AI data center spending drives growth in semiconductor market - Construction Dive

According to a recent Gartner forecast, AI data center spending will account for more than half of global semiconductor revenue within the next four years, up from 36.5% in 2026. This shift reflects a... Read original →

AI data center spending is fundamentally reshaping the semiconductor industry, with its revenue share projected to exceed 50% within four years. This shift is driven by explosive demand for compute and memory from large language models, pushing 3nm and below EUV processes into mainstream adoption, while memory chip revenues are expected to account for 54% of total semiconductor income in 2026. However, enterprises are grappling with escalating AI costs, particularly from repetitive data processing via continuous looping agents, prompting a transition from flat-rate to usage-based pricing models. Vendors like Snowflake, AWS, and Oracle are responding with cost-control features, while companies such as Travelers Insurance are developing in-house LLMs to reduce reliance on expensive frontier models. China’s progress in open-source AI models offers viable alternatives. In the near term, a multi-tiered computing architecture—leveraging local devices for routine tasks and high-performance systems for complex workloads—will become standard for optimizing performance and cost efficiency. This transformation not only redefines technical strategies but also reshapes global supply chains, especially pressuring US and European vendors.