covering all 60 minerals on the U.S. Geological Survey (USGS) critical list plus uranium, with a 180-day tariff negotiation window expiring ~July 2026. Project Vault: $12 billion in Export-Import Bank (EXIM) -backed strategic stockpile commitments, with Boeing, GE Vernova, and Western Digital as confirmed participants. A U.S. Trade Representative (USTR) price floor proposal establishing legally binding minimum returns for domestic mining and processing (comment period closes March 19). And ~$1 billion in Department of Energy (DOE) grants spanning the full technology readiness spectrum. That’s a lot of movement!

Notably, each addresses a distinct failure mode that has kept domestic mining, processing, and recycling ventures from reaching bankability. The Section 232 proclamation creates urgency; Project Vault creates offtake. The price floor attacks revenue uncertainty, and the DOE grants reduce capital risk. Together, they reshape the funding environment for tough tech founders working across the supply chain, from earthward extraction through the starward applications that increasingly drive demand.

The startup response across that supply chain, however, is wildly uneven. Some nodes are dense with activity; others are alarmingly vacant.

I mapped all fifty minerals on the USGS critical list by domestic processing status, startup activity, and demand trajectory into an interactive I’m calling The Periodic Table of Who’s Actually Building.

Venture investment in U.S. rare earth startups jumped 3,000% in a single year (2024 to 2025), capturing 90% of global funding in the category. That concentration tells a clear story about where conviction is flowing. Rare earth processing and magnets is where the most consequential buildout is manifesting: Phoenix Tailings running zero-emission refining in Massachusetts, MP Materials building a campus targeting 10,000 metric tons of NdFeB magnets per year with a Pentagon offtake at ~2x market price, Noveon Magnetics producing 2,000 tonnes/year with GM and ABB supply agreements already signed. Battery recycling is close behind: Redwood Materials building closed-loop supply chains at EV-battery scale, Ascend Elements advancing graphite recycling under a DOE cost-sharing grant, Princeton NuEnergy scaling black mass production to 15,000 tons/year. And the intelligence community is paying attention: In-Q-Tel backed Alta Resource Technologies’ protein-based mineral extraction from brines, a technique that could turn geothermal wastewater into lithium feedstock and that, notably, doesn’t care whether the brine is in California or on Mars.

The vacancies on or periodic table are harder to explain. Let’s look at Gallium: Asian processors produce 98.7% of global supply, and no domestic facility has recovered primary gallium in nearly four decades, despite gallium nitride semiconductor demand growing at 29.5% CAGR through 2030. The DOE put $6 million toward the problem. I have not seen a single venture-backed startup is working on gallium recovery at scale. There are efforts to be sure, such as Anactisis in Pennsylvania, Supra Elemental Recovery in Texas, and Flash Metals / Metallium (Texas / Perth, Australia), but not commensurate with the criticality of the mineral. The same is true for germanium (Chinese processors control ~60% of global output, essential for night vision and secure military communications), despite the fact that novel extraction pathways for both elements are exactly the kind of separation chemistry that transfers to regolith beneficiation off-planet. For magnet-to-magnet recycling, the first generation of EV motors and wind turbines is reaching end-of-life; JL MAG alone produces 25,000 tonnes of NdFeB per year in China, roughly equal to the entire non-Chinese magnet industry. The UK’s HyProMag is further ahead than any American company on this. I have colleagues in the U.S. who literally build their own specialized magnets because there is simply no great domestic manufacturer from which to buy.

And there are opportunities that don’t need all the mucky-muck and occupational hazards of running a mine or a furnace: mineral supply chain software (scheduling, provenance-tracking, FEOC [foreign entity of concern] compliance, logistics optimization). As the domestic minerals industry scales from a handful of demonstration projects to dozens of commercial facilities, someone will need to build those liminal services. That’s an Atoms-meets-Bits opportunity with federal tailwinds and almost no startup activity, though it’s certainly less sexy than making yet another dog-walking app, right?

None of this is easy, of course. Producing refined critical minerals outside the established Chinese processing base still costs ~50% more due to labor, energy, and permitting differentials. Mining projects take 10-16 years from discovery to production; policy cycles last four. I think there is a longer arc worth holding in mind: every mineral on the USGS critical list exists in abundance on near-Earth asteroids and in lunar regolith, which means the processing techniques developed earthward become the knowledge base for extracting these same materials off-planet, a bridge I explored in Space for Earthlings and one that reframes today’s “rare earth” problem as tomorrow’s “common space” opportunity. The question is whether founders move into the white space before the window closes.

Comment period closes March 19. Section 232 negotiations expire ~July 2026. For the best policy overview of how we got here, 🎥 CSIS's year-in-review on critical minerals is essential viewing.

And be sure to explore the full interactive: The Periodic Table of Who's Actually Building →

It’s got fifty critical minerals and gives a look at domestic processing status. Click any square to see who's building and where there’re vacancies.

The U.S. Army consolidated 120+ separate procurement actions into a single 10-year enterprise vehicle for Anduril Industries, and this ‘consolidation model’ is a great signal: DoD is learning to buy from commercial-first companies the way commercial-first companies want to sell. (🎥 Palmer Luckey on The Axios Show, days after the announcement, is worth the watch.) • Form Energy's iron-air chemistry will get its first real stress test at scale: Google and Xcel Energy are deploying a 30 GWh system in Pine Island, Minnesota, storing energy by rusting and de-rusting iron pellets for up to 100 hours continuously. • The NRC quietly reclassified fusion reactors under the same regulatory framework as particle accelerators rather than nuclear plants… this single administrative move compresses a potential 20-year licensing timeline into a defined, finite pathway. • Samsung Electronics began volume production of 2nm Gate-All-Around chips, the first foundry at this node; more interesting is that the Taylor, Texas fab is coming online simultaneously, which means the U.S. will have domestic leading-edge capacity from a non-TSMC source for the first time in a generation.

No One Builds Alone.

/N1BA

Telemetry is written by JMill of The End Effector.