# EveryNeocloud > Every neocloud and AI compute provider in the world. A free guide to 259 providers, filed by kind (neocloud, inference, sovereign, hyperscaler, developer, marketplace, miner, chipmaker, supercomputer, research); chip (nvidia-blackwell, nvidia-hopper, nvidia-ampere, nvidia-other, amd-instinct, tpu, trainium, gaudi, wafer, lpu, custom); offer (bare-metal, vm, clusters, kubernetes, serverless, api, colocation, marketplace, research); ownership (startup, listed, state, subsidiary, public, acquired, closed); by GPU fleet on a 0 to 5 scale (0 boutique, under 1,000 GPUs; 1 growing, 1,000 to 10,000; 2 large, 10,000 to 50,000; 3 very large, 50,000 to 200,000; 4 frontier, 200,000 to 1 million; 5 million-GPU class, 1 million and more); by backers and partners, region, country and date. With hubs, a calendar, a timeline, a world map and a glossary. Published by ClearTrust. Each page gives: name, founded, headquarters, founders, gpus, data centres, power, funding, listed, customers, people, performance, formerly, what it is, how it came about, the compute, where the gpus are, customers and deals, money and ownership, its place in ai, questions and criticism, a short FAQ, further reading and source links. Pages are derived from Wikipedia and Wikidata with AI assistance and may contain mistakes; cite the page URL. ## Browse - [All providers](https://everyneocloud.cpn.cleartrust.cc/providers) - [Timeline](https://everyneocloud.cpn.cleartrust.cc/timeline) - [AI compute hubs](https://everyneocloud.cpn.cleartrust.cc/hubs) - [Regions](https://everyneocloud.cpn.cleartrust.cc/regions) - [Countries](https://everyneocloud.cpn.cleartrust.cc/countries) - [World map](https://everyneocloud.cpn.cleartrust.cc/map) - [Neoclouds](https://everyneocloud.cpn.cleartrust.cc/kinds/neocloud) - [Inference platforms](https://everyneocloud.cpn.cleartrust.cc/kinds/inference) - [Sovereign and national AI clouds](https://everyneocloud.cpn.cleartrust.cc/kinds/sovereign) - [Hyperscale clouds](https://everyneocloud.cpn.cleartrust.cc/kinds/hyperscaler) - [Developer clouds](https://everyneocloud.cpn.cleartrust.cc/kinds/developer) - [GPU marketplaces](https://everyneocloud.cpn.cleartrust.cc/kinds/marketplace) - [Miners turned AI hosts](https://everyneocloud.cpn.cleartrust.cc/kinds/miner) - [Chipmaker clouds](https://everyneocloud.cpn.cleartrust.cc/kinds/chipmaker) - [AI supercomputers](https://everyneocloud.cpn.cleartrust.cc/kinds/supercomputer) - [Research and university clusters](https://everyneocloud.cpn.cleartrust.cc/kinds/research) - [Nvidia Blackwell](https://everyneocloud.cpn.cleartrust.cc/chips/nvidia-blackwell) - [Nvidia Hopper](https://everyneocloud.cpn.cleartrust.cc/chips/nvidia-hopper) - [Nvidia Ampere](https://everyneocloud.cpn.cleartrust.cc/chips/nvidia-ampere) - [Other Nvidia GPUs](https://everyneocloud.cpn.cleartrust.cc/chips/nvidia-other) - [AMD Instinct](https://everyneocloud.cpn.cleartrust.cc/chips/amd-instinct) - [Google TPU](https://everyneocloud.cpn.cleartrust.cc/chips/tpu) - [AWS Trainium and Inferentia](https://everyneocloud.cpn.cleartrust.cc/chips/trainium) - [Intel Gaudi](https://everyneocloud.cpn.cleartrust.cc/chips/gaudi) - [Cerebras wafer-scale](https://everyneocloud.cpn.cleartrust.cc/chips/wafer) - [Groq LPU](https://everyneocloud.cpn.cleartrust.cc/chips/lpu) - [Other and custom chips](https://everyneocloud.cpn.cleartrust.cc/chips/custom) - [Bare-metal GPUs](https://everyneocloud.cpn.cleartrust.cc/offers/bare-metal) - [GPU virtual machines](https://everyneocloud.cpn.cleartrust.cc/offers/vm) - [Training clusters](https://everyneocloud.cpn.cleartrust.cc/offers/clusters) - [Managed Kubernetes](https://everyneocloud.cpn.cleartrust.cc/offers/kubernetes) - [Serverless GPUs](https://everyneocloud.cpn.cleartrust.cc/offers/serverless) - [Model APIs](https://everyneocloud.cpn.cleartrust.cc/offers/api) - [Colocation and hosting](https://everyneocloud.cpn.cleartrust.cc/offers/colocation) - [Marketplace](https://everyneocloud.cpn.cleartrust.cc/offers/marketplace) - [Research allocation](https://everyneocloud.cpn.cleartrust.cc/offers/research) - [Private, venture-backed](https://everyneocloud.cpn.cleartrust.cc/ownership/startup) - [Listed companies](https://everyneocloud.cpn.cleartrust.cc/ownership/listed) - [State-owned or state-backed](https://everyneocloud.cpn.cleartrust.cc/ownership/state) - [Part of a larger group](https://everyneocloud.cpn.cleartrust.cc/ownership/subsidiary) - [Public research bodies](https://everyneocloud.cpn.cleartrust.cc/ownership/public) - [Acquired or merged](https://everyneocloud.cpn.cleartrust.cc/ownership/acquired) - [Closed](https://everyneocloud.cpn.cleartrust.cc/ownership/closed) ## Live - [Today](https://everyneocloud.cpn.cleartrust.cc/today): on this day, provider of the day - [Calendar](https://everyneocloud.cpn.cleartrust.cc/calendar) - [News](https://everyneocloud.cpn.cleartrust.cc/news) ## Learn - [Concepts](https://everyneocloud.cpn.cleartrust.cc/concepts) ## For agents - [MCP server](https://everyneocloud.cpn.cleartrust.cc/mcp): Streamable HTTP, no key needed - [JSON API](https://everyneocloud.cpn.cleartrust.cc/developers): GET https://everyneocloud.cpn.cleartrust.cc/api/v1/ - [Sitemap](https://everyneocloud.cpn.cleartrust.cc/sitemap.xml) - [How complete the catalogue is](https://everyneocloud.cpn.cleartrust.cc/coverage) ## AI compute hubs - [Northern Virginia](https://everyneocloud.cpn.cleartrust.cc/hub/northern-virginia): Loudoun and Prince William counties: the largest data centre market in the world and the home region of the big clouds. - [Abilene (Stargate)](https://everyneocloud.cpn.cleartrust.cc/hub/abilene): West Texas town that hosts the first Stargate campus, built by Crusoe and leased to Oracle for OpenAI. - [Memphis (Colossus)](https://everyneocloud.cpn.cleartrust.cc/hub/memphis): Home of Colossus, the AI training cluster built for xAI in a former Electrolux factory, and its larger successors across the state line. - [Dallas-Fort Worth](https://everyneocloud.cpn.cleartrust.cc/hub/dallas-fort-worth): One of the busiest data centre markets in the US, with GPU halls in Denton, Plano and the suburbs, and the head office of Applied Digital. - [Plano](https://everyneocloud.cpn.cleartrust.cc/hub/plano): Dallas suburb where CoreWeave opened its first Texas data centre, a $1.6 billion build announced in 2023. - [Atlanta](https://everyneocloud.cpn.cleartrust.cc/hub/atlanta): Fast-growing data centre market and site of Microsoft's second Fairwater AI datacenter, linked to Wisconsin as one "AI superfactory". - [Ellendale (Polaris Forge)](https://everyneocloud.cpn.cleartrust.cc/hub/ellendale): Prairie town where Applied Digital builds Polaris Forge 1, an AI campus with 400 MW leased to CoreWeave. - [Childress (IREN)](https://everyneocloud.cpn.cleartrust.cc/hub/childress): Texas Panhandle campus where IREN turned a bitcoin mining site into AI capacity for Microsoft. - [Sweetwater (IREN)](https://everyneocloud.cpn.cleartrust.cc/hub/sweetwater): West Texas wind town where IREN is building a 2 GW hub for AI compute. - [Lake Mariner (TeraWulf)](https://everyneocloud.cpn.cleartrust.cc/hub/lake-mariner): Former power plant site on Lake Ontario where TeraWulf hosts AI compute for Fluidstack, backed by Google. - [New York and New Jersey](https://everyneocloud.cpn.cleartrust.cc/hub/new-york-new-jersey): Where CoreWeave began as a crypto miner and keeps its head office, near Wall Street and the finance that funds the neoclouds. - [San Francisco Bay Area](https://everyneocloud.cpn.cleartrust.cc/hub/san-francisco-bay-area): Where the AI labs, the chip start-ups and many neoclouds are run from, and home to the El Capitan and Perlmutter supercomputers. - [Seattle](https://everyneocloud.cpn.cleartrust.cc/hub/seattle): Home of Amazon Web Services and Microsoft Azure, the two largest clouds, which buy more GPUs than any neocloud. - [Denver and the Front Range](https://everyneocloud.cpn.cleartrust.cc/hub/denver): Where Crusoe began by turning wasted oil-field gas into computing power, and near the NCAR supercomputer in Cheyenne. - [Richland Parish (Hyperion)](https://everyneocloud.cpn.cleartrust.cc/hub/richland-parish): Rural Louisiana parish where Meta is building Hyperion, an AI training data centre planned at gigawatt scale. - [Oak Ridge](https://everyneocloud.cpn.cleartrust.cc/hub/oak-ridge): US Department of Energy lab that ran Summit and runs Frontier, the world's first exascale supercomputer. - [Montreal](https://everyneocloud.cpn.cleartrust.cc/hub/montreal): A centre of deep learning research, home of Mila and of Bell Canada, which is building a national AI compute network. - [Quebec](https://everyneocloud.cpn.cleartrust.cc/hub/quebec): Hydro-powered province drawing AI data centres and miners, including Telus's sovereign AI factory in Rimouski. - [Iceland](https://everyneocloud.cpn.cleartrust.cc/hub/iceland): Geothermal and hydro power and a cool climate make Iceland a home for GPU clusters from Nscale, Crusoe and Nebius. - [Narvik (Stargate Norway)](https://everyneocloud.cpn.cleartrust.cc/hub/narvik): Arctic Norwegian town where Nscale and Aker are building Stargate Norway, OpenAI's first European data centre project. - [Mäntsälä (Nebius)](https://everyneocloud.cpn.cleartrust.cc/hub/mantsala): Finnish town where Nebius runs its first owned AI data centre, built by Yandex and expanded for GPUs. - [Amsterdam](https://everyneocloud.cpn.cleartrust.cc/hub/amsterdam): Head office of Nebius and Leaseweb, and one of Europe's main internet exchange and data centre cities. - [London and Loughton](https://everyneocloud.cpn.cleartrust.cc/hub/london): Head office of Nscale and Fluidstack, with GPU halls in the Docklands and Slough and a planned AI campus at Loughton. - [Bristol (Isambard-AI)](https://everyneocloud.cpn.cleartrust.cc/hub/bristol): Home of Isambard-AI, the UK's national AI supercomputer, run by the University of Bristol. - [Jülich (JUPITER)](https://everyneocloud.cpn.cleartrust.cc/hub/julich): German research centre that hosts JUPITER, Europe's first exascale supercomputer. - [Paris](https://everyneocloud.cpn.cleartrust.cc/hub/paris): Home of Mistral AI, Scaleway and the Jean Zay supercomputer, and the centre of France's push for sovereign AI compute. - [Barcelona (MareNostrum)](https://everyneocloud.cpn.cleartrust.cc/hub/barcelona): Home of the Barcelona Supercomputing Center and MareNostrum 5, a EuroHPC machine with thousands of Nvidia H100s. - [Bologna (Leonardo)](https://everyneocloud.cpn.cleartrust.cc/hub/bologna): Italy's supercomputing centre, CINECA, runs Leonardo, a EuroHPC machine with close to 14,000 Nvidia A100s. - [Lugano (Alps)](https://everyneocloud.cpn.cleartrust.cc/hub/lugano): Seat of the Swiss National Supercomputing Centre and Alps, a machine with 10,752 Nvidia GH200 superchips. - [Kajaani (LUMI)](https://everyneocloud.cpn.cleartrust.cc/hub/kajaani): Former paper mill town in Finland that hosts LUMI, a EuroHPC supercomputer powered by hydro and wind. - [Navi Mumbai](https://everyneocloud.cpn.cleartrust.cc/hub/navi-mumbai): India's largest data centre market and home of Yotta's NM1, where India's first large Nvidia H100 cluster arrived in 2024. - [Delhi NCR](https://everyneocloud.cpn.cleartrust.cc/hub/delhi-ncr): National Capital Region, home of E2E Networks and of Yotta's Greater Noida campus, with GPU clusters for IndiaAI. - [Bengaluru](https://everyneocloud.cpn.cleartrust.cc/hub/bengaluru): India's technology capital, home of its AI start-ups, the IISc PARAM Pravega machine and the buyers of Indian GPU time. - [Pune (C-DAC)](https://everyneocloud.cpn.cleartrust.cc/hub/pune): Home of C-DAC, maker of the PARAM supercomputers, and of AIRAWAT, India's national AI research machine. - [Hyderabad](https://everyneocloud.cpn.cleartrust.cc/hub/hyderabad): Fast-growing Indian cloud market with regions from AWS and Microsoft and the head office of CtrlS. - [Abu Dhabi (Stargate UAE)](https://everyneocloud.cpn.cleartrust.cc/hub/abu-dhabi): Home of G42 and of the 5 GW UAE-US AI Campus, where Stargate UAE is being built for OpenAI and Oracle. - [Riyadh (Humain)](https://everyneocloud.cpn.cleartrust.cc/hub/riyadh): Saudi capital and base of Humain, the national AI company set up by the Public Investment Fund in 2025. - [Singapore](https://everyneocloud.cpn.cleartrust.cc/hub/singapore): Southeast Asia's cloud capital, with tight limits on new data centres and the head offices of Bitdeer and Singtel's AI cloud. - [Johor](https://everyneocloud.cpn.cleartrust.cc/hub/johor): Malaysian state across from Singapore that became Southeast Asia's fastest-growing data centre market, including YTL's Nvidia AI cloud. - [Tokyo](https://everyneocloud.cpn.cleartrust.cc/hub/tokyo): Home of SoftBank, of the ABCI 3.0 national AI cloud at Kashiwa and of the TSUBAME supercomputers. - [Kobe (Fugaku)](https://everyneocloud.cpn.cleartrust.cc/hub/kobe): RIKEN's computing centre on Port Island, home of the K computer and Fugaku, the world's fastest machine from 2020 to 2022. - [Seoul](https://everyneocloud.cpn.cleartrust.cc/hub/seoul): Capital of a country that secured 260,000 Nvidia GPUs in 2025 for Naver, SK, Samsung, Hyundai and the government. - [Taipei and Hsinchu](https://everyneocloud.cpn.cleartrust.cc/hub/taipei): Where the servers for the world's AI clusters are designed, and home of Taiwan's Taiwania national supercomputers. - [Beijing](https://everyneocloud.cpn.cleartrust.cc/hub/beijing): Home of Baidu AI Cloud and many of China's AI labs, which run on a mix of Nvidia chips and domestic accelerators. - [Hangzhou](https://everyneocloud.cpn.cleartrust.cc/hub/hangzhou): Home of Alibaba Cloud, China's largest cloud, and of DeepSeek, whose models made efficient training a global story. - [Sydney](https://everyneocloud.cpn.cleartrust.cc/hub/sydney): Head office of IREN and AirTrunk and Australia's largest data centre market. - [Melbourne](https://everyneocloud.cpn.cleartrust.cc/hub/melbourne): Site of the first Firmus AI factory under Project Southgate, with thousands of Nvidia GB300 GPUs at a CDC campus. - [Canberra](https://everyneocloud.cpn.cleartrust.cc/hub/canberra): Home of Australia's National Computational Infrastructure and its Gadi supercomputer at the Australian National University. ## Calendar - [Nvidia GTC San Jose](https://everyneocloud.cpn.cleartrust.cc/events/nvidia-gtc-san-jose): Nvidia's main yearly conference, where its chief executive sets out the next GPUs, racks and networks that AI clouds will buy. - [Nvidia GTC Washington, D.C.](https://everyneocloud.cpn.cleartrust.cc/events/nvidia-gtc-dc): Nvidia's policy-facing event in the US capital, on public-sector AI, national labs and American AI infrastructure. - [Nvidia GTC Paris at VivaTech](https://everyneocloud.cpn.cleartrust.cc/events/nvidia-gtc-paris): Nvidia's European keynote, held inside the VivaTech fair, where Europe's sovereign AI clouds and neoclouds make their announcements. - [GTC Taipei at Computex](https://everyneocloud.cpn.cleartrust.cc/events/gtc-taipei-computex): Asia's largest computer trade show, where Taiwan's server makers and Nvidia show the racks that fill AI data centres. - [AWS re:Invent](https://everyneocloud.cpn.cleartrust.cc/events/aws-reinvent): Amazon Web Services' yearly conference, where it launches new instances, Trainium chips and AI services. - [Microsoft Ignite](https://everyneocloud.cpn.cleartrust.cc/events/microsoft-ignite): Microsoft's yearly conference for IT and cloud customers, where Azure sets out its AI infrastructure and chips. - [Microsoft Build](https://everyneocloud.cpn.cleartrust.cc/events/microsoft-build): Microsoft's developer conference, with news on Azure AI services, model hosting and the infrastructure behind them. - [Google Cloud Next](https://everyneocloud.cpn.cleartrust.cc/events/google-cloud-next): Google Cloud's yearly conference, where it announces TPU generations, GPU instances and AI services. - [Oracle AI World](https://everyneocloud.cpn.cleartrust.cc/events/oracle-ai-world): Oracle's yearly customer conference, renamed from CloudWorld in 2025, with news on its AI clusters and Stargate sites. - [NeurIPS](https://everyneocloud.cpn.cleartrust.cc/events/neurips): The largest research conference in machine learning, where much of the work that drives demand for compute is presented. - [ICML](https://everyneocloud.cpn.cleartrust.cc/events/icml): The International Conference on Machine Learning, one of the field's two largest research meetings, held in South America in 2027. - [CVPR](https://everyneocloud.cpn.cleartrust.cc/events/cvpr): The leading conference on computer vision, where image and video models, and the compute they need, are presented. - [ISC High Performance](https://everyneocloud.cpn.cleartrust.cc/events/isc-high-performance): Europe's main supercomputing conference, held in Hamburg, where the June TOP500 list is released. - [TOP500 list, June edition](https://everyneocloud.cpn.cleartrust.cc/events/top500-june): The first of the year's two rankings of the world's fastest supercomputers, released at the opening of ISC. - [SC, the International Conference for High Performance Computing](https://everyneocloud.cpn.cleartrust.cc/events/sc-conference): The world's largest supercomputing conference and exhibition, where the November TOP500 list is released. - [TOP500 list, November edition](https://everyneocloud.cpn.cleartrust.cc/events/top500-november): The second of the year's two supercomputer rankings, released on the Monday of the SC conference. - [MLPerf Inference results](https://everyneocloud.cpn.cleartrust.cc/events/mlperf-inference): MLCommons' spring round of AI inference benchmarks, comparing how fast chips and clouds run standard models. - [MLPerf Training results](https://everyneocloud.cpn.cleartrust.cc/events/mlperf-training): MLCommons' round of AI training benchmarks, where clouds show how fast thousands of GPUs train standard models. - [AI Infra Summit](https://everyneocloud.cpn.cleartrust.cc/events/ai-infra-summit): A Silicon Valley conference on the full stack of AI infrastructure, from chips and networks to clouds and data centres. - [OCP Global Summit](https://everyneocloud.cpn.cleartrust.cc/events/ocp-global-summit): The Open Compute Project's yearly meeting, where hyperscalers share open designs for AI racks, power and cooling. - [Hot Chips](https://everyneocloud.cpn.cleartrust.cc/events/hot-chips): The yearly symposium at Stanford where chip designers present the inner workings of new AI processors. - [Bengaluru Tech Summit](https://everyneocloud.cpn.cleartrust.cc/events/bengaluru-tech-summit): Karnataka's yearly technology summit and exhibition, a meeting point for India's AI start-ups, chip designers and clouds. - [World Artificial Intelligence Conference](https://everyneocloud.cpn.cleartrust.cc/events/world-ai-conference): China's largest AI conference and exhibition, held in Shanghai each summer, where Chinese clouds and chipmakers show their compute. - [LEAP](https://everyneocloud.cpn.cleartrust.cc/events/leap-riyadh): Saudi Arabia's large technology conference, where Gulf AI compute plans and data centre deals are announced. - [India AI Impact Summit anniversary](https://everyneocloud.cpn.cleartrust.cc/events/india-ai-impact-summit): On 16 February 2026 the India AI Impact Summit opened in New Delhi, the first summit in the series held in the Global South. - [Amazon EC2 launch anniversary](https://everyneocloud.cpn.cleartrust.cc/events/ec2-launch): On 25 August 2006 Amazon opened a public beta of EC2, the service that made renting computers by the hour normal. - [Windows Azure availability anniversary](https://everyneocloud.cpn.cleartrust.cc/events/azure-launch): On 1 February 2010 Microsoft made its Windows Azure cloud platform commercially available. - [Google App Engine anniversary](https://everyneocloud.cpn.cleartrust.cc/events/app-engine-launch): On 7 April 2008 Google released App Engine in preview, its first step into selling cloud computing. - [ChatGPT launch anniversary](https://everyneocloud.cpn.cleartrust.cc/events/chatgpt-launch): OpenAI released ChatGPT on 30 November 2022, setting off the rush for GPUs that created today's neocloud industry. - [Stargate announcement anniversary](https://everyneocloud.cpn.cleartrust.cc/events/stargate-announced): On 21 January 2025 OpenAI, SoftBank, Oracle and MGX announced Stargate, a plan for up to 500 billion US dollars of AI infrastructure. - [First exascale supercomputer anniversary](https://everyneocloud.cpn.cleartrust.cc/events/frontier-exascale): On 30 May 2022 Frontier, at Oak Ridge National Laboratory, was named the world's first exascale supercomputer. - [Fugaku takes first place anniversary](https://everyneocloud.cpn.cleartrust.cc/events/fugaku-number-one): On 22 June 2020 Japan's Fugaku became the world's fastest supercomputer, the first Arm-based machine to top the TOP500. - [El Capitan takes first place anniversary](https://everyneocloud.cpn.cleartrust.cc/events/el-capitan-number-one): On 18 November 2024 El Capitan, at Lawrence Livermore National Laboratory, was named the world's fastest supercomputer. - [JUPITER inauguration anniversary](https://everyneocloud.cpn.cleartrust.cc/events/jupiter-inauguration): On 5 September 2025 JUPITER, Europe's first exascale supercomputer, went into operation at Forschungszentrum Jülich. - [Alps inauguration anniversary](https://everyneocloud.cpn.cleartrust.cc/events/alps-inauguration): On 14 September 2024 Switzerland inaugurated Alps, its national supercomputer for science and AI, at CSCS in Lugano. - [CoreWeave listing anniversary](https://everyneocloud.cpn.cleartrust.cc/events/coreweave-listing): On 28 March 2025 CoreWeave began trading on the Nasdaq, one of the first neoclouds to list on a US stock exchange. ## Concepts - **Neocloud**: A cloud company built mainly to rent out GPUs for AI, rather than to offer the full range of services a general cloud sells. (https://everyneocloud.cpn.cleartrust.cc/concepts/neocloud) - **Hyperscaler**: One of the few very large general cloud companies, such as AWS, Microsoft Azure and Google Cloud, that run data centres across the world. (https://everyneocloud.cpn.cleartrust.cc/concepts/hyperscaler) - **AI factory**: A term popularised by Nvidia for a data centre built to produce AI output, measured in tokens, at industrial scale. (https://everyneocloud.cpn.cleartrust.cc/concepts/ai-factory) - **Sovereign AI**: AI compute, data and models kept under a country's own control, inside its borders and under its laws. (https://everyneocloud.cpn.cleartrust.cc/concepts/sovereign-ai) - **Data residency**: A rule or promise that data is stored and processed inside a particular country or region. (https://everyneocloud.cpn.cleartrust.cc/concepts/data-residency) - **GPU marketplace**: A platform that lists GPUs from many owners and matches buyers to them, usually by price. (https://everyneocloud.cpn.cleartrust.cc/concepts/gpu-marketplace) - **Decentralised compute**: A network that pools GPUs owned by many independent operators, often coordinated and paid for with a cryptocurrency token. (https://everyneocloud.cpn.cleartrust.cc/concepts/decentralised-compute) - **Chipmaker cloud**: A cloud run by a chip company to sell access to its own accelerators, rather than selling the chips alone. (https://everyneocloud.cpn.cleartrust.cc/concepts/chipmaker-cloud) - **HPC hosting**: Renting space, power and cooling for high-performance computing, a business many former bitcoin miners moved into. (https://everyneocloud.cpn.cleartrust.cc/concepts/hpc-hosting) - **Stranded power**: Electricity capacity that exists at a site but has no nearby demand, such as a remote wind farm or a gas field. (https://everyneocloud.cpn.cleartrust.cc/concepts/stranded-power) - **Compute deal**: A contract in which a customer commits to buy a set amount of AI compute over several years. (https://everyneocloud.cpn.cleartrust.cc/concepts/compute-deal) - **Anchor tenant**: The large customer whose long contract makes a new cluster or data centre possible to finance and build. (https://everyneocloud.cpn.cleartrust.cc/concepts/anchor-tenant) - **Take-or-pay contract**: A contract in which the buyer pays for the agreed capacity whether or not it actually uses it. (https://everyneocloud.cpn.cleartrust.cc/concepts/take-or-pay) - **GPU-backed debt**: Loans secured on GPUs and on the contracts to rent them out, used by neoclouds to buy hardware at scale. (https://everyneocloud.cpn.cleartrust.cc/concepts/gpu-backed-debt) - **Sale-leaseback**: Selling an asset, such as a data centre or a fleet of servers, to an investor and renting it back at once. (https://everyneocloud.cpn.cleartrust.cc/concepts/sale-leaseback) - **Depreciation**: The accounting charge that spreads the cost of hardware over the years it is expected to stay useful. (https://everyneocloud.cpn.cleartrust.cc/concepts/depreciation) - **Capital expenditure**: Money spent on long-lasting assets such as data centres, power equipment and servers, rather than on running costs. (https://everyneocloud.cpn.cleartrust.cc/concepts/capex) - **GPU**: A chip with thousands of small cores working in parallel, first built for graphics and now the main engine of AI. (https://everyneocloud.cpn.cleartrust.cc/concepts/gpu) - **Accelerator**: Any chip built to speed up a particular kind of work, such as AI, beyond what an ordinary processor can do. (https://everyneocloud.cpn.cleartrust.cc/concepts/accelerator) - **TPU**: Google's own AI chip, used in its data centres since 2015 and rented to others through Google Cloud. (https://everyneocloud.cpn.cleartrust.cc/concepts/tpu) - **Custom accelerator**: An AI chip designed by a cloud or internet company for its own data centres, such as Trainium, Maia or MTIA. (https://everyneocloud.cpn.cleartrust.cc/concepts/custom-accelerator) - **Trainium**: Amazon Web Services' own AI chips: Trainium for training models and Inferentia for running them. (https://everyneocloud.cpn.cleartrust.cc/concepts/trainium) - **Wafer-scale engine**: Cerebras's design that turns a whole silicon wafer into one giant chip instead of cutting it into many small ones. (https://everyneocloud.cpn.cleartrust.cc/concepts/wafer-scale) - **LPU**: Groq's chip for running language models, which keeps model weights in fast on-chip memory to generate tokens quickly. (https://everyneocloud.cpn.cleartrust.cc/concepts/lpu) - **Tensor core**: A specialised unit inside a GPU that multiplies small blocks of numbers in one step, doing most of the work in AI. (https://everyneocloud.cpn.cleartrust.cc/concepts/tensor-core) - **HBM**: Memory chips stacked on top of one another and placed right next to the processor, giving AI chips their huge memory speed. (https://everyneocloud.cpn.cleartrust.cc/concepts/hbm) - **Memory bandwidth**: How many bytes per second a chip can read from its memory, often the true limit on how fast a model answers. (https://everyneocloud.cpn.cleartrust.cc/concepts/memory-bandwidth) - **SRAM**: Very fast memory built into the processor itself, much quicker than HBM but holding far less. (https://everyneocloud.cpn.cleartrust.cc/concepts/sram) - **Superchip**: Nvidia's name for a module that joins its Grace processor and one or two GPUs with a fast link on one board. (https://everyneocloud.cpn.cleartrust.cc/concepts/superchip) - **CUDA**: Nvidia's software platform for programming its GPUs, released in 2007 and a key reason most AI runs on Nvidia. (https://everyneocloud.cpn.cleartrust.cc/concepts/cuda) - **ROCm**: AMD's open-source software platform for programming its Instinct accelerators, its answer to CUDA. (https://everyneocloud.cpn.cleartrust.cc/concepts/rocm) - **Node**: One server in a cluster, in AI usually a box holding eight GPUs, two processors, memory and network cards. (https://everyneocloud.cpn.cleartrust.cc/concepts/node) - **DGX and HGX**: Nvidia's complete AI servers (DGX) and the eight-GPU baseboard it supplies to other server makers (HGX). (https://everyneocloud.cpn.cleartrust.cc/concepts/dgx) - **Rack**: The standard steel frame, about two metres tall, that holds servers, switches and power units in a data centre. (https://everyneocloud.cpn.cleartrust.cc/concepts/rack) - **Rack density**: How much electrical power a single rack draws, the number that most shapes an AI data centre's design. (https://everyneocloud.cpn.cleartrust.cc/concepts/rack-density) - **NVLink**: Nvidia's very fast link that lets GPUs share data with each other directly, much faster than through the network. (https://everyneocloud.cpn.cleartrust.cc/concepts/nvlink) - **NVSwitch**: A switch chip that connects many GPUs over NVLink so that every GPU can talk to every other at full speed. (https://everyneocloud.cpn.cleartrust.cc/concepts/nvswitch) - **NVL72**: Nvidia's rack-scale system that joins 72 Blackwell GPUs and 36 Grace CPUs over NVLink so they act as one. (https://everyneocloud.cpn.cleartrust.cc/concepts/nvl72) - **Scale-up network**: The fastest network tier, joining a small group of GPUs, such as a server or rack, so tightly that they work as one. (https://everyneocloud.cpn.cleartrust.cc/concepts/scale-up) - **Scale-out network**: The network that joins thousands of servers or racks into one cluster, usually InfiniBand or Ethernet. (https://everyneocloud.cpn.cleartrust.cc/concepts/scale-out) - **InfiniBand**: A high-speed, low-delay network standard long used in supercomputers and now in many AI clusters. (https://everyneocloud.cpn.cleartrust.cc/concepts/infiniband) - **RDMA**: A way for one computer to read or write another's memory directly over the network, without involving its processor. (https://everyneocloud.cpn.cleartrust.cc/concepts/rdma) - **RoCE**: A way to run RDMA, the fast memory-to-memory transfer used in AI clusters, over ordinary Ethernet networks. (https://everyneocloud.cpn.cleartrust.cc/concepts/roce) - **Ethernet fabric**: An AI cluster network built from Ethernet switches, tuned to carry GPU traffic with little loss and delay. (https://everyneocloud.cpn.cleartrust.cc/concepts/ethernet-fabric) - **Ultra Ethernet**: An industry effort, begun in 2023, to update Ethernet for AI and high-performance computing. (https://everyneocloud.cpn.cleartrust.cc/concepts/ultra-ethernet) - **Spine-leaf**: A network layout where every server switch connects to every core switch, so any two servers are only a few hops apart. (https://everyneocloud.cpn.cleartrust.cc/concepts/spine-leaf) - **Rail-optimised network**: A cluster wiring scheme where GPU number one in every server joins one switch group, GPU two another, and so on. (https://everyneocloud.cpn.cleartrust.cc/concepts/rail-optimised) - **DPU**: A network card with its own processors that handles networking, storage and security work so the main server does not have to. (https://everyneocloud.cpn.cleartrust.cc/concepts/dpu) - **Optical transceiver**: A small module that turns electrical signals into light and back, so data can travel over fibre between switches and servers. (https://everyneocloud.cpn.cleartrust.cc/concepts/optical-transceiver) - **Co-packaged optics**: Putting the optical parts of a network link right next to the switch chip, instead of in separate plug-in modules. (https://everyneocloud.cpn.cleartrust.cc/concepts/co-packaged-optics) - **Cluster**: Many servers joined by a fast network and shared storage so they can work together on one job. (https://everyneocloud.cpn.cleartrust.cc/concepts/cluster) - **Pod**: A standard building block of a cluster: a set number of racks and GPUs with their own network layer. (https://everyneocloud.cpn.cleartrust.cc/concepts/pod) - **SuperPOD**: Nvidia's reference design for a complete AI cluster built from DGX systems, network, storage and software. (https://everyneocloud.cpn.cleartrust.cc/concepts/superpod) - **Supercomputer**: One of the fastest computers of its time, usually a very large cluster run by a national lab, university or company. (https://everyneocloud.cpn.cleartrust.cc/concepts/supercomputer) - **High-performance computing**: Using clusters and supercomputers to solve problems too big for ordinary computers, such as weather, physics and AI. (https://everyneocloud.cpn.cleartrust.cc/concepts/hpc) - **Exaflop**: A billion billion (10 to the power 18) calculations per second, the scale of today's fastest computers. (https://everyneocloud.cpn.cleartrust.cc/concepts/exaflop) - **Petaflop**: A thousand million million (10 to the power 15) calculations per second, one thousandth of an exaflop. (https://everyneocloud.cpn.cleartrust.cc/concepts/petaflop) - **FLOPS**: The number of calculations with decimal numbers a chip can do each second, the basic measure of compute. (https://everyneocloud.cpn.cleartrust.cc/concepts/flops) - **Numerical precision**: How many bits a computer uses to store each number; fewer bits mean faster sums but less exact results. (https://everyneocloud.cpn.cleartrust.cc/concepts/precision) - **BF16**: A 16-bit number format created at Google Brain that keeps the range of 32-bit numbers with less detail, now standard for AI training. (https://everyneocloud.cpn.cleartrust.cc/concepts/bf16) - **FP8**: An 8-bit number format, supported since Nvidia's Hopper GPUs in 2022, that doubles AI speed over 16-bit formats. (https://everyneocloud.cpn.cleartrust.cc/concepts/fp8) - **FP4**: A 4-bit number format introduced for AI with Nvidia's Blackwell GPUs, mainly used to make inference faster and cheaper. (https://everyneocloud.cpn.cleartrust.cc/concepts/fp4) - **Sparsity**: Skipping calculations on values that are zero; chip makers quote speeds with sparsity that real models often cannot reach. (https://everyneocloud.cpn.cleartrust.cc/concepts/sparsity) - **MFU**: The share of a chip's peak arithmetic speed that a training job actually turns into useful model work. (https://everyneocloud.cpn.cleartrust.cc/concepts/mfu) - **Utilisation**: How much of the time, or capacity, a provider's GPUs are actually doing paid or useful work. (https://everyneocloud.cpn.cleartrust.cc/concepts/utilisation) - **Goodput**: The share of a cluster's time that produces real progress on a job, after losses to failures, restarts and waiting. (https://everyneocloud.cpn.cleartrust.cc/concepts/goodput) - **Training**: The process of building an AI model by showing it vast amounts of data and adjusting its weights to reduce its errors. (https://everyneocloud.cpn.cleartrust.cc/concepts/training) - **Pre-training**: The first, largest phase of training a model, on trillions of words or images, before it is adapted to tasks. (https://everyneocloud.cpn.cleartrust.cc/concepts/pre-training) - **Post-training**: The phases after pre-training that shape a model's behaviour, such as instruction tuning and reinforcement learning. (https://everyneocloud.cpn.cleartrust.cc/concepts/post-training) - **Fine-tuning**: Further training of an existing model on a smaller, specialised dataset to adapt it to a task or a company's data. (https://everyneocloud.cpn.cleartrust.cc/concepts/fine-tuning) - **Inference**: Running a trained model to produce answers, images or other output for users, as opposed to training it. (https://everyneocloud.cpn.cleartrust.cc/concepts/inference) - **Tokens**: The small chunks of text, roughly three-quarters of an English word each, that language models read and write. (https://everyneocloud.cpn.cleartrust.cc/concepts/tokens) - **Tokens per second**: How fast a model produces text, measured for one user or for a whole server, the key speed figure for inference. (https://everyneocloud.cpn.cleartrust.cc/concepts/tokens-per-second) - **Time to first token**: How long a user waits between sending a prompt and seeing the first word of the reply. (https://everyneocloud.cpn.cleartrust.cc/concepts/time-to-first-token) - **Latency**: The time something takes to happen: a network message to arrive, or a model to answer. (https://everyneocloud.cpn.cleartrust.cc/concepts/latency) - **Batching**: Processing several requests or data samples together, so the chip does more useful work per read of the model. (https://everyneocloud.cpn.cleartrust.cc/concepts/batching) - **KV cache**: Memory a language model keeps of the conversation so far, so it does not recompute it for every new token. (https://everyneocloud.cpn.cleartrust.cc/concepts/kv-cache) - **Prefill and decode**: The two phases of answering a prompt: reading it all at once (prefill), then writing the reply one token at a time (decode). (https://everyneocloud.cpn.cleartrust.cc/concepts/prefill-and-decode) - **Context window**: The most tokens a model can take into account at once, covering the prompt, any documents and its reply. (https://everyneocloud.cpn.cleartrust.cc/concepts/context-window) - **Test-time compute**: Letting a model spend more computation while answering, for example by reasoning step by step, to get better results. (https://everyneocloud.cpn.cleartrust.cc/concepts/test-time-compute) - **Quantisation**: Storing a model's numbers in fewer bits, such as 8 or 4 instead of 16, to make it smaller and faster to run. (https://everyneocloud.cpn.cleartrust.cc/concepts/quantisation) - **Speculative decoding**: Speeding up text generation by letting a small model guess several tokens ahead and a large model check them at once. (https://everyneocloud.cpn.cleartrust.cc/concepts/speculative-decoding) - **Parameters**: The numbers inside a model that training adjusts; their count, in billions, is the usual measure of a model's size. (https://everyneocloud.cpn.cleartrust.cc/concepts/parameters) - **Mixture of experts**: A model design where each token is sent to only a few of many smaller sub-networks, so a huge model runs at the cost of a smaller one. (https://everyneocloud.cpn.cleartrust.cc/concepts/mixture-of-experts) - **Scaling laws**: Observed rules showing that models get predictably better as compute, data and size grow, the reason labs build ever larger clusters. (https://everyneocloud.cpn.cleartrust.cc/concepts/scaling-laws) - **Foundation model**: A large model trained on broad data that can be adapted to many tasks; the most capable are called frontier models. (https://everyneocloud.cpn.cleartrust.cc/concepts/foundation-model) - **Open-weight model**: A model whose trained parameters are published for anyone to download and run, such as Llama, Mistral, Qwen or DeepSeek. (https://everyneocloud.cpn.cleartrust.cc/concepts/open-weight-model) - **Data parallelism**: Training on many GPUs at once by giving each a copy of the model and a different slice of the data, then combining what they learn. (https://everyneocloud.cpn.cleartrust.cc/concepts/data-parallelism) - **Tensor parallelism**: Splitting each layer of a model across several GPUs so they compute it together, needing very fast links between them. (https://everyneocloud.cpn.cleartrust.cc/concepts/tensor-parallelism) - **Pipeline parallelism**: Splitting a model's layers into stages on different GPUs and passing data along them like an assembly line. (https://everyneocloud.cpn.cleartrust.cc/concepts/pipeline-parallelism) - **Expert parallelism**: Placing the different experts of a mixture-of-experts model on different GPUs and routing each token to the right one. (https://everyneocloud.cpn.cleartrust.cc/concepts/expert-parallelism) - **Collective communication**: Standard group operations, such as all-reduce, through which many GPUs share and combine results during training. (https://everyneocloud.cpn.cleartrust.cc/concepts/collective-communication) - **Checkpointing**: Saving the full state of a training job at intervals so it can resume from there after a failure. (https://everyneocloud.cpn.cleartrust.cc/concepts/checkpointing) - **Burn-in testing**: Running a new cluster hard for days or weeks before handing it over, to find faulty GPUs, cables and nodes early. (https://everyneocloud.cpn.cleartrust.cc/concepts/burn-in) - **Job scheduler**: Software that queues jobs on a shared cluster and decides which run where and when. (https://everyneocloud.cpn.cleartrust.cc/concepts/job-scheduler) - **Slurm**: The open-source job scheduler that runs most of the world's supercomputers and many AI training clusters. (https://everyneocloud.cpn.cleartrust.cc/concepts/slurm) - **Kubernetes**: Open-source software, first released by Google in 2014, that runs and manages containers across many servers. (https://everyneocloud.cpn.cleartrust.cc/concepts/kubernetes) - **Containers**: A packaged bundle of an application and everything it needs to run, so it behaves the same on any machine. (https://everyneocloud.cpn.cleartrust.cc/concepts/containers) - **ML framework**: The software library, such as PyTorch or JAX, in which AI models are written, trained and run. (https://everyneocloud.cpn.cleartrust.cc/concepts/ml-framework) - **Inference server**: Software that loads a model onto GPUs and serves requests to it efficiently, handling batching and memory. (https://everyneocloud.cpn.cleartrust.cc/concepts/inference-server) - **Model API**: A web service that runs a model for you and charges by the token, so you never manage GPUs yourself. (https://everyneocloud.cpn.cleartrust.cc/concepts/model-api) - **Serverless GPU**: GPUs that start when a request arrives and shut down when it ends, billed by the second, so idle time costs nothing. (https://everyneocloud.cpn.cleartrust.cc/concepts/serverless-gpu) - **Cold start**: The delay when a service must start a new machine and load a model before it can answer the first request. (https://everyneocloud.cpn.cleartrust.cc/concepts/cold-start) - **Bare metal**: Renting a whole physical server with no virtualisation layer, so the customer gets every bit of its performance. (https://everyneocloud.cpn.cleartrust.cc/concepts/bare-metal) - **Virtual machine**: A software-defined computer that runs on a share of a physical server, rented by the hour from a cloud. (https://everyneocloud.cpn.cleartrust.cc/concepts/virtual-machine) - **Hypervisor**: The software that creates and runs virtual machines on a physical server and keeps them apart. (https://everyneocloud.cpn.cleartrust.cc/concepts/hypervisor) - **Multi-Instance GPU**: An Nvidia feature that splits one data centre GPU into as many as seven smaller, isolated GPUs. (https://everyneocloud.cpn.cleartrust.cc/concepts/mig) - **Multi-tenancy**: Serving many customers on shared infrastructure while keeping their work and data separate. (https://everyneocloud.cpn.cleartrust.cc/concepts/multi-tenancy) - **Confidential computing**: Hardware that keeps data and models encrypted even while they are being processed, so even the cloud cannot see them. (https://everyneocloud.cpn.cleartrust.cc/concepts/confidential-computing) - **Region and availability zone**: A cloud region is a geographic area of data centres; an availability zone is one isolated site or group of sites within it. (https://everyneocloud.cpn.cleartrust.cc/concepts/availability-zone) - **Egress fees**: Charges a cloud makes for data leaving its network, which can make moving large datasets between clouds costly. (https://everyneocloud.cpn.cleartrust.cc/concepts/egress-fees) - **Service level agreement**: The part of a contract that sets the availability and support a provider promises and the credits owed if it falls short. (https://everyneocloud.cpn.cleartrust.cc/concepts/sla) - **Parallel file system**: Storage spread across many servers so thousands of GPUs can read data and save checkpoints at the same time. (https://everyneocloud.cpn.cleartrust.cc/concepts/parallel-file-system) - **Object storage**: Cheap, very large storage that keeps data as objects reached over the web, used for datasets, models and backups. (https://everyneocloud.cpn.cleartrust.cc/concepts/object-storage) - **NVMe**: A fast standard for connecting flash drives directly to a server, used for local scratch space and storage arrays in AI clusters. (https://everyneocloud.cpn.cleartrust.cc/concepts/nvme) - **GPUDirect**: Nvidia technologies that let network cards and storage move data straight into GPU memory, skipping the CPU. (https://everyneocloud.cpn.cleartrust.cc/concepts/gpudirect) - **On-demand**: Renting GPUs by the hour or second with no commitment, at the highest list price but with the most freedom. (https://everyneocloud.cpn.cleartrust.cc/concepts/on-demand) - **Reserved capacity**: Compute a customer commits to use for months or years in return for a lower price and guaranteed access. (https://everyneocloud.cpn.cleartrust.cc/concepts/reserved-capacity) - **Spot instances**: Spare cloud capacity sold at a deep discount that the provider can take back at short notice. (https://everyneocloud.cpn.cleartrust.cc/concepts/spot-instances) - **GPU-hour**: One GPU used for one hour, the basic unit in which cloud compute is priced and training runs are measured. (https://everyneocloud.cpn.cleartrust.cc/concepts/gpu-hour) - **Colocation**: Renting space, power and cooling in someone else's data centre to house your own servers. (https://everyneocloud.cpn.cleartrust.cc/concepts/colocation) - **Powered shell**: A data centre building delivered with its structure and power connection ready, which the tenant then fits out. (https://everyneocloud.cpn.cleartrust.cc/concepts/powered-shell) - **Megawatt**: A million watts; data centres and AI clusters are measured by the megawatts of power their computers can draw. (https://everyneocloud.cpn.cleartrust.cc/concepts/megawatt) - **PUE**: A data centre's total power use divided by the power that reaches its computers; 1.0 would mean no overhead at all. (https://everyneocloud.cpn.cleartrust.cc/concepts/pue) - **Water usage effectiveness**: The litres of water a data centre uses for cooling per kilowatt-hour of computing, a growing concern for AI. (https://everyneocloud.cpn.cleartrust.cc/concepts/wue) - **Liquid cooling**: Removing heat from chips with liquid rather than air, needed for the densest AI racks. (https://everyneocloud.cpn.cleartrust.cc/concepts/liquid-cooling) - **Direct-to-chip cooling**: Liquid cooling where metal plates with fluid channels sit right on top of the GPUs and processors. (https://everyneocloud.cpn.cleartrust.cc/concepts/direct-to-chip) - **Immersion cooling**: Cooling servers by submerging them entirely in a fluid that does not conduct electricity. (https://everyneocloud.cpn.cleartrust.cc/concepts/immersion-cooling) - **Grid interconnection**: The process of connecting a large new power user or generator to the electricity grid, often the slowest step in building AI compute. (https://everyneocloud.cpn.cleartrust.cc/concepts/grid-interconnection) - **Behind-the-meter power**: Electricity generated at or next to a data centre and supplied directly, rather than drawn through the public grid. (https://everyneocloud.cpn.cleartrust.cc/concepts/behind-the-meter) - **Power purchase agreement**: A long contract to buy electricity from a specific power plant or wind or solar farm, often to back clean-energy claims. (https://everyneocloud.cpn.cleartrust.cc/concepts/power-purchase-agreement) - **AI gigafactory**: The European Union's plan for very large AI compute sites, each with around 100,000 advanced chips. (https://everyneocloud.cpn.cleartrust.cc/concepts/ai-gigafactory) - **EuroHPC Joint Undertaking**: The European Union body, set up in 2018, that co-funds Europe's large supercomputers, including JUPITER, LUMI and Leonardo. (https://everyneocloud.cpn.cleartrust.cc/concepts/eurohpc) - **Research allocation**: Compute time granted to researchers through a competitive call, rather than sold at a price. (https://everyneocloud.cpn.cleartrust.cc/concepts/research-allocation) - **IndiaAI Mission**: India's national AI programme, approved in March 2024, which subsidises GPU access and funds Indian models and datasets. (https://everyneocloud.cpn.cleartrust.cc/concepts/indiaai-mission) - **IndiaAI compute portal**: The online platform through which Indian start-ups, researchers and agencies apply for subsidised GPUs under the IndiaAI Mission. (https://everyneocloud.cpn.cleartrust.cc/concepts/indiaai-compute-portal) - **National Supercomputing Mission**: India's programme, launched in 2015, to build PARAM supercomputers across research institutions, run by C-DAC and IISc. (https://everyneocloud.cpn.cleartrust.cc/concepts/national-supercomputing-mission) - **AIRAWAT**: India's national AI supercomputer at C-DAC Pune, built on Nvidia A100 GPUs and ranked 75th on the TOP500 list in 2023. (https://everyneocloud.cpn.cleartrust.cc/concepts/airawat) - **Export controls**: Government rules that restrict selling advanced chips and chip-making tools to certain countries, mainly applied by the United States to China. (https://everyneocloud.cpn.cleartrust.cc/concepts/export-controls) - **AI diffusion rule**: A US rule issued in January 2025 to cap AI chip exports by country in three tiers, withdrawn before it took effect. (https://everyneocloud.cpn.cleartrust.cc/concepts/ai-diffusion-rule) - **Stargate**: A joint venture announced in January 2025 by OpenAI, SoftBank, Oracle and MGX to build AI data centres in the United States. (https://everyneocloud.cpn.cleartrust.cc/concepts/stargate) - **TOP500**: The twice-yearly ranking of the world's 500 fastest supercomputers, published since 1993. (https://everyneocloud.cpn.cleartrust.cc/concepts/top500) - **HPL**: The benchmark used to rank the TOP500, which measures how fast a computer solves a huge system of equations. (https://everyneocloud.cpn.cleartrust.cc/concepts/hpl) - **HPCG**: A supercomputer benchmark that tests memory and network speed, which better reflects many real applications than HPL. (https://everyneocloud.cpn.cleartrust.cc/concepts/hpcg) - **Green500**: A ranking, published with the TOP500 since 2007, that orders supercomputers by energy efficiency rather than raw speed. (https://everyneocloud.cpn.cleartrust.cc/concepts/green500) - **MLPerf**: A set of industry benchmarks, run by the MLCommons group, that compares how fast systems train and run AI models. (https://everyneocloud.cpn.cleartrust.cc/concepts/mlperf) ## Providers - [SpaceX Super Intelligence](https://everyneocloud.cpn.cleartrust.cc/p/spacexai) (2023), United States: Musk's AI company behind Grok and the Colossus data centres, now part of SpaceX. - [G42](https://everyneocloud.cpn.cleartrust.cc/p/g42) (2018), United Arab Emirates: Abu Dhabi AI holding group running cloud, data centres and national compute projects - [Rackspace Technology](https://everyneocloud.cpn.cleartrust.cc/p/rackspace-technology) (1998), United States: Managed cloud and AI services firm running data centres worldwide - [Linode](https://everyneocloud.cpn.cleartrust.cc/p/linode) (2003), United States: Linux virtual machines and cloud infrastructure, now part of Akamai - [OVHcloud](https://everyneocloud.cpn.cleartrust.cc/p/ovhcloud) (1999), France: French cloud and hosting provider running its own servers and data centres - [Google Compute Engine](https://everyneocloud.cpn.cleartrust.cc/p/google-compute-engine) (2012), United States: Google's infrastructure-as-a-service for launching virtual machines on demand - [Titan](https://everyneocloud.cpn.cleartrust.cc/p/titan) (2012), United States: A hybrid supercomputer built by Cray for Oak Ridge National Laboratory. - [IBM Cloud](https://everyneocloud.cpn.cleartrust.cc/p/ibm-cloud), United States: IBM's hybrid cloud platform spanning data centres, bare metal and managed Kubernetes - [PARAM](https://everyneocloud.cpn.cleartrust.cc/p/param) (1991), India: India's home-grown supercomputer line, built by C-DAC in Pune since the early 1990s - [Kabru](https://everyneocloud.cpn.cleartrust.cc/p/kabru), India: A Linux cluster of Pentium Xeon nodes built at IMSc in Chennai for physics simulations - [K computer](https://everyneocloud.cpn.cleartrust.cc/p/k-computer) (2011), Japan: Fujitsu-built Japanese supercomputer at RIKEN, active 2011 to 2019 - [Aurora](https://everyneocloud.cpn.cleartrust.cc/p/aurora) (2023), United States: An exascale supercomputer at Argonne National Laboratory built by Intel and Cray. - [JUWELS](https://everyneocloud.cpn.cleartrust.cc/p/juwels) (2020), Germany: Modular supercomputer at Jülich with an AMD and Nvidia booster module - [Pittsburgh Supercomputing Center](https://everyneocloud.cpn.cleartrust.cc/p/pittsburgh-supercomputing-center) (1986), United States: Joint Carnegie Mellon and Pittsburgh university centre for research computing and networking - [Texas Advanced Computing Center](https://everyneocloud.cpn.cleartrust.cc/p/texas-advanced-computing-center) (2001), United States: University of Texas centre running supercomputers and storage for U.S. research - [Oak Ridge Leadership Computing Facility](https://everyneocloud.cpn.cleartrust.cc/p/oak-ridge-leadership-computing-facility) (1992), United States: US Department of Energy supercomputing user facility at Oak Ridge National Laboratory - [NCAR-Wyoming Supercomputing Center](https://everyneocloud.cpn.cleartrust.cc/p/ncar-wyoming-supercomputing-center) (2003), United States: A high-performance computing and data archival facility in Cheyenne, Wyoming. - [National Computational Infrastructure](https://everyneocloud.cpn.cleartrust.cc/p/national-computational-infrastructure) (2007), Australia: Australia's national high-performance computing and data facility at ANU in Canberra - [Taiwania 3](https://everyneocloud.cpn.cleartrust.cc/p/taiwania-3) (2020), Taiwan: A CPU-based supercomputer for scientific research and biomedical analysis in Taiwan. - [MareNostrum](https://everyneocloud.cpn.cleartrust.cc/p/marenostrum) (2005), Spain: Spain's main supercomputer at the Barcelona Supercomputing Center - [Barcelona Supercomputing Center](https://everyneocloud.cpn.cleartrust.cc/p/barcelona-supercomputing-center) (2005), Spain: Spanish national supercomputing centre running the MareNostrum machine in Barcelona - [Abr Arvan](https://everyneocloud.cpn.cleartrust.cc/p/abr-arvan) (2015), Iran: Iranian CDN and cloud provider offering IaaS, PaaS and object storage - [CINECA](https://everyneocloud.cpn.cleartrust.cc/p/cineca) (1969), Italy: Italian university consortium running supercomputers for scientific research - [Swiss National Supercomputing Centre](https://everyneocloud.cpn.cleartrust.cc/p/swiss-national-supercomputing-centre) (1992), Switzerland: Switzerland's national high-performance computing centre, run by ETH Zurich in Lugano. - [CoreWeave](https://everyneocloud.cpn.cleartrust.cc/p/coreweave) (2017), United States: US GPU cloud running AI data centres for hyperscalers and model labs - [Colossus](https://everyneocloud.cpn.cleartrust.cc/p/colossus-data-center) (2024), United States: An AI supercomputer cluster in Tennessee and Mississippi built by xAI. - [Centre for Development of Advanced Computing](https://everyneocloud.cpn.cleartrust.cc/p/centre-for-development-of-advanced-computing) (1988), India: Indian government society building supercomputers and advanced computing systems - [Leonardo](https://everyneocloud.cpn.cleartrust.cc/p/leonardo) (2022), Italy: Italian petascale supercomputer run by CINECA for European research - [Singtel](https://everyneocloud.cpn.cleartrust.cc/p/singtel) (1879), Singapore: Singapore telecom group offering AI cloud services and data centre capacity - [Oracle Cloud](https://everyneocloud.cpn.cleartrust.cc/p/oracle-cloud) (2016), United States: Oracle's global cloud for servers, storage, databases and GPU compute - [IREN](https://everyneocloud.cpn.cleartrust.cc/p/iren) (2018), Australia: Australian GPU cloud and data centre operator for AI workloads - [Cerebras](https://everyneocloud.cpn.cleartrust.cc/p/cerebras-systems) (2016), United States: Developer of wafer-scale AI chips and supercomputers for inference and training. - [YTL Corporation](https://everyneocloud.cpn.cleartrust.cc/p/ytl-corporation) (1955), Malaysia: Malaysian infrastructure group running AI data centres and Nvidia cloud capacity - [SoftBank Group](https://everyneocloud.cpn.cleartrust.cc/p/softbank-group) (1981), Japan: Japanese investment group behind the Stargate AI infrastructure project - [Google data centers](https://everyneocloud.cpn.cleartrust.cc/p/google-data-centers), United States: Hyperscale campuses running Search, YouTube, Gemini and Google Cloud AI workloads - [LUMI](https://everyneocloud.cpn.cleartrust.cc/p/lumi) (2023), Finland: EuroHPC petascale machine in Kajaani, Finland, built by HPE on AMD hardware - [Amazon Elastic Compute Cloud](https://everyneocloud.cpn.cleartrust.cc/p/amazon-elastic-compute-cloud) (2006), United States: AWS service for renting virtual servers by the second - [NextDC](https://everyneocloud.cpn.cleartrust.cc/p/nextdc) (2010), Australia: Australian data centre operator building AI campuses and GPU superclusters - [The Stargate Project](https://everyneocloud.cpn.cleartrust.cc/p/stargate-llc) (2025), United States: US AI infrastructure venture building data centres for OpenAI - [Pawsey Supercomputing Centre](https://everyneocloud.cpn.cleartrust.cc/p/pawsey-supercomputing-centre) (2014), Australia: National research computing centre in Perth with supercomputers, storage and cloud. - [Nebius Group](https://everyneocloud.cpn.cleartrust.cc/p/nebius-group) (2004), Netherlands: AI infrastructure company offering GPU cloud services, listed on Nasdaq. - [Perlmutter](https://everyneocloud.cpn.cleartrust.cc/p/perlmutter) (2021), United States: A DOE supercomputer at NERSC for fusion and climate research. - [Bitdeer](https://everyneocloud.cpn.cleartrust.cc/p/bitdeer) (2021), Singapore: Singapore crypto-mining group that rents out GPU capacity for AI work - [Hyperion](https://everyneocloud.cpn.cleartrust.cc/p/hyperion-data-center), United States: A planned AI data center in Louisiana owned by Meta Platforms. - [Tencent Cloud](https://everyneocloud.cpn.cleartrust.cc/p/tencent-cloud) (2013), China: Tencent's global cloud arm: 66 availability zones, 700+ services, AI and media tools - [Forschungszentrum Jülich](https://everyneocloud.cpn.cleartrust.cc/p/forschungszentrum-julich) (1956), Germany: German research centre running supercomputers and national HPC services - [Cloudflare](https://everyneocloud.cpn.cleartrust.cc/p/cloudflare) (2009), United States: Edge network running Nvidia GPUs for AI inference via Workers AI - [DigitalOcean](https://everyneocloud.cpn.cleartrust.cc/p/digitalocean) (2011), United States: Alternative cloud for developers, with GPU Droplets for AI work - [Yandex Cloud](https://everyneocloud.cpn.cleartrust.cc/p/yandex) (1997), Russia: Russian cloud computing and AI infrastructure provider - [Hetzner Online](https://everyneocloud.cpn.cleartrust.cc/p/hetzner) (1997), Germany: German data centre operator selling servers, colocation and cloud capacity - [Netweb Technologies](https://everyneocloud.cpn.cleartrust.cc/p/netweb-technologies) (1996), India: Indian maker of supercomputers, AI systems and data centre servers - [MARA Holdings](https://everyneocloud.cpn.cleartrust.cc/p/mara-holdings) (2010), United States: Public bitcoin miner and digital asset company based in Florida - [Core Scientific](https://everyneocloud.cpn.cleartrust.cc/p/core-scientific) (2017), United States: US data centre operator selling GPU cloud and colocation for AI work - [Groq](https://everyneocloud.cpn.cleartrust.cc/p/groq) (2016), United States: Chip designer and inference cloud built on its own Language Processing Unit - [Selene](https://everyneocloud.cpn.cleartrust.cc/p/selene) (2020), United States: Nvidia's own DGX SuperPOD supercomputer, built in three months in 2020 - [Alps](https://everyneocloud.cpn.cleartrust.cc/p/alps) (2024), Switzerland: Swiss national supercomputer using Nvidia GH200 chips for AI and scientific research. - [Sequoia](https://everyneocloud.cpn.cleartrust.cc/p/sequoia) (2011), United States: IBM Blue Gene/Q supercomputer built for US nuclear weapons simulation at Livermore - [NERSC](https://everyneocloud.cpn.cleartrust.cc/p/national-energy-research-scientific-computing-center) (1974), United States: US Department of Energy supercomputing centre for open science research - [Baidu](https://everyneocloud.cpn.cleartrust.cc/p/baidu) (2000), China: Chinese search and AI group with its own cloud, chips and foundation models - [Tesla Dojo](https://everyneocloud.cpn.cleartrust.cc/p/tesla-dojo) (2023), United States: Tesla's custom supercomputer for training its self-driving vision models - [TeraWulf](https://everyneocloud.cpn.cleartrust.cc/p/terawulf) (2021), United States: American bitcoin miner expanding into high-performance computing and AI infrastructure. - [Fugaku](https://everyneocloud.cpn.cleartrust.cc/p/fugaku) (2020), Japan: A Japanese CPU-based supercomputer built by Fujitsu for RIKEN research work. - [Northern Data](https://everyneocloud.cpn.cleartrust.cc/p/northern-data) (2019), Germany: German GPU cloud and data centre group for AI and high-performance computing - [Summit](https://everyneocloud.cpn.cleartrust.cc/p/summit) (2018), United States: A supercomputer built by IBM for the Oak Ridge National Laboratory. - [Taiwania](https://everyneocloud.cpn.cleartrust.cc/p/taiwania) (2018), Taiwan: A series of supercomputers in Taiwan for scientific research and public access. - [Frontier](https://everyneocloud.cpn.cleartrust.cc/p/frontier) (2021), United States: The world's first exascale supercomputer, hosted at Oak Ridge National Laboratory in Tennessee. It uses AMD processors a - [Sierra](https://everyneocloud.cpn.cleartrust.cc/p/sierra) (2018), United States: A US nuclear weapons simulation supercomputer at Lawrence Livermore - [Starcloud](https://everyneocloud.cpn.cleartrust.cc/p/starcloud) (2023), United States: Orbital data centres built and operated by a US space computing company - [Ionos](https://everyneocloud.cpn.cleartrust.cc/p/ionos) (1988), Germany: German web hosting and cloud computing provider, part of United Internet - [Aruba](https://everyneocloud.cpn.cleartrust.cc/p/aruba-s-p-a) (1994), Italy: Italian data centre and cloud provider running its own campuses across Italy and Europe - [Hivenet](https://everyneocloud.cpn.cleartrust.cc/p/hivenet) (2022), Switzerland: Swiss distributed cloud using crowdsourced storage and GPU capacity - [HPC5](https://everyneocloud.cpn.cleartrust.cc/p/hpc5) (2020), Italy: Eni's Dell-built GPU supercomputer in northern Italy - [Piz Daint](https://everyneocloud.cpn.cleartrust.cc/p/piz-daint) (2012), Switzerland: Swiss national supercomputer at CSCS, upgraded to Cray XC50/XC40 with Tesla P100 GPUs - [Nebulae](https://everyneocloud.cpn.cleartrust.cc/p/nebulae) (2010), China: A petascale supercomputer in Shenzhen built by Sugon - [Santos Dumont](https://everyneocloud.cpn.cleartrust.cc/p/santos-dumont) (2024), Brazil: Brazilian public supercomputer run by LNCC in Petrópolis - [Deucalion](https://everyneocloud.cpn.cleartrust.cc/p/deucalion) (2023), Portugal: Portugal's fastest supercomputer, hosted at MACC in Guimarães - [DEGIMA](https://everyneocloud.cpn.cleartrust.cc/p/degima), Japan: GPU cluster at Nagasaki University for astrophysical N-body simulations - [Galileo](https://everyneocloud.cpn.cleartrust.cc/p/galileo) (2015), Italy: Italian national Tier-1 supercomputer for scientific and technical computing - [Genesis Mission](https://everyneocloud.cpn.cleartrust.cc/p/genesis-mission) (2025), United States: US federal programme using AI computing to speed up scientific research - [Spanish Supercomputing Network](https://everyneocloud.cpn.cleartrust.cc/p/spanish-supercomputing-network) (2007), Spain: Spain's linked supercomputers offering free HPC time to research groups - [Tsubame](https://everyneocloud.cpn.cleartrust.cc/p/tsubame) (2006), Japan: A long-running Japanese supercomputer series built for AI and scientific computing. - [Humain](https://everyneocloud.cpn.cleartrust.cc/p/humain) (2025), Saudi Arabia: Saudi AI company building data centres, cloud services and Arabic models - [Deutsche Telekom](https://everyneocloud.cpn.cleartrust.cc/p/deutsche-telekom) (1995), Germany: German telecoms group building industrial AI cloud infrastructure with Nvidia - [YTL AI Cloud](https://everyneocloud.cpn.cleartrust.cc/p/ytl-power) (2024), Malaysia: Malaysian AI cloud and supercomputer run from a solar-backed Johor data centre park - [Amazon Bedrock](https://everyneocloud.cpn.cleartrust.cc/p/amazon-bedrock) (2023), United States: Serverless platform giving one API to many foundation models - [El Capitan](https://everyneocloud.cpn.cleartrust.cc/p/el-capitan) (2024), United States: Exascale supercomputer for US nuclear stockpile stewardship built by HPE and AMD. - [Azure](https://everyneocloud.cpn.cleartrust.cc/p/microsoft-azure) (2010), United States: Microsoft's global cloud computing platform for infrastructure and AI services. - [Nscale](https://everyneocloud.cpn.cleartrust.cc/p/nscale) (2024), United Kingdom: UK-based GPU cloud provider and data centre operator founded in 2024. - [SambaNova Systems](https://everyneocloud.cpn.cleartrust.cc/p/sambanova) (2017), United States: American provider of custom AI silicon and cloud inference services based on Reconfigurable Dataflow Units. - [Naver Cloud](https://everyneocloud.cpn.cleartrust.cc/p/naver-corporation) (1999), South Korea: Cloud arm of Korean internet group Naver, running its own data centre - [OpenRouter](https://everyneocloud.cpn.cleartrust.cc/p/openrouter) (2023), United States: One API that routes requests to hundreds of AI models - [Scaleway](https://everyneocloud.cpn.cleartrust.cc/p/scaleway) (2000), France: French cloud and dedicated server provider run by the Iliad group - [Alibaba Cloud](https://everyneocloud.cpn.cleartrust.cc/p/alibaba-cloud) (2009), China: China's largest cloud provider, running global data centres and the Qwen models - [Gcore](https://everyneocloud.cpn.cleartrust.cc/p/gcore) (2011), Luxembourg: Luxembourg edge AI, cloud, CDN and security provider with GPU clusters in Europe - [Jio Platforms](https://everyneocloud.cpn.cleartrust.cc/p/jio-platforms) (2019), India: Reliance's holding group for Jio telecom, apps, cloud and AI services - [Amazon Web Services](https://everyneocloud.cpn.cleartrust.cc/p/amazon-web-services) (2006), United States: Amazon's on-demand cloud platform, offering compute, storage and AI infrastructure worldwide - [Schwarz Digits](https://everyneocloud.cpn.cleartrust.cc/p/schwarz-digits) (2023), Germany: German retail group's IT arm running the STACKIT sovereign cloud - [Yandex Cloud](https://everyneocloud.cpn.cleartrust.cc/p/yandex-cloud) (2018), Russia: Yandex's public cloud, offering compute, data and ML services to Russian businesses - [DiRAC](https://everyneocloud.cpn.cleartrust.cc/p/dirac) (2009), United Kingdom: UK supercomputing service for particle physics, astronomy and cosmology research - [Christofari](https://everyneocloud.cpn.cleartrust.cc/p/christofari) (2019), Russia: Sberbank's Nvidia-based supercomputers in Moscow for neural network training - [Taipei-1](https://everyneocloud.cpn.cleartrust.cc/p/taipei-1) (2024), Taiwan: Nvidia-owned supercomputer in Taiwan's Kaohsiung Software Park - [Voyager-EUS2](https://everyneocloud.cpn.cleartrust.cc/p/voyager-eus2), United States: Microsoft Azure supercomputer in the East US 2 region - [National Supercomputer Center in Guangzhou](https://everyneocloud.cpn.cleartrust.cc/p/national-supercomputer-center-in-guangzhou) (2014), China: Guangzhou centre running the Tianhe-3 supercomputer for Chinese government use - [Gauss Centre for Supercomputing](https://everyneocloud.cpn.cleartrust.cc/p/gauss-centre-for-supercomputing) (2007), Germany: Coordinates Germany's three national Tier-0 supercomputing centres - [Megaspeed](https://everyneocloud.cpn.cleartrust.cc/p/megaspeed) (2015), Singapore: Singapore AI cloud firm investigated over Nvidia chip diversion to China - [Earth Simulator](https://everyneocloud.cpn.cleartrust.cc/p/earth-simulator) (2002), Japan: Japanese vector supercomputer series built for global climate and solid earth simulation. - [Blue Waters](https://everyneocloud.cpn.cleartrust.cc/p/blue-waters) (2012), United States: A US petascale research supercomputer at the University of Illinois Urbana-Champaign - [Akamai Technologies](https://everyneocloud.cpn.cleartrust.cc/p/akamai-technologies) (1998), United States: Edge platform and cloud compute provider, built on the Linode hosting business - [Gyoukou](https://everyneocloud.cpn.cleartrust.cc/p/gyoukou) (2017), Japan: Immersion-cooled Japanese supercomputer built by ExaScaler and PEZY Computing - [Irish Centre for High-End Computing](https://everyneocloud.cpn.cleartrust.cc/p/irish-centre-for-high-end-computing) (2005), Ireland: Ireland's national high-performance computing centre, running supercomputers for research. - [National Supercomputing Center of Tianjin](https://everyneocloud.cpn.cleartrust.cc/p/national-supercomputing-center-of-tianjin) (1984), China: National supercomputing facility in Tianjin running the Tianhe-1A machine - [Research and Technology Computing Center](https://everyneocloud.cpn.cleartrust.cc/p/research-and-technology-computing-center-france) (2003), France: French public supercomputing centre running the Tera 100 machine - [National Center for Supercomputing Applications](https://everyneocloud.cpn.cleartrust.cc/p/national-center-for-supercomputing-applications) (1986), United States: US university centre supplying high-performance computing to researchers nationwide - [CESGA](https://everyneocloud.cpn.cleartrust.cc/p/cesga) (1993), Spain: Galicia's public high performance computing centre in Santiago de Compostela - [National Supercomputer Centre in Sweden](https://everyneocloud.cpn.cleartrust.cc/p/national-supercomputer-centre-in-sweden), Sweden: Swedish national supercomputer centre in Linköping, running Triolith. - [T Cloud Public](https://everyneocloud.cpn.cleartrust.cc/p/t-cloud-public) (2016), Germany: Deutsche Telekom's OpenStack-based infrastructure-as-a-service cloud, run by T-Systems. - [GENCI](https://everyneocloud.cpn.cleartrust.cc/p/genci) (2007), France: French state-backed body funding and coordinating national HPC centres - [High Performance Computing Center, Stuttgart](https://everyneocloud.cpn.cleartrust.cc/p/high-performance-computing-center-stuttgart) (1995), Germany: University of Stuttgart supercomputer centre running the Hawk system - [Leibniz Supercomputing Centre](https://everyneocloud.cpn.cleartrust.cc/p/leibniz-supercomputing-centre) (1962), Germany: Bavarian academy-run centre offering supercomputing resources to research - [Shaheen](https://everyneocloud.cpn.cleartrust.cc/p/shaheen) (2009), Saudi Arabia: KAUST's flagship supercomputer series in Saudi Arabia - [Tianhe-2](https://everyneocloud.cpn.cleartrust.cc/p/tianhe-2) (2013), China: A 33.86-petaflops supercomputer in Guangzhou, built by NUDT with Inspur. - [EuroHPC Joint Undertaking](https://everyneocloud.cpn.cleartrust.cc/p/european-high-performance-computing-joint-undertaking) (2018), Luxembourg: EU public-private partnership building a pan-European supercomputing and AI infrastructure - [Roadrunner](https://everyneocloud.cpn.cleartrust.cc/p/roadrunner) (2008), United States: IBM-built hybrid supercomputer at Los Alamos, first to sustain a petaflop - [Tianhe-1](https://everyneocloud.cpn.cleartrust.cc/p/tianhe-1) (2009), China: Chinese petascale supercomputer at the National Supercomputing Center in Tianjin - [Crusoe](https://everyneocloud.cpn.cleartrust.cc/p/crusoe), United States: AI cloud and modular data centre operator that also builds its own power supply - [Radiant](https://everyneocloud.cpn.cleartrust.cc/p/ori), United Kingdom: AI compute infrastructure company building repeatable 'AI factories' powered by its FlightDeck platform. - [CUDO Compute](https://everyneocloud.cpn.cleartrust.cc/p/cudo-compute), United Kingdom: European AI infrastructure builder deploying and operating NVIDIA GPU clusters - [Massed Compute](https://everyneocloud.cpn.cleartrust.cc/p/massed-compute), United States: GPU cloud with per-second pricing and no bandwidth billing - [GreenNode](https://everyneocloud.cpn.cleartrust.cc/p/greennode), Vietnam: Sovereign AI and public cloud with GPU instances across Vietnam and Thailand - [Highreso](https://everyneocloud.cpn.cleartrust.cc/p/highreso), Japan: Japanese GPU cloud operator running its own GPU-only data centres - [Ubilink](https://everyneocloud.cpn.cleartrust.cc/p/ubilink), Taiwan: Taiwanese AI supercomputing firm running an NVIDIA H100 cluster - [Ubitus](https://everyneocloud.cpn.cleartrust.cc/p/ubitus) (2012), Japan: GPU virtualisation and cloud gaming company running AI compute centres across Asia - [Sesterce](https://everyneocloud.cpn.cleartrust.cc/p/sesterce), France: European builder of dedicated bare-metal GPU supercomputers on its own campuses - [Verda](https://everyneocloud.cpn.cleartrust.cc/p/verda-formerly-datacrunch), Finland: European full-stack AI cloud offering GPU clusters, instances and model APIs - [Fal](https://everyneocloud.cpn.cleartrust.cc/p/fal) (2021), United States: Serverless GPU platform for generative AI inference - [Anyscale](https://everyneocloud.cpn.cleartrust.cc/p/anyscale), United States: Managed Ray platform for distributed AI workloads - [Novita AI](https://everyneocloud.cpn.cleartrust.cc/p/novita-ai), United States: AI cloud platform offering model APIs and GPU instances for developers. - [Modal](https://everyneocloud.cpn.cleartrust.cc/p/modal), United States: Serverless cloud platform for AI development and inference - [DeepInfra](https://everyneocloud.cpn.cleartrust.cc/p/deepinfra) (2022), United States: Inference cloud for open-source AI models - [Featherless AI](https://everyneocloud.cpn.cleartrust.cc/p/featherless-ai) (2023), United States: Serverless inference platform for open-source AI models - [Cerebrium](https://everyneocloud.cpn.cleartrust.cc/p/cerebrium), South Africa: Serverless GPU infrastructure for real-time AI workloads - [Inference.net](https://everyneocloud.cpn.cleartrust.cc/p/inference-net), United States: Marketplace for low-cost AI inference via unused GPU capacity - [PPIO](https://everyneocloud.cpn.cleartrust.cc/p/ppio) (2018), China: Independent AI cloud for agent infrastructure and GPU compute - [Fireworks AI](https://everyneocloud.cpn.cleartrust.cc/p/fireworks-ai), United States: Platform for running and training open AI models - [Runware](https://everyneocloud.cpn.cleartrust.cc/p/runware) (2023), United States: Serverless AI inference platform with custom hardware - [Koyeb](https://everyneocloud.cpn.cleartrust.cc/p/koyeb) (2021), France: Serverless cloud platform for AI inference and applications - [Siam AI](https://everyneocloud.cpn.cleartrust.cc/p/siam-ai), Thailand: GPU cloud provider based in Thailand for enterprise workloads - [RunPod](https://everyneocloud.cpn.cleartrust.cc/p/runpod), United States: AI developer cloud for training and inference - [Saturn Cloud](https://everyneocloud.cpn.cleartrust.cc/p/saturn-cloud), United States: Managed inference platform for open AI models on dedicated GPUs - [Lightning AI](https://everyneocloud.cpn.cleartrust.cc/p/lightning-ai) (2026), United States: AI cloud platform with owned GPU infrastructure - [Exoscale](https://everyneocloud.cpn.cleartrust.cc/p/exoscale) (2011), Switzerland: European cloud provider for developers and AI workloads - [Zenlayer](https://everyneocloud.cpn.cleartrust.cc/p/zenlayer) (2014), United States: Distributed cloud infrastructure for global AI inference - [Hostkey](https://everyneocloud.cpn.cleartrust.cc/p/hostkey) (2007), Netherlands: Dedicated GPU and bare-metal server provider in Europe and the US - [Hivelocity](https://everyneocloud.cpn.cleartrust.cc/p/hivelocity), United States: Bare metal and GPU infrastructure for AI workloads - [TensorDock](https://everyneocloud.cpn.cleartrust.cc/p/tensordock), United States: Marketplace for on-demand GPU servers and cloud instances - [Vast.ai](https://everyneocloud.cpn.cleartrust.cc/p/vast-ai) (2016), United States: Marketplace for distributed GPU cloud compute - [GPU Mart](https://everyneocloud.cpn.cleartrust.cc/p/gpu-mart) (2021), United States: Dedicated GPU VPS and bare-metal servers for AI and rendering. - [Hyperbolic](https://everyneocloud.cpn.cleartrust.cc/p/hyperbolic), United States: Unified platform for global GPU compute access and orchestration - [Compute Exchange](https://everyneocloud.cpn.cleartrust.cc/p/compute-exchange), United States: Marketplace for buying and selling GPU compute capacity and hardware. - [Render Network](https://everyneocloud.cpn.cleartrust.cc/p/render-network): Decentralised GPU network for rendering and AI compute - [Nosana](https://everyneocloud.cpn.cleartrust.cc/p/nosana): Decentralised marketplace for on-demand GPU compute - [Io.net](https://everyneocloud.cpn.cleartrust.cc/p/io-net) (2023): Decentralised network for GPU compute - [Spheron Network](https://everyneocloud.cpn.cleartrust.cc/p/spheron-network): Marketplace for renting enterprise NVIDIA and AMD GPUs - [SaladCloud](https://everyneocloud.cpn.cleartrust.cc/p/saladcloud) (2018), United States: Distributed GPU cloud using idle consumer hardware - [Clore.ai](https://everyneocloud.cpn.cleartrust.cc/p/clore-ai): Decentralised marketplace for renting NVIDIA GPUs - [Lium](https://everyneocloud.cpn.cleartrust.cc/p/lium), Saint Kitts and Nevis: GPU rental marketplace with per-second billing - [Livepeer](https://everyneocloud.cpn.cleartrust.cc/p/livepeer): Open network for AI video inference workloads - [Theta EdgeCloud](https://everyneocloud.cpn.cleartrust.cc/p/theta-edgecloud): Distributed GPU network for AI inference and agents - [Phala Cloud](https://everyneocloud.cpn.cleartrust.cc/p/phala-cloud), United States: Verifiable private AI compute on GPU TEEs - [Ornn](https://everyneocloud.cpn.cleartrust.cc/p/ornn): Marketplace for buying and trading GPU compute capacity - [Kinesis Network](https://everyneocloud.cpn.cleartrust.cc/p/kinesis-network), United States: Control plane for distributed GPU compute - [Applied Digital](https://everyneocloud.cpn.cleartrust.cc/p/applied-digital) (2021), United States: Builder and operator of large-scale AI data centres - [Mawson Infrastructure Group](https://everyneocloud.cpn.cleartrust.cc/p/mawson-infrastructure-group), United States: Digital infrastructure provider for high-performance compute and colocation. - [DigiPowerX](https://everyneocloud.cpn.cleartrust.cc/p/digi-power-x), United States: Vertically integrated AI infrastructure with owned power and GPU compute. - [Keel Infrastructure](https://everyneocloud.cpn.cleartrust.cc/p/bitfarms) (2017), United States: Energy-backed digital infrastructure for high-performance computing - [CleanSpark](https://everyneocloud.cpn.cleartrust.cc/p/cleanspark), United States: Digital infrastructure developer for AI and bitcoin mining - [Bitzero](https://everyneocloud.cpn.cleartrust.cc/p/bitzero), Norway: Nordic AI data centre operator with over 1 GW of power capacity. - [Nvidia DGX Cloud Lepton](https://everyneocloud.cpn.cleartrust.cc/p/nvidia-dgx-cloud-lepton), United States: Unified platform for multi-cloud GPU compute access - [Eos](https://everyneocloud.cpn.cleartrust.cc/p/eos-nvidia) (2023), United States: Nvidia's DGX SuperPOD AI supercomputer for internal research - [Isambard-AI](https://everyneocloud.cpn.cleartrust.cc/p/isambard-ai) (2025), United Kingdom: UK’s most powerful AI supercomputer for research and industry - [Doudna](https://everyneocloud.cpn.cleartrust.cc/p/doudna) (2026), United States: US Department of Energy supercomputer for AI and science - [Alice Recoque](https://everyneocloud.cpn.cleartrust.cc/p/alice-recoque) (2026), France: France's first exascale supercomputer for research and AI - [Horizon](https://everyneocloud.cpn.cleartrust.cc/p/horizon-tacc) (2026), United States: Largest academic supercomputer in the United States - [Gefion](https://everyneocloud.cpn.cleartrust.cc/p/gefion): Denmark's sovereign AI supercomputer for science and industry - [Equinox](https://everyneocloud.cpn.cleartrust.cc/p/equinox-argonne) (2026), United States: 10,000 GPU AI supercomputer at Argonne National Laboratory - [ASPIRE 2A+](https://everyneocloud.cpn.cleartrust.cc/p/aspire-2a-plus), Singapore: Singapore national AI supercomputer with 320 Nvidia H100 GPUs - [Lux](https://everyneocloud.cpn.cleartrust.cc/p/lux) (2026), United States: AMD-powered AI supercomputer at Oak Ridge National Lab - [Discovery](https://everyneocloud.cpn.cleartrust.cc/p/discovery-oak-ridge) (2028), United States: AI supercomputer for the U.S. Department of Energy - [Ionstream](https://everyneocloud.cpn.cleartrust.cc/p/ionstream), United States: Bare-metal NVIDIA GPU servers with automated provisioning and optimised networking - [Cortex](https://everyneocloud.cpn.cleartrust.cc/p/cortex-tesla), United States: Tesla's AI training cluster for autonomous driving and robotics. - [Stargate Norway](https://everyneocloud.cpn.cleartrust.cc/p/stargate-norway) (2025), Norway: 100,000 GPU AI infrastructure project in Northern Norway - [AI Research Resource](https://everyneocloud.cpn.cleartrust.cc/p/ai-research-resource-uk) (2023), United Kingdom: UK national supercomputing programme for AI research and industry - [EuroHPC AI Factories](https://everyneocloud.cpn.cleartrust.cc/p/eurohpc-ai-factories) (2024): Pan-European network of AI-optimised supercomputing hubs - [AIRAWAT](https://everyneocloud.cpn.cleartrust.cc/p/airawat) (2023), India: India's national AI supercomputer for research and public welfare - [Yotta Shakti Cloud](https://everyneocloud.cpn.cleartrust.cc/p/yotta-shakti-cloud), India: India's sovereign AI cloud platform with NVIDIA GPU infrastructure - [Cyfuture Cloud](https://everyneocloud.cpn.cleartrust.cc/p/cyfuture-cloud), India: Indian cloud provider offering GPU and virtual machine hosting. - [IndiaAI Compute Portal](https://everyneocloud.cpn.cleartrust.cc/p/indiaai-compute-portal), India: National cloud platform for affordable AI compute in India - [ESDS](https://everyneocloud.cpn.cleartrust.cc/p/esds) (2005), India: Indian-owned cloud and data centre provider with AI infrastructure. - [Vensysco](https://everyneocloud.cpn.cleartrust.cc/p/vensysco) (2016), India: Indian AI compute and secure exam solutions provider - [RackBank](https://everyneocloud.cpn.cleartrust.cc/p/rackbank), India: AI data centre provider in India offering bare metal GPU hosting. - [NAIRR Pilot](https://everyneocloud.cpn.cleartrust.cc/p/national-ai-research-resource-pilot) (2024), United States: US federal pilot providing shared AI compute and data to researchers - [Latitude.sh](https://everyneocloud.cpn.cleartrust.cc/p/latitude-sh), United States: Global bare metal cloud for AI workloads - [Prometheus](https://everyneocloud.cpn.cleartrust.cc/p/prometheus-meta) (2026), United States: Meta's first multi-gigawatt AI training cluster - [Thunder Compute](https://everyneocloud.cpn.cleartrust.cc/p/thunder-compute), United States: GPU sandboxes and instances for AI developers - [Solstice](https://everyneocloud.cpn.cleartrust.cc/p/solstice), United States: DOE AI supercomputer with 100,000 Nvidia Blackwell GPUs at Argonne - [WhiteFiber](https://everyneocloud.cpn.cleartrust.cc/p/bit-digital), United States: AI and HPC infrastructure provider owned by Bit Digital - [HIVE Digital Technologies](https://everyneocloud.cpn.cleartrust.cc/p/hive-digital-technologies) (2017), Canada: Clean-energy data centres for crypto mining and AI compute - [Flux](https://everyneocloud.cpn.cleartrust.cc/p/flux): Decentralised cloud network for GPU workloads and web hosting - [NodeShift](https://everyneocloud.cpn.cleartrust.cc/p/nodeshift) (2024), United Kingdom: Decentralized cloud platform for excess compute and storage - [Aethir](https://everyneocloud.cpn.cleartrust.cc/p/aethir): Distributed cloud GPU network for AI and gaming - [Seeweb](https://everyneocloud.cpn.cleartrust.cc/p/seeweb), Italy: European cloud provider with GPU and Kubernetes services - [SF Compute](https://everyneocloud.cpn.cleartrust.cc/p/sf-compute), United States: GPU cloud with a resale market for reserved capacity - [NHN Cloud](https://everyneocloud.cpn.cleartrust.cc/p/nhn-cloud), South Korea: South Korean cloud provider with large GPU clusters for AI training - [Parasail](https://everyneocloud.cpn.cleartrust.cc/p/parasail), United States: Inference cloud for AI startups, billed by token. - [Baseten](https://everyneocloud.cpn.cleartrust.cc/p/baseten), United States: Inference cloud for running AI models in production - [Inferless](https://everyneocloud.cpn.cleartrust.cc/p/inferless), United States: Serverless GPU inference platform for custom models - [Blue Lion](https://everyneocloud.cpn.cleartrust.cc/p/blue-lion) (2027), Germany: German supercomputer for AI and climate science - [Golem Network](https://everyneocloud.cpn.cleartrust.cc/p/golem-network) (2016), Switzerland: Decentralized marketplace for GPU and CPU compute - [Canadian Sovereign AI Compute Strategy](https://everyneocloud.cpn.cleartrust.cc/p/canadian-sovereign-ai-compute-strategy) (2024), Canada: National programme to build domestic AI infrastructure in Canada - [Krutrim Cloud](https://everyneocloud.cpn.cleartrust.cc/p/krutrim-cloud), India: India-native AI cloud platform for training and inference - [Hydra Host](https://everyneocloud.cpn.cleartrust.cc/p/hydra-host), United States: Bare metal GPU platform and Brokkr software for independent AI Factories - [Shadeform](https://everyneocloud.cpn.cleartrust.cc/p/shadeform), United States: Marketplace for GPU cloud access across multiple providers - [Lambda](https://everyneocloud.cpn.cleartrust.cc/p/lambda) (2012), United States: AI-only GPU cloud running NVIDIA supercomputers for training and inference - [Avian](https://everyneocloud.cpn.cleartrust.cc/p/avian), United States: Serverless and dedicated GPU inference for open-source models - [Riot Platforms](https://everyneocloud.cpn.cleartrust.cc/p/riot-platforms), United States: Bitcoin miner offering data centre power for AI workloads - [Hot Aisle](https://everyneocloud.cpn.cleartrust.cc/p/hot-aisle) (2023): Developer-first cloud for sovereign inference on isolated AMD GPUs - [Beam](https://everyneocloud.cpn.cleartrust.cc/p/beam): Developer cloud for AI agents and GPU workloads - [Microsoft Fairwater](https://everyneocloud.cpn.cleartrust.cc/p/microsoft-fairwater) (2025), United States: Purpose-built AI supercomputer campus in Wisconsin - [FriendliAI](https://everyneocloud.cpn.cleartrust.cc/p/friendliai), South Korea: Accelerated generative AI inference serving platform - [Oblivus](https://everyneocloud.cpn.cleartrust.cc/p/oblivus): On-demand and reserved GPU cloud for training, inference and rendering - [Soluna](https://everyneocloud.cpn.cleartrust.cc/p/soluna), United States: Renewable energy-powered AI and Bitcoin cloud - [Sharon AI](https://everyneocloud.cpn.cleartrust.cc/p/sharon-ai), Australia: Australian firm building AI data centres and GPU cloud capacity - [GMI Cloud](https://everyneocloud.cpn.cleartrust.cc/p/gmi-cloud): AI-native cloud for production inference, GPU clusters and agent runtimes - [Fluence](https://everyneocloud.cpn.cleartrust.cc/p/fluence) (2017), Switzerland: Decentralised marketplace for GPU and CPU cloud compute - [Mithril](https://everyneocloud.cpn.cleartrust.cc/p/mithril-formerly-foundry), United States: A GPU VM marketplace with spot bids and flexible reservations for AI workloads - [Coatlicue](https://everyneocloud.cpn.cleartrust.cc/p/coatlicue), Mexico: Mexican national supercomputer for AI and climate modelling - [Arc Compute](https://everyneocloud.cpn.cleartrust.cc/p/arc-compute) (2018), United States: Turnkey NVIDIA GPU cluster design, deployment and management for AI workloads - [Hon Hai Kaohsiung Super Computing Center](https://everyneocloud.cpn.cleartrust.cc/p/hon-hai-kaohsiung-super-computing-center) (2025), Taiwan: Taiwan’s fastest AI supercomputer built by Foxconn and Nvidia - [Replicate](https://everyneocloud.cpn.cleartrust.cc/p/replicate): Run, fine-tune and deploy AI models through one API - [Firmus](https://everyneocloud.cpn.cleartrust.cc/p/firmus-sustainable-metal-cloud), Australia: Neocloud operator building immersion-cooled AI Factories across Asia-Pacific - [Core42](https://everyneocloud.cpn.cleartrust.cc/p/core42), United Arab Emirates: Sovereign AI cloud and infrastructure provider based in the UAE. - [Akash Network](https://everyneocloud.cpn.cleartrust.cc/p/akash-network): Decentralized marketplace for GPU and CPU compute power - [Israel-1](https://everyneocloud.cpn.cleartrust.cc/p/israel-1) (2023), Israel: Nvidia AI supercomputer in Israel for research and partners - [Simplismart](https://everyneocloud.cpn.cleartrust.cc/p/simplismart), India: AI inference platform for custom model deployment and scaling - [Targon](https://everyneocloud.cpn.cleartrust.cc/p/targon), United States: On-demand NVIDIA GPU cloud for training and inference - [Berzelius](https://everyneocloud.cpn.cleartrust.cc/p/berzelius), Sweden: Swedish AI supercomputer at Linköping University - [Hyperstack](https://everyneocloud.cpn.cleartrust.cc/p/hyperstack-nexgen-cloud) (2020), United Kingdom: Self-serve GPU cloud from UK-owned NexGen Cloud, sold as Hyperstack - [Dawn](https://everyneocloud.cpn.cleartrust.cc/p/dawn-cambridge), United Kingdom: UK national AI supercomputer for research and industry - [Nvidia DGX Cloud](https://everyneocloud.cpn.cleartrust.cc/p/nvidia-dgx-cloud), United States: Nvidia's internal cloud for AI development and operations - [TensorWave](https://everyneocloud.cpn.cleartrust.cc/p/tensorwave): AMD-exclusive AI cloud for training and running large models - [E2E Networks](https://everyneocloud.cpn.cleartrust.cc/p/e2e-networks) (2009), India: India's listed GPU cloud for AI training and inference. - [Voltage Park](https://everyneocloud.cpn.cleartrust.cc/p/voltage-park), United States: US GPU cloud selling on-demand and reserved AI compute from its own hardware - [WhiteFiber](https://everyneocloud.cpn.cleartrust.cc/p/whitefiber), United States: AI data centres, private GPU clouds and reserved GPU capacity in North America - [Polarise](https://everyneocloud.cpn.cleartrust.cc/p/polarise) (2024), Germany: German sovereign AI cloud running its own data centres and open-weight model APIs - [Jarvislabs](https://everyneocloud.cpn.cleartrust.cc/p/jarvislabs), India: GPU cloud and inference platform based in India - [Neysa](https://everyneocloud.cpn.cleartrust.cc/p/neysa), India: GPU cloud and AI platform for enterprises in India - [Chutes](https://everyneocloud.cpn.cleartrust.cc/p/chutes): Serverless, decentralised compute for running open-source AI models in production - [Together AI](https://everyneocloud.cpn.cleartrust.cc/p/together-ai), United States: AI cloud running open models on demand, with GPU clusters and inference research - [Fluidstack](https://everyneocloud.cpn.cleartrust.cc/p/fluidstack): Compute infrastructure company building large-scale AI data centres