The rapid expansion of artificial intelligence, cloud computing, and large-scale data storage is accelerating the retirement of older data center equipment. Millions of servers and hard disk drives (HDDs) are entering the Information Technology Asset Disposition (ITAD) stream, creating a significant urban-mining opportunity. In addition to familiar materials such as gold, copper, aluminum, and palladium, this equipment contains a less visible but increasingly strategic metal: ruthenium.
Ruthenium is one of Earth’s rarest stable elements and belongs to the platinum group metals. It is exceptionally hard, chemically resistant, and capable of maintaining reliable electrical and magnetic performance at nanometer scales. These properties make it valuable in chemical catalysts, hydrogen-production systems, semiconductors, capacitors, advanced memory, and data-storage technologies.
Its supply is also highly vulnerable. There are no dedicated ruthenium mines; it is recovered entirely as a byproduct of platinum and palladium production. Approximately 85% to 90% of primary output comes from South Africa, with smaller quantities produced in Russia, Zimbabwe, and North America. Because ruthenium contributes only a small share of a platinum mine’s revenue, higher ruthenium prices cannot quickly stimulate additional production. Recent estimates indicate a significant supply deficit. In 2025, global ruthenium demand reached approximately 1.224 million troy ounces, compared with a primary supply of about 954,000 ounces, a shortfall of 270,000 ounces. Ruthenium is not traded on a major public exchange and has no established futures market, making its price opaque and difficult for manufacturers to hedge. Reported prices rose to around $800 per ounce in 2025, with later estimates reaching as high as $1,750 per ounce.
The main source of ruthenium in retired servers is the Perpendicular Magnetic Recording, or PMR, hard drive. PMR was introduced to overcome the physical limits of older longitudinal magnetic recording. As magnetic bits became smaller, ordinary thermal energy could destabilize them, risking data loss. PMR addressed this problem by orienting the magnetic regions vertically rather than horizontally.
Ruthenium is essential to the multilayer structure of a PMR platter. One ultrathin layer, sometimes only a few atoms thick, separates two magnetic films and forces them to align in opposite directions. This antiferromagnetic coupling stabilizes tiny magnetic bits and was famously described by IBM as “pixie dust.”
A second ruthenium layer acts as a crystalline foundation for the magnetic recording medium. Because ruthenium has a hexagonal close-packed structure, it promotes the growth of cobalt-platinum-chromium alloys into uniform, isolated columns. This allows data to be stored at much higher densities while remaining thermally stable. Although these coatings are only nanometers thick, they cover both sides of every platter. A representative 3.5-inch platter may contain approximately 2.46 milligrams of ruthenium, assuming a combined effective ruthenium-layer thickness of 15 nanometers. An enterprise HDD containing eight platters would therefore hold roughly 19.65 milligrams.
Servers also contain smaller amounts of ruthenium in thick-film resistors on printed circuit boards. These components often use ruthenium dioxide because it resists moisture, remains stable during high-temperature manufacturing, and provides reliable electrical resistance. A single resistor contains very little, but thousands of components can produce a measurable secondary source.
A representative storage server with 12 eight-platter HDDs may contain approximately 236 milligrams of ruthenium across its drives. Adding an estimated 19 milligrams from circuit boards produces a modeled total of about 255 milligrams per server.
At a reference price of $1,750 per troy ounce, equivalent to about $56.26 per gram, the total represents a theoretical contained-metal value of roughly $430 per rack. This figure is not the same as recoverable profit. It does not account for equipment collection, secure data destruction, dismantling, transportation, chemical processing, refining losses, or environmental compliance. Drive designs also vary by model, age, platter count, and ruthenium-layer composition. Laboratory assays are therefore necessary before commercial decisions are made.
Nevertheless, hyperscale retirement programs can involve thousands or tens of thousands of racks. At that scale, ruthenium can represent millions of dollars in previously overlooked material value, particularly when recovered alongside gold, palladium, aluminum, copper, cobalt alloys, and rare-earth magnets.
A typical hard drive weighs more than 500 grams, but it may contain only a few tens of milligrams of ruthenium. Most of its mass consists of aluminum, steel, magnets, motors, and electronic assemblies. Processing complete drives without separation would therefore waste energy and chemical reagents.
Commercial recovery begins with dismantling or controlled mechanical processing. Platters are separated from aluminum chassis, motors, magnets, and circuit boards. Selected circuit-board components may be treated independently. Concentrating the ruthenium-bearing fractions is essential before advanced refining begins. Ruthenium is also extremely resistant to conventional leaching. Processes commonly used to dissolve gold, copper, or palladium may be ineffective against metallic ruthenium and ruthenium dioxide. Industrial recovery generally requires aggressive oxidative treatment that converts the material into soluble ruthenate compounds.
The dissolved ruthenium can then be oxidized to ruthenium tetroxide, a volatile compound that can be separated from most other metals by controlled distillation. Its vapor is captured in sealed absorption equipment and converted into a chloride solution. The ruthenium is then precipitated and reduced to produce high-purity metallic sponge suitable for reuse. This chemistry is extremely hazardous. Ruthenium tetroxide is highly toxic, strongly oxidizing, and capable of reacting violently with organic materials. The process may also involve chlorine, corrosive acids, strong alkalis, and high temperatures. Ruthenium recovery must therefore occur in specialized, licensed facilities with sealed reactors, gas scrubbers, continuous monitoring, and trained personnel. It is not suitable for informal or small-scale recycling.
The future of HDD technology will influence the amount of ruthenium available. Microwave-Assisted Magnetic Recording, or MAMR, generally retains a media structure similar to PMR and is therefore expected to preserve demand for ruthenium layers.
Heat-Assisted Magnetic Recording, or HAMR, represents a greater change. HAMR briefly heats the recording surface with a microscopic optical system, allowing data to be written to high-coercivity iron-platinum media. This architecture may eliminate the traditional ruthenium layers used in PMR drives. As HAMR expands in high-capacity enterprise storage, demand for ruthenium in new HDDs could gradually decline.
That transition does not eliminate the urban-mining opportunity. Instead, it creates a finite and potentially valuable recovery window. Billions of PMR drives produced over the past two decades are now approaching retirement. At the same time, primary ruthenium supply remains geographically concentrated and unable to respond quickly to growing demand from catalysts, semiconductors, hydrogen technologies, and advanced electronics. Old server racks are therefore more than obsolete hardware. They are distributed above-ground deposits containing one of the technology economy’s rarest metals. For ITAD companies and refiners capable of identifying PMR drives, separating their platters, aggregating sufficient volume, and processing the material safely, ruthenium could become an important source of additional value within the circular economy.
