The smartest AI trade right now is upstream: molecules and powders
Last Updated: Feb. 12, 2026 at 8:37 a.m. ET
Citrini Research calls the highest-conviction AI trade in 2026 an upstream play: raw materials—specialty molecules and engineered powders—that feed semiconductor fabs, advanced packaging, batteries and precision manufacturing. The firm argues that while software moats are debated, the physical inputs that enable AI hardware are a clearer place to find durable demand.
Key thesis in one sentence
The current, most defensible AI investment thesis is not primarily in software platforms but in the raw materials (molecules and powders) that are essential to producing chips, cooling systems, and energy storage for AI compute.
Why molecules and powders matter for AI
- Semiconductors and advanced packaging require high-purity chemicals and specialty molecules (e.g., photoresists, deposition precursors, etchants). These inputs are critical to wafer fabrication and yield control.
- Metal and ceramic powders enable additive manufacturing, heat-sink production, and advanced interconnects used in GPUs and custom AI accelerators.
- Battery electrode powders and precursor molecules are central to energy-dense, high-cycle battery systems that power cloud-region data centers and edge infrastructure where power density matters.
- Thermal interface materials and phase-change compounds—which are molecular in nature—play a growing role in managing the power and cooling demands of next-generation AI hardware.
Each of these categories represents a node in the AI hardware supply chain where incremental increases in compute demand translate into sustained demand for materials rather than one-off software transitions.
Practical implications for traders and allocators
- Focus upstream: Evaluate companies that produce high-purity chemicals, specialty powders, and advanced materials used by fabs and battery makers rather than only software names.
- Monitor capacity constraints: Materials suppliers with constrained capacity or high barriers to scale can exhibit margin expansion when AI-driven orders increase.
- Watch procurement timelines: Materials often have longer lead times and qualification cycles; supply disruptions can create outsized pricing power for established suppliers.
- Consider portfolio exposure: Allocate across materials classes (chemical precursors, metal/ceramic powders, battery materials, thermal materials) to capture different demand vectors from AI growth.
Risks and counterarguments
- Technological substitution: Alternative fabrication methods or material chemistries could shift demand patterns, underscoring the need to evaluate supplier adaptability.
- Cyclical demand and capital cycles: Materials demand is tied to capital expenditures at fabs and data centers; downturns in capex can reduce near-term demand.
- Concentration and geopolitics: A small number of suppliers or geographic concentration can create policy and supply-risk exposure for investors.
How to evaluate companies in this trade
- Product qualification: Prioritize suppliers with certified, high-purity products already qualified by major fabs or battery OEMs.
- Scaling capability: Look for demonstrated ability to scale production without compromising quality—this includes proprietary processes and modular capacity expansion.
- Customer diversification: Firms with multiple end-market exposures (semiconductors, automotive batteries, additive manufacturing) can offer more stable earnings.
- Pricing power and margin profiles: Historical gross margins and the ability to pass through raw material inflation are key indicators of resilience.
Tickers and coverage to watch
Relevant tickers mentioned in trading conversations include AI and AFP. Traders should map these tickers to sector ETFs, materials suppliers, and industrial names as part of a broader supply-chain strategy.
Actionable checklist for institutional investors
- Conduct a supply-chain audit: Identify which molecules and powders are used in the AI hardware stack most relevant to your thesis.
- Engage with procurement teams: Qualify timelines and inventory policies at major customers to estimate lead indicators of materials demand.
- Stress-test scenarios: Model both accelerated AI adoption and slower capex cycles to understand demand elasticity for materials.
- Rebalance exposures: Allocate a portion of hardware exposure from software and services into upstream materials where conviction is higher.
Conclusion
Citrini Research's central insight reframes the AI investment debate: instead of exclusively hunting for long-term software moats, institutional capital should also consider upstream materials that are physically necessary to produce the compute backbone of generative AI. Molecules and powders represent a tangible, industrial layer of the AI value chain where demand growth can be more predictable and harder to disrupt than some downstream software economics.
Investors targeting this theme should emphasize suppliers with qualified products, scalable capacity, and diversified end markets, while managing supply-concentration and cyclical risks.
