MP Materials Locks In Nine-Figure Gadolinium Deal as Litinsky Warns of Spreading Rare-Earth Supply Panic in Aerospace
Q2 2026 earnings call, August 6, 2026
MP Materials used its second-quarter call to unveil a new heavy rare earth supply agreement and a fresh industrial coordination initiative for the drone sector, while management delivered some of its most pointed commentary yet on how tight the global rare earth supply chain has become. The results themselves were solid but unremarkable within the range investors have come to expect; the more consequential disclosures were about what is happening in the market around MP, and how the company is positioning to capture it.
A quiet nine-figure deal signals heavy rare earth optionality
In July, MP signed a long-term agreement to supply gadolinium oxide to a large U.S. aerospace and defense manufacturer, a deal CEO James Litinsky described as "expected to be a sizable nine-figure deal in total over multiple years." CFO Ryan Corbett was explicit that the contract carries locked-in economics rather than spot exposure, calling it evidence of MP's "ability to continue to drive incremental value out of the world-class Mountain Pass asset." The gadolinium byproduct emerges from separating out dysprosium and terbium ahead of the company's planned 2028 samarium ramp, and management signaled more such deals are likely, specifically flagging yttrium as the next element under commercial evaluation. Notably, Corbett said MP had held gadolinium product available for sale "for a very, very long time" but waited to announce until it found a partner offering the right risk-adjusted return, a comment that suggests the company is deliberately withholding disclosure until commercial terms are optimized rather than rushing headlines.
Project Swarm: MP moves to coordinate drone-industry demand
Perhaps the most novel disclosure was Project Swarm, an initiative MP has run over the past several months with U.S. and allied drone manufacturers to aggregate and standardize future magnet demand. Litinsky explained the rationale bluntly: dozens of venture-backed drone companies are individually too small to matter, noting "the entire American drone industry would still be materially smaller than just our GM business as it stands today." Rather than requiring early-stage companies to commit to long-term purchase volumes before their products mature, Swarm lets them reserve future manufacturing capacity through subscription agreements while continuing to iterate. Litinsky tied this directly to the 10X facility's magnet grades, which will be standardized around Department of War preferred specifications, and described a strategic obligation behind the initiative, saying MP wants to ensure "the instruments of warfare and national security are taken care of" before fully allocating 10X capacity to commercial customers.
Evidence of real supply chain stress beyond MP's own numbers
Management pointed to outside evidence that the rare earth shortage is beginning to bite in ways the market hasn't fully priced. Litinsky cited Honeywell Aerospace losing roughly a quarter of its market value in a single session after attributing the disruption to a shortfall of just $15 million in upstream supply, framing it as proof of how disproportionately small material shortages can cascade through downstream value. "That is the extent of the problem," he said, arguing this is not isolated to MP's own customer base but reflects a broader "controlled scarcity" regime emerging from China's export licensing on heavies and magnets. He cautioned against complacency, describing the current equilibrium as an "uneasy detente" rather than a stable state.
Magnetics ramp on track, but near-term financials will stay lumpy
MP reaffirmed that commercial magnet shipments to General Motors will begin in the fourth quarter, following a quarter in which magnets were delivered for in-vehicle qualification testing. Corbett cautioned that the transition from precursor product sales to finished magnet sales will be "definitionally lumpy and nonlinear," with roughly $46 million of prepaid precursor revenue still to be recognized over the next three to four quarters on a declining basis. Magnetics segment EBITDA actually dipped sequentially this quarter even as the company posted its best-ever metal production quarter, because start-up costs tied to magnet production are not yet inventoriable and are being expensed rather than capitalized. Management pushed back on a question implying weak new contract flow for Independence and 10X capacity, clarifying that Independence is essentially fully sold out between GM and Apple, and 10X is fully committed to the Department of War, with the expectation that most of that capacity will eventually be re-contracted to commercial industrial customers at attractive terms once the facility is operational. Construction news capped the call: Litinsky announced the company achieved vertical construction on the 10X site during the call itself, a milestone he treated as a real-time proof point of execution pace.
Financial performance: modest beat with a cost drag from timing
Consolidated revenue and PPA income more than doubled year-over-year to $126.1 million, driven by a 127% increase in NdPr sales volumes, with adjusted EBITDA of $28.5 million, up $41 million year-over-year. NdPr production came in at 840 metric tons, up 41%, despite an extended planned maintenance shutdown at Mountain Pass in April. Materials segment adjusted EBITDA was $32.5 million on $113.2 million of revenue plus PPA income. Management guided to NdPr oxide realized pricing in the high $90s per kilogram for the third quarter, with PPA income of roughly $10 per kilogram, a step down from current spot pricing near $110 per kilogram, largely because stockpiled inventory carries minimal additional PPA benefit. Corbett guided Q3 materials sales volumes to be roughly flat sequentially, while Chief Operating Officer Michael Rosenthal said he expects NdPr production to exceed 1,000 metric tons in the third quarter. The balance sheet remains strong, with $1.45 billion in cash and short-term investments after $230.3 million of capital expenditure in the quarter, including an $80 million acquisition of the 10X site. Full-year capex guidance was maintained at $500 million to $600 million.
Heavy rare earth buildout and cost trajectory
Rosenthal detailed meaningful operational progress: mechanical completion of the first heavy rare earth separation circuit in May, with feed introduction imminent and terbium and dysprosium production still expected later this year. The company is also progressing engineering and procurement on both its samarium program, targeting first production in 2028, and its newly announced gadolinium separation project on a similar timeline. Ground will be broken this month on an expanded facility area to house magnet recycling alongside additional heavy rare earth separation capacity. On costs, Corbett said the path to a lower cost structure is "very clear," contingent on consistent production at target throughput, process efficiency gains, reduced maintenance intensity, and benefits from the chlor-alkali recommissioning, with the bulk of those improvements expected to build through 2027.
The NdPr binding constraint thesis, and why heavies may matter less over time
Despite growing attention to heavy rare earths, Litinsky reiterated his conviction that NdPr, not heavies, remains the true bottleneck across the industry. He argued that most large future demand pools for magnets, including robotics, disk drives, and other physical AI applications, are increasingly "no heavy or very limited heavy," a shift he credits to advances in Grain Boundary Diffusion technology and broader manufacturing intellectual property. He drew a historical parallel to Micron, whose enterprise value went from $50 billion to $1 trillion within three years of ChatGPT's launch as a memory squeeze materialized, suggesting a similar dynamic could hit specific rare earth magnet end-markets without clear advance warning. On defense-specific demand, Litinsky estimated global drone production near 12 million units annually could translate to 500 to 1,000 tons of magnet demand today, current demand he called "immaterial" relative to total industry volume, but one he expects to become "quite large" over a three-to-five-year horizon given MP's position as the only qualified domestic supplier for that dual-use application.
Management's broader framing: an infrastructure phase, not the end state
Litinsky closed prepared remarks with an extended historical analogy comparing the current AI infrastructure debate to earlier general purpose technology cycles, including railroads, electrification, and the internet, arguing that markets historically underprice the eventual downstream economic impact while fixating on capex return questions during the buildout phase. "The larger opportunity often emerges when that infrastructure becomes pervasive enough for entrepreneurs to build businesses that redefine entire industries," he said, framing MP's investment thesis around the transition from AI compute to physical deployment in manufacturing, transport, and defense equipment, a transition he characterizes as a structural rather than cyclical shift favoring domestic critical materials production.