One year after its introduction, AlphaEvolve, an AI-powered evolutionary algorithm developed by Google DeepMind and Google Cloud, has transitioned from a research prototype to a critical tool for solving real-world challenges. Originally designed to iteratively discover optimized algorithms for complex mathematical problems, the system now accelerates scientific discovery and delivers significant commercial value. AlphaEvolve has become a core component of Google's internal infrastructure. It optimized the design of next-generation Tensor Processing Units (TPUs) by proposing a counterintuitive circuit design that was directly integrated into silicon. The system also discovered more efficient cache replacement policies, accomplishing in two days work that previously required months of human effort. Additionally, AlphaEvolve improved the efficiency of Google Spanner by refining log-structured merge-tree compaction heuristics, reducing write amplification by 20%. It further provided insights for compiler strategies that reduced software storage footprints by nearly 9%. Beyond internal tools, the technology is driving results for external enterprises across diverse sectors. In financial services, Klarna utilized AlphaEvolve to optimize a major transformer model, doubling training speeds while enhancing model quality. Semiconductor manufacturer Substrate applied the system to its computational lithography framework, achieving a multi-fold increase in runtime speed that enables larger, more advanced simulations. In logistics, FM Logistic used the algorithm to solve complex routing problems, improving efficiency by 10.4% and saving over 15,000 kilometers in annual distance traveled. Marketing firm WPP achieved a 10% accuracy gain in AI model components for high-dimensional campaign data. In the life sciences sector, Schrödinger applied AlphaEvolve to machine-learned force fields, achieving a fourfold speedup in training and inference, which significantly shortens research and development cycles for drug discovery and materials design. Pushmeet Kohli, Chief Scientist of Google Cloud, and Amin Vahdat, SVP and Chief Technologist of Google Cloud, highlighted that AlphaEvolve demonstrates the potential of self-improving algorithms to tackle societal and industrial problems. The system has helped refine DNA sequencing error correction, improved disaster prediction accuracy, and simulated power grid stabilization. It is also accelerating neuroscience research and molecular simulations. Looking ahead, the developers plan to expand AlphaEvolve's capabilities to address a broader array of external challenges. The system represents a shift toward using AI that can learn, evolve, and optimize itself to drive future breakthroughs. The development involved collaboration between Google DeepMind, Google Cloud, and Google Research teams, supported by a wide network of academic and industry partners who have applied the technology to critical problems.