Microsoft and Google are reshaping their search APIs, signaling a shift in how these tech giants control access to web data. Microsoft is retiring the Bing Search API, while Google has limited its API to just 10 results per query. These changes reflect a strategic move to funnel developers toward their own cloud ecosystems, particularly in the context of generative AI. Instead of providing broad access to search results, both companies are focusing on AI-driven applications, restricting external access to manage costs and minimize automated scraping. As traditional search APIs fade, new players like Perplexity and Parallel are stepping in, specifically designed for AI workloads. These companies prioritize retrieval quality and low latency, crucial for retrieval-augmented generation (RAG) systems. Perplexity has been noted for outperforming Google in relevance for RAG tasks, while Parallel, founded by former Twitter CEO Parag Agrawal, recently claimed better results using Perplexity’s own evaluation tool. This competitive environment indicates a resurgence in the search API market, driven by AI-native infrastructure. Underpinning this new generation of search APIs is Vespa, an open-source engine that supports large-scale retrieval and machine learning inference. The architecture is evolving, integrating search infrastructure directly into AI systems. As search technology adapts, the focus is shifting from merely returning links to delivering context and answers tailored for AI applications. The dynamics of search are changing, with incumbents tightening control and innovators pushing for more open access.