The visible layer of the AI trade is crowded and expensive. Investors have piled into accelerator designers, hyperscalers building data center campuses, and software companies renting intelligence to the rest of us. The physical inputs that make any of it possible sit further upstream, and almost no one in a typical growth portfolio owns them directly. That gap is what a recent Animal Spirits Talk Your Book episode with Sprott’s Steve Schofstall explored, alongside hosts Michael Batnick and Ben Carlson.
The vehicle in question is the Sprott Critical Materials ETF (NYSEARCA:SETM), which sits in a corner of the market that has quietly performed well amid the intensifying AI capex cycle. The fund is up 53.7% over the trailing year and 14.14% in the past month alone, which is the kind of move that usually attracts attention only after it happens. The question worth answering is whether critical materials genuinely deserve the picks and shovels label, or whether that phrase dresses up an old-fashioned commodity cycle in fashionable clothing.
How the Trade Is Framed
Sprott is a specialist in this territory. The firm manages roughly $65 billion in assets and is best known for physical metal trusts covering precious metals, copper, and uranium, as well as 13 US ETFs and 5 UCITS-wrapped funds in Europe. Its identity is built around the economy’s physical layer.
Batnick set the framing bluntly. “You wouldn’t have ever really thought of commodities as a technology play. Back in the day,” he said, and that instinct is exactly why the trade is under-owned. Commodities have historically been treated as a hedge on inflation or a bet on China, not as a claim on a computing buildout.
Carlson extended the argument past mining into anything the buildout physically consumes. “all of the glass they make for fiber optic cables, all the data center stuff, that all needs a gigantic amount of materials, some of them critical,” he said. A company like Corning becomes part of the same story because its optical products are effectively a pass-through for materials.
The word critical carries specific weight. The designation is issued by a governing body rather than being marketing language, and that matters because it ties the fund to a supply chain that policymakers already treat as strategic.
Rare Earths Are Not Actually Rare
The most useful correction came from Carlson. “rare earths are not that rare,” he said, and that one sentence rewrites the entire investment case.
The binding constraint is refining and processing capacity. European auditors have documented that China produces 95% of the world’s refined gallium, and the pattern repeats across several inputs that semiconductor and power electronics supply chains cannot quickly substitute away from.
That distinction separates this from an ordinary commodity cycle. A copper bull market ends when price draws out enough new supply, because the mines and smelters exist. Refining capacity for specialty materials takes years to permit and build, and the geographic concentration of that capacity means Western buyers will pay whatever the marginal ton cost is during a demand surge.
This is where the picks-and-shovels framing earns its keep. The AI buildout is a bet that a supply chain built for a smaller industrial world will get repriced as data center construction, grid upgrades, and electrified transport all pull from the same processing bottleneck at once. The materials sit one layer below the suppliers most investors think of first, and we profiled seven of those non-chip AI beneficiaries, from power to cooling to networking, in a free report you can grab here.
The Verdict and the Risks
The real weakness in this thesis is that commodity producers are cyclical, capital-intensive, and price takers with almost no control over the prices they receive. A refiner that mints money at today’s prices can look completely different two years into a downturn, and the equities move harder than the underlying commodities in both directions.
Demand forecasts tied to a technology buildout have also historically overshot the actual buildout. Fiber optic capacity in the late 1990s is the reference case, and anyone extrapolating current AI capex straight through the next decade repeats an old mistake. The materials thesis survives that risk better than the chip thesis does, because the processing bottleneck is real regardless of how quickly hyperscaler orders grow.
Critical materials genuinely qualify as picks and shovels, but the specific edge is the refining chokepoint, and an investor who owns this exposure is really underwriting the slow rebuild of Western processing capacity. That is a durable position to hold, provided the investor accepts that the ride will look nothing like owning software.
The stock is up 11.78% year-to-date after a sharp recent run, which is worth keeping in mind before treating any entry point as urgent. The idea is more interesting than the price action at this exact moment, and the case for owning it rests on the next several years rather than the next several weeks.
Contact [email protected] for any questions or corrections.
Top Gaining Stocks
Top Losing Stocks