AI supercomputer
In plain wordsA huge network of AI chips — often tens or hundreds of thousands — wired together in a data center to train the largest AI models.
A tightly networked cluster of AI acceleratorsAI accelerator, n.: Any chip built or tuned to speed up AI — graphics chips, Google's TPUs, phone 'neural engines' and many startup designs. Such chips do the heavy math of training and running models., usually housed in one data centerAI data center, n.: A building or campus packed with AI chips, power equipment and cooling, built to train and run AI models. The biggest use as much electricity as a city. or campus, built and scheduled as a single machine for large training runstraining run, n.: One complete attempt at training a model with a fixed recipe, from start to finish. For the largest AI models a single run can occupy tens of thousands of chips for months. or inferenceinference, n.: Drawing a conclusion from evidence. In today's AI it most often means running an already-trained model to get an answer, as opposed to training it. at scale. Its capability depends not only on chip count and peak FLOPSFLOPS, abbr.: Floating-point operations per second: how many arithmetic steps a computer can do each second. It is the standard speed rating for supercomputers and AI chips. but on interconnectinterconnect, n.: The wiring and networking that connect chips to each other — inside a server, across a data center, or within a single chip — so they can share data and work as one. topology and bandwidth, storage, power, cooling and reliability, since one failure can stall a synchronous job (hence the attention to goodputgoodput, n.: The share of a big training job's time and computing that actually moves training forward, after subtracting time lost to crashes, restarts, waiting and repeated work. When serving users, it means the rate of requests answered fast enough.). Unlike classic supercomputers, ranked on the TOP500 by 64-bit LINPACK performance, AI supercomputers are rated in low-precision FLOPS and compared with benchmarks such as MLPerfMLPerf, prop. n.: A standard set of speed tests for AI hardware and software, where chip makers and cloud companies compare how fast their systems train and run AI models.. Examples include Microsoft's 2020 system for OpenAI (10,000 GPUs), Meta's AI Research SuperCluster (2022) and xAI's Colossus, which reached 100,000 H100sH100, prop. n.: Nvidia's flagship AI chip of 2022–2024, so scarce and sought-after during the generative-AI boom that it became a unit of measure for AI computing power. in 2024; by 2025–2026, clusters of several hundred thousand accelerators drawing a gigawatt or more were under construction.
Origin
'Supercomputer' dates to the 1960s and 1970s (CDC and Cray machines). Nvidia marketed its DGX-1 (2016) as a deep-learning supercomputer, Google built TPU pods from 2017, and 'AI supercomputer' became common as frontier training clusters grew from thousands of GPUs (2020) to hundreds of thousands (2024–2026).
Usage note
Many AI supercomputers would rank modestly, or not at all, on the TOP500, which measures 64-bit arithmetic; their headline AI FLOPS assume 8- or 4-bit precision.
Related
- AI data centerAI data center, n.: A building or campus packed with AI chips, power equipment and cooling, built to train and run AI models. The biggest use as much electricity as a city.
- rack-scale systemrack-scale system, n.: An AI computer designed as one whole cabinet of chips, cooling and wiring, rather than separate servers, so all the chips inside can work together like one big chip.
- interconnectinterconnect, n.: The wiring and networking that connect chips to each other — inside a server, across a data center, or within a single chip — so they can share data and work as one.
- distributed trainingdistributed training, n.: Training one AI model on many computer chips at once, splitting up the data or the model itself among them and keeping them in sync, so that jobs that would take a single chip thousands of years finish in weeks.
- FLOPSFLOPS, abbr.: Floating-point operations per second: how many arithmetic steps a computer can do each second. It is the standard speed rating for supercomputers and AI chips.
- goodputgoodput, n.: The share of a big training job's time and computing that actually moves training forward, after subtracting time lost to crashes, restarts, waiting and repeated work. When serving users, it means the rate of requests answered fast enough.
- StargateStargate, prop. n.: A giant venture announced in 2025 by OpenAI, SoftBank and Oracle to build data centers in the United States for training and running AI, with a pledged $500 billion.
- computecompute, n.: The raw computing power used to train and run AI, usually measured by how many specialized chips are used and for how long. With data and algorithms, it is a main driver of AI progress.
Mentioned in
AI data centerAI data center, n.: A building or campus packed with AI chips, power equipment and cooling, built to train and run AI models. The biggest use as much electricity as a city., StargateStargate, prop. n.: A giant venture announced in 2025 by OpenAI, SoftBank and Oracle to build data centers in the United States for training and running AI, with a pledged $500 billion., xAIxAI, prop. n.: An artificial intelligence company founded by Elon Musk in 2023 that makes the Grok chatbot. It is now part of Musk's rocket company, SpaceX., AI factoryAI factory, n.: A data center built to turn electricity and data into AI models and AI answers at industrial scale; in Europe, also a public supercomputing hub that helps companies and researchers build AI., distributed trainingdistributed training, n.: Training one AI model on many computer chips at once, splitting up the data or the model itself among them and keeping them in sync, so that jobs that would take a single chip thousands of years finish in weeks., FLOPSFLOPS, abbr.: Floating-point operations per second: how many arithmetic steps a computer can do each second. It is the standard speed rating for supercomputers and AI chips., sovereign AIsovereign AI, n.: A country's ability to build and run AI on its own terms, so that the computers, models and know-how it relies on cannot be withheld or dictated by foreign companies or governments., TPUTPU, abbr.: Google's own family of chips built specifically for AI, used to train and run its Gemini models and rented to other companies through its cloud., Genesis MissionGenesis Mission, prop. n.: A US government program launched in 2025 to speed up science with AI, using the national laboratories' supercomputers and data to build AI tools for research., InfiniBandInfiniBand, prop. n.: A very fast networking technology used to connect the thousands of servers in supercomputers and AI training clusters, designed so data moves with minimal delay., MLPerfMLPerf, prop. n.: A standard set of speed tests for AI hardware and software, where chip makers and cloud companies compare how fast their systems train and run AI models., National AI Research ResourceNational AI Research Resource, prop. n.: A US program to give university researchers, students and small companies shared access to the costly computers, data and AI models that are otherwise mostly available to big technology firms., rack-scale systemrack-scale system, n.: An AI computer designed as one whole cabinet of chips, cooling and wiring, rather than separate servers, so all the chips inside can work together like one big chip., scale-up and scale-outscale-up and scale-out, phr.: Two ways to get more computing power: make each machine bigger and more tightly linked (scale up), or add more machines connected by a network (scale out)., stragglerstraggler, n.: One slow computer or chip in a large group working in lockstep. Because everyone must wait for the slowest member at each step, a single straggler can slow down an entire AI training job.