This increase is a direct reflection of the supply and demand imbalance in the global AI supply chain. All available silicon in the world is being bought up by hyperscalers for accelerator production. Samsung, recognizing its dominant role in the value chain (as one of the few alternative foundries to TSMC), is leveraging its market power to maximize margins. For the B2B market, this means an automatic increase in the cost of all future developments in AI hardware. The era of "cheap chips" has ended, and now every new model training cycle is becoming more expensive for corporations due to production capacity inflation.
Source: Samsung / Reuters
HardwareSamsungSemiconductorsB2BMacroeconomics