While hyperscalers compete in LLM architectures, the hardware sector faces a severe shortage of raw materials. Ruthenium is critical for the production of thick-film resistors in advanced microchips and high-capacity hard drives (which are bought en masse to store training datasets). Supply chains torn apart by geopolitics cannot keep up with the demand from TSMC and other foundries. This means the cost of AI accelerators and data storage systems will continue to rise, inevitably affecting the prices cloud providers (AWS, Azure) charge end developers.
Source: Reuters
EconomyHardwareRutheniumSupply ChainCommodities