Google makes AlphaEvolve AI optimization agent widely available

July 10, 2026

Google launched AlphaEvolve, a code-optimization agent built with DeepMind, to all Google Cloud customers on Wednesday.cloud.google+1
Early adopters reported gains including BASF improving forecasting models by over 80% and FM Logistic cutting warehouse routing distances.cloud.google
The release sharpens competition with Microsoft and Amazon Web Services, which are expanding their own agentic AI platforms.investing

Google Alphabet Inc. on Wednesday made AlphaEvolve, its Gemini-powered code-optimization agent, generally available to all Google Cloud customers through the Gemini Enterprise Agent Platform. The tool, developed in partnership with Google DeepMind, is designed to help organizations tackle complex computational and engineering challenges by autonomously exploring vast algorithmic search spaces to find optimized solutions.cloud.google+1

The broad rollout marks a transition from private preview, where AlphaEvolve was first introduced last year, to full general availability. The system has been tested across domains including logistics, semiconductors, genomics, high-performance computing, and financial services during its early access period.cloud.google+2

How It Works

AlphaEvolve follows a four-step process: users provide a baseline algorithm and problem definition, establish a scoring function to evaluate candidate solutions, run the optimization agent, and then deploy the resulting code into production workloads. The system requires only two core inputs — a seed program and a deterministic evaluator that scores mutated candidates.cloud.google

"AI is moving beyond acting as a productivity assistant that accelerates how we work to a discovery engine that expands what we can achieve," said Pushmeet Kohli, Chief Scientist at Google Cloud and Vice President of Science at Google DeepMind.cloud.google

Early Adopters Report Gains

Several organizations shared results from AlphaEvolve's early access program. BASF said it used the tool to improve planning and forecasting models by over 80 percent after previous attempts at building a supply chain digital twin had failed. FM Logistic reported a 10.4 percent improvement in warehouse routing, saving over 15,000 kilometers in staff travel. Klarna said it doubled throughput in one of its largest machine learning training pipelines while improving model quality.cloud.google

Oak Ridge National Laboratory deployed AlphaEvolve on Frontier, the world's first exascale supercomputer, to optimize GPU kernels for scientific computing.cloud.google

Internal Use and Competition

Google itself has used AlphaEvolve to optimize next-generation Tensor Processing Unit designs, reduce write amplification in its Spanner database by 20 percent, and discover quantum circuits with 10 times lower error rates for molecular simulations on its Willow quantum processor.cloud.google

The launch intensifies competition in the enterprise AI platform market, where rivals including Microsoft and Amazon Amazon.com, Inc. Web Services are also expanding agentic AI offerings. AlphaEvolve is accessible through the Gemini Enterprise Agent Platform, which Google introduced at Cloud Next 2026 in April as an evolution of its Vertex AI platform.siliconangle+1

Sources (14)