Google DeepMind AlphaEvolve Agent Designs TPU Silicon and Optimizes Global Logistics

May 7, 2026 · Updated May 15, 2026

Google DeepMind shared the first year of impact for AlphaEvolve, a Gemini-powered coding agent for autonomous algorithmic discovery following the launch of the Gemini Enterprise Agent Platform. Unlike standard assistants, it evolves efficient code. It has already optimized internal hardware design and helped solve long-standing mathematical challenges.

DNA sequencing error reduction: 30%
Quantum circuit error reduction: 10x
Google Spanner write amplification reduction: 20%
FM Logistic routing efficiency gain: 10.4%
Schrödinger molecular simulation speedup: 4x
Availability: Private preview on Google Cloud

This shift marks the transition of agents from reactive chat tools to autonomous engineers optimizing AI foundations. By reducing write amplification (the ratio of data written to storage versus the original request) in Google Spanner by 20% and cutting quantum circuit errors by 10x, AlphaEvolve shows that agentic systems can outperform human-intensive optimization.

You can now access AlphaEvolve through a private preview on Google Cloud, where it is used by companies like Klarna to double model training speeds. In logistics, the agent improved routing efficiency by 10.4% for FM Logistic. For scientific workflows, it provides a 4x speedup in molecular simulations.

View the full update on deepmind.google

Google DeepMind

@GoogleDeepMindMay 7

Algorithms are part of nearly every aspect of life, from the physics of the natural world to planning shipping routes. Our Gemini-powered coding agent AlphaEvolve has been accelerating progress over the last year - from quantum and biotechnology to logistics and @Google’s AI infrastructure. ↓ https://t.co/CAjvAqJiod

View on X

Still wondering? A few quick answers below.

AlphaEvolve is a Gemini-powered coding agent developed by Google DeepMind for designing and optimizing advanced algorithms. Unlike general AI assistants that help write software, AlphaEvolve is an autonomous system that explores mathematical and computational spaces to discover more efficient solutions for complex scientific, engineering, and business problems across various industries.

AlphaEvolve functions as an autonomous research partner that uses evolutionary techniques to refine and improve algorithms. It can navigate high-dimensional data, test potential mathematical inequalities for counterexamples, and propose counterintuitive designs. The system is capable of learning and optimizing its own code to solve open problems in fields like quantum physics, genomics, and microeconomics.

AlphaEvolve is currently available in private preview for commercial enterprises through Google Cloud. It is being used by a variety of organizations, including financial services like Klarna, logistics firms like FM Logistic, and life sciences companies like Schrodinger. Interested businesses can engage with the Google Cloud team to explore specific algorithmic optimization applications.

AlphaEvolve has achieved significant results, including a 30% reduction in DNA sequencing errors and a 10x error reduction in quantum circuits. In commercial settings, it helped Klarna double its model training speed and enabled FM Logistic to improve routing efficiency by 10.4%, saving over 15,000 kilometers of travel annually through optimized Traveling Salesman Problem solutions.

AlphaEvolve is a core component of Googles infrastructure, used to optimize next-generation TPU chip designs and cache replacement policies. It improved Google Spanners efficiency by reducing write amplification—the ratio of data written to storage versus the original request—by 20%. It also discovered compiler optimization strategies that reduced the storage footprint of software by nearly 9%.