Top news
Investing in the public benefit OfficialAnthropic NewsOct 18Popular Musk Pledges SpaceX Shares to ‘Every Kid in America’ WebBloomberg TechnologyOct 19Popular Websites are closing their doors to AI agents WebThe Verge AIOct 17 This is the new ChatGPT Voice, powered by GPT Live OfficialOpenAI on YouTubeOct 18 How AI giants will reshape global philanthropy WebAxios AIOct 19 Gemini 4 Argon Really Surprised Me WebTwo Minute PapersOct 19 Google Gemini 4 agents just beat Minecraft with no human intervention WebEngadget AIOct 19 Who could it be? 🧐 OfficialGoogle on YouTubeOct 18 Anthropic gives outside AI watchdog “employee-like” access WebArs Technica AIOct 18 What are self-replicating ‘AI worms’, and why is Anthropic worried about them? WebThe Guardian AIOct 18
GlossaryHardwareNew to AI

Compute

Also written: Computational power, Compute Advantage, Computational demands, computing demands, processing demands, Computations, Compute allocation, Supercomputers

The processing power and specialized hardware needed to train and run AI models.

In plain terms

Compute is the brainpower for AI. It's the processing power and specialized computer chips, like GPUs, that AI models need to learn and operate.

Think of it like building a skyscraper. You need a lot of heavy machinery and skilled workers (compute) to construct it. Once built, you still need power and maintenance to keep it running. Similarly, training a large AI model requires immense compute, and running it for users also needs significant power.

The expert version

Compute refers to the computational resources, primarily specialized hardware like GPUs and TPUs, required for AI model development and operation. This includes the processing power for training large neural networks on vast datasets and the resources for running inference, where the trained model makes predictions or generates outputs.

Efficiency in compute is a major focus, as both training and inference can be extremely resource-intensive. Algorithmic improvements can reduce the compute needed, as seen in the significant decrease in training compute per model since 2012, even as models have grown larger.

Why it matters now

Frontier AI labs are heavily focused on acquiring and managing computational power. Recent news highlights efforts to build massive computing infrastructure, expand hardware, and even have AI agents manage their own compute resources. This intense focus on compute reflects its critical role in training and deploying increasingly sophisticated AI models.

Where it shows up 40 items, newest first
How each company uses it
Anthropic Claude
Anthropic sees continued AI compute spending driven by inference demand, even with safety concerns.
Google Gemini
Google Gemini is offering faster, lighter models for developers needing efficient compute.
Microsoft Copilot
Microsoft Azure provides tools to help estimate the cost of AI workloads and manage compute expenses.
OpenAI ChatGPT
OpenAI is investing in new leadership for its data center and computing units, and has paused training to focus on resource management and safety.
xAI Grok
xAI Grok is rapidly expanding its AI hardware, aiming to significantly boost its computing power.
Related terms

GPUTPUData CenterTrainingInferenceAI Infrastructure

Often confused with
Algorithm
An algorithm is a set of instructions for a computer to follow, while
Data
Data is the information AI models learn from, whereas compute is the proce
Model
A model is the AI program itself, which requires compute to be trained an
Origin

The term 'compute' in the context of AI has evolved from general computing terminology to specifically denote the processing power and hardware essential for AI development and deployment.

Try this about 5 minutes

Use ChatGPT or Google Gemini to generate a short story about a robot learning to cook. Observe how quickly the AI produces the text; this speed is a direct result of the compute power allocated to running the model for your request.

Try it: Playbook entries that use thisPlaybook →

The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.

Questions people ask

What kind of hardware is used for AI compute?
AI compute primarily uses specialized chips like Graphics Processing Units (GPUs) and Tensor Processing Units (TPUs) because they are highly efficient at performing the parallel calculations needed for AI.
Is compute the same as a computer?
Compute refers to the processing power and resources, often involving many specialized computers or chips working together, rather than a single standard computer.

First seen on Subvolts May 5, 2020. Updated Oct 17, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.