OVERCLOCK.news
Live Feature
Feature ai

The AI you never see is redesigning chips, playtesting games and tuning your PC

Google DeepMind's AlphaEvolve, now open to anyone on Google Cloud, has already rewritten part of a Google chip and freed up data center capacity. It is one of several places machine learning does real work with no chatbot in sight.

By OVERCLOCK September 16, 2026 4 min read 6 sources
BlueskyXRedditMail

Most of the AI people talk about talks back. The more useful kind often doesn't. It sits inside a chip design, a data center scheduler, a pair of headphones or a game studio's test lab, doing one narrow job well, and nobody notices it at all. One of the best examples went on general release this summer: on 9 July Google made AlphaEvolve, a system from Google DeepMind that writes and improves code, available to everyone on Google Cloud.

Google DeepMind's abstract illustration for AlphaEvolve: a glowing wireframe landscape of peaks in orange, green, blue and purple rising from a dark grid.
An illustration, not a chip. Google DeepMind's artwork for AlphaEvolve, which has been generally available on Google Cloud since 9 July.Image: Google DeepMind

AlphaEvolve is not a chatbot. DeepMind calls it an evolutionary coding agent. Two Gemini models do the writing, the fast Gemini Flash to come up with lots of ideas and the more powerful Gemini Pro to go deeper, and between them they propose changes to a program. Automated evaluators then run every proposal and score it against a goal, a faster result or a smaller circuit. The best-scoring programs are kept and used to prompt the next round, and the loop goes again. It is evolution, with code in place of DNA and a test suite in place of the environment.

Google DeepMind's diagram of the AlphaEvolve loop: a scientist or engineer supplies a prompt, models, evaluation code and a starting program; a prompt sampler, an ensemble of language models, a pool of evaluators and a program database pass programs around a loop that outputs the best program.
The loop, as DeepMind draws it. Language models propose programs, evaluators score them, and a database keeps the best to seed the next round. A person supplies the goal and the test.Image: Google DeepMind

That loop has already been inside Google's own hardware. AlphaEvolve proposed a rewrite, in Verilog, the language chip designers use to describe circuits, that removed unnecessary bits from a key arithmetic circuit for matrix multiplication, the operation at the heart of AI. After checks that the modified circuit still did exactly what it should, the change was built into an upcoming Tensor Processing Unit, Google's own AI accelerator. An AI helped design part of a chip that runs AI.

It has also been quietly running Google's data centers. AlphaEvolve found a heuristic, a rule of thumb, for Borg, the system that decides which of Google's machines runs which job. DeepMind said in May 2025 that it had been in production for more than a year and recovers, on average, 0.7% of Google's worldwide computing resources. That sounds small until you remember what it is a percentage of: the same work, done on less hardware, all the time.

It has sped up the software that trains AI models as well. AlphaEvolve made the FlashAttention kernel, a heavily tuned piece of code inside Transformer models, up to 32.5% faster, in the kind of low-level GPU code that engineers normally leave to compilers. When Google opened it to everyone this summer it listed two more results: refined compaction rules in Google's Spanner database that cut write amplification by 20%, and new compiler strategies that shrank software storage footprints by nearly 9%.

What AlphaEvolve changed inside Google
Where What it changed Result Google reports
A Google TPURewrote an arithmetic circuit for matrix multiplication, in VerilogBuilt into an upcoming TPU
Borg, Google's cluster schedulerA new scheduling heuristic0.7% of worldwide compute recovered, on average
FlashAttention kernelLow-level GPU code for Transformer modelsUp to 32.5% faster
Google Spanner databaseCompaction heuristics20% less write amplification
Software storageCompiler optimisation strategiesNearly 9% smaller footprint
Small percentages, enormous scale. Every result here is Google's own report of its own systems, from DeepMind's May 2025 post and Google Cloud's July 2026 launch post.Sources: Google DeepMind, 14 May 2025; Google Cloud, 9 July 2026.

Write amplification is worth a sentence, because it affects every SSD. Databases like Spanner don't write your data just once. They keep reorganising their files in the background, and every reorganisation writes the same data again. Write amplification is the ratio of what actually gets written to disk to what you asked to store. Cutting it by a fifth means less disk traffic and less wear on the drives, across a very large number of drives.

Other companies are trying it now. During its preview, Google says, it was tested in logistics, semiconductors, genomics, high performance computing and financial services, and the chipmaker Infineon says its first experiments suggest it could transform how chips are designed. Pushmeet Kohli, Google DeepMind's vice president of science, describes the shift as AI becoming "a discovery engine that expands what we can achieve".

Your headphones may be doing something similar on a much smaller scale. Sony's WH-1000XM6 use what Sony calls an AI-based beamforming system, with six microphones, to pick out the wearer's voice from background noise on calls. Beamforming is the trick of combining several microphones: a sound reaches each one at a slightly different moment depending on where it comes from, and by lining up those timings the headphones can favour sound from the direction of your mouth and play down the rest. Sony's own description puts an AI model in that system. The noise cancelling runs on a dedicated processor, the QN3, which Sony says is seven times faster than the previous model's and handles twelve microphones.

Game studios use it to play their games for them. A modern shooter has an enormous amount to test: researchers at EA's SEED group estimated that Battlefield V has 601 features that need testing, adding up to around half a million hours, some 300 years of work, if it were all done by hand. So SEED has been building test bots with reinforcement learning, where an agent learns by trial and reward instead of following a script, and imitation learning, where it learns from how people play. One of its papers shows how much the plumbing matters too: a better path-following method for AI-driven test vehicles in Battlefield 2042 cut the average time to drive a route by 39% and left them stuck far less often.

There may even be one on your graphics card. NVIDIA's Project G-Assist is an experimental assistant that tunes an RTX PC from typed or spoken requests. It can chart frame rates, latency and temperatures, adjust GPU settings, even change keyboard lighting. It runs a Llama-based model with 8 billion parameters on the graphics card itself, so it is free and works offline, and an update last year cut its memory needs by 40%, bringing it to any RTX card with 6GB or more, laptops included. The AI's job here is translation, from plain English to the right setting.

What these have in common is a clear score. A circuit is smaller or it isn't, a scheduler wastes less compute or it doesn't, a test bot finishes the route or gets stuck. AlphaEvolve is built around exactly that: its evaluators have to be able to measure every proposal automatically. That is the quiet lesson of quiet AI. It does its best work where success can be measured by a machine, and those turn out to be places nobody thinks to look.

Sources

Every figure in this article traces to one of the following. Where a manufacturer and an independent test disagree, both are listed and the disagreement is stated in the text rather than resolved silently.

  1. Google Cloud Solve harder problems with AlphaEvolve, now available to everyone on Google Cloud 9 July 2026: general availability; testing in logistics, semiconductors, genomics, high performance computing and financial services; Spanner compaction heuristics cutting write amplification by 20%; storage footprints cut by nearly 9%; Infineon's comment on transforming the chip design lifecycle; Pushmeet Kohli's quote.
  2. Google DeepMind AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms 14 May 2025: Gemini Flash and Gemini Pro with automated evaluators and an evolutionary framework; the Verilog rewrite integrated into an upcoming TPU; the Borg heuristic in production for over a year recovering 0.7% of worldwide compute; up to 32.5% speedup of the FlashAttention kernel. Also the images in figures 1 and 2.
  3. Sony Electronics Sony Introduces the Best Noise Cancellation with the WH-1000XM6 Headphones A six-microphone AI-based beamforming system that isolates the user's voice from background noise; the QN3 processor, seven times faster than the previous model, using twelve microphones.
  4. Electronic Arts SEED SEED ML research: AAA game testing Battlefield V's 601 features and around 0.5 million hours, about 300 work years, of manual testing; reinforcement and imitation learning research for game testing; a path-following method in Battlefield 2042 cutting mean route time by 39%.
  5. NVIDIA New Lightweight AI Model for Project G-Assist Brings Support for 6GB NVIDIA GeForce RTX and RTX PRO GPUs 18 August 2025: an experimental on-device assistant for tuning RTX systems by voice and text; a new model using 40% less VRAM, supporting RTX GPUs with 6GB or more including laptops; charting frame rates, latency and temperatures and adjusting GPU and peripheral settings.
  6. NVIDIA Project G-Assist: an AI assistant for GeForce RTX AI PCs A Llama-based Instruct model with 8 billion parameters running locally on GeForce RTX hardware, free to use and able to run offline.

This is original work by OVERCLOCK, not a summary of another outlet's reporting. Specifications are as published by the manufacturer unless a measured figure is named as such; vendor performance claims are labelled with their precision and are not comparable across different numeric formats. Prices and availability change, and are accurate as of publication.

More hardware