Ruthenium, a rare platinum-group metal recovered almost exclusively as a byproduct of platinum and palladium mining, touched $760 per troy ounce this month — surpassing the prior record of roughly $580 set in early 2022 — as AI infrastructure investment drives demand for the materials underpinning data center hardware. The metal plays a critical role in hard disk drive manufacturing, specifically as a thin underlayer on magnetic recording platters that increases storage density, making it indispensable for the mass-storage arrays used in AI training and inference deployments. Ruthenium is also used in advanced semiconductor fabrication as an electrode and barrier layer material, further compressing available supply as chip fabs ramp capacity to meet AI accelerator demand.

The supply constraint is structurally difficult to resolve. Because ruthenium can only be recovered as a byproduct of primary platinum and palladium extraction, output cannot simply be scaled up in response to higher prices without also expanding production of those metals. The world's dominant producers — Anglo American Platinum, Impala Platinum, and Sibanye-Stillwater in South Africa, along with Russia's Norilsk Nickel — face their own operational headwinds including energy costs, labor disputes, and geopolitical risk. Global annual ruthenium production is measured in just a few tonnes, leaving minimal buffer against demand spikes of this magnitude.

The AI infrastructure boom has created sustained demand across the full hardware stack — servers, storage, and the networking gear that ties data centers together. Hyperscalers including Microsoft, Google, Amazon, and Meta have committed hundreds of billions of dollars in data center capital expenditure, and HDD manufacturers Seagate and Western Digital have reported strong order backlogs driven specifically by AI storage workloads. High-capacity drives built on newer recording technologies such as HAMR and MAMR — both of which rely on ruthenium-based platter coatings — are particularly in demand for the dense storage arrays that feed large-scale model training and retrieval pipelines.

The pattern rhymes with the cobalt and lithium supply crunches that blindsided the electric vehicle industry in the early 2020s. The AI sector is walking into a similar strategic problem: geographically concentrated, slow-to-expand <a href="/news/2026-03-14-qatar-helium-disruption-threatens-ai-chip-supply-chain-tsmc-and-hynix-most-exposed">critical mineral dependencies</a> that GPU availability projections and power grid forecasts rarely account for. The ruthenium price signal is worth tracking precisely because it sits outside the usual AI infrastructure conversation — and that blind spot is exactly how commodity crunches tend to develop. Seagate's last earnings call flagged tight platter material sourcing as a watch item; if HDD manufacturers start managing allocation, the knock-on effects for hyperscaler storage build-outs could arrive faster than the market currently expects.