Amazon's custom silicon business has surpassed an annual revenue run rate of USD 25 billion, a figure that primarily encompasses chips developed for Amazon Web Services (AWS). This disclosure offers a rare financial insight into a closely observed segment of the cloud market, as major technology firms intensify investments in processors designed for artificial intelligence (AI) and data centre workloads. Custom silicon has emerged as a strategic imperative for cloud providers, driven by escalating demand for AI computing, allowing companies to manage costs, enhance efficiency, and mitigate dependence on external semiconductor vendors. Amazon's chip development supports various AWS product lines, including general cloud computing processors and specialised chips for machine learning, positioned as part of a strategy to provide customers with greater flexibility in infrastructure used to train and deploy AI models. The USD 25 billion run-rate figure measures the current pace of sales, rather than a booked annual revenue total. Rival hyperscalers, including Microsoft, Google, and Meta, are pursuing similar strategies, increasing investments in internally designed chips as AI demand strains capacity, pricing, and supply chains. While Nvidia remains a dominant supplier of AI processors, major cloud groups are seeking alternative solutions. Amazon's custom silicon project predates the recent surge in generative AI, reflecting a sustained, long-term investment by AWS. Customers are likely to assess this milestone based on availability and cost, as deeper investment in proprietary hardware by cloud providers often creates a trade-off for customers between using tailored tools and maintaining flexibility.