The article explores the evolving relationship between edge computing and traditional cloud computing, emphasizing the gradual shift of computational power from centralized data centers to users' devices. It outlines how computing technology tends to concentrate resources initially before diffusing them across a wider landscape. For instance, AI functions that once relied heavily on centralized servers are increasingly being handled on local devices, making them more efficient and cost-effective. A smartphone today can perform tasks that previously required expensive workstations, illustrating this trend. Figma serves as a key example where local computation enhances user experience by rendering graphics on the device to avoid lag, while still leveraging cloud storage for collaboration. This blend of edge and cloud computing is evident in other applications too, such as Netflix and Google Maps, where local resources are essential for performance. However, the article clarifies that edge computing is distinct from local access; the former refers to where computations occur, while the latter is about what data the computation can utilize. Many users still depend on cloud-based inference, as their local machines often serve to orchestrate actions rather than perform complex calculations. The potential for local access is growing, driven by innovations like Claude Code and Cowork, which aim to provide AI models direct access to local file systems. While some applications today utilize remote local access, the trend suggests a future where more computation occurs on devices themselves. The article notes that demand for inference is skyrocketing, outpacing the capacity of existing data centers, which are rapidly expanding to meet this need. Edge inference is seen as a looming threat to traditional data center operations, with companies like Apple and Microsoft investing in hardware capable of running advanced models locally. As consumer hardware becomes more capable, the ability to perform inference on devices will significantly impact user experience. The gap between cloud-based and on-device performance is narrowing, with research indicating that state-of-the-art results from the cloud can be replicated on consumer hardware within a year. This shift will likely redefine how users interact with AI, making local capabilities increasingly important as the technology continues to evolve.