CoreWeave
American cloud-computing company specializing in GPU-accelerated infrastructure for artificial intelligence and other demanding workloads.
Last updated October 14, 2026
AI crawler reads
No AI crawler has read this profile since Oct 28, 2026.
Overview
CoreWeave is an American cloud-computing company focused on supplying graphics processing unit infrastructure and related software to artificial-intelligence developers, enterprises, research organizations, and other high-performance-computing users. Based in Livingston, New Jersey, the company operates a specialized cloud platform built around large-scale GPU clusters rather than the broad general-purpose infrastructure model associated with traditional hyperscalers. Its services support AI model training and inference, machine learning, visual effects and rendering, scientific computing, and other applications that require substantial parallel processing capacity. The company was established in New Jersey in 2017 under the name Atlantic Crypto by commodities traders Michael Intrator, Brian Venturo, Brannin McBee, and Peter Salanki. Its original business involved mining Ethereum with graphics processors. After the cryptocurrency market deteriorated in 2018, the founders redirected the company toward renting its accumulated GPU capacity to external customers. The business adopted the CoreWeave name in 2019 and gradually evolved from a crypto-mining operation into a dedicated GPU cloud provider. CoreWeave’s growth accelerated as demand for generative AI increased in 2022 and 2023. The company invested heavily in Nvidia’s high-end accelerators, including the H100, and expanded its data-center footprint. Its customer base included AI companies such as Stability AI, while Nvidia became both an investor and a major technology partner. CoreWeave also developed a financing model in which specialized GPU hardware could support debt facilities, enabling it to acquire equipment and construct data centers at a rapid pace. In 2023, it secured a multibillion-dollar debt facility backed by Nvidia H100 systems, a transaction widely noted because of the novelty of using AI accelerators as collateral. The CoreWeave Cloud Platform uses a Kubernetes-oriented architecture intended for containerized, distributed workloads. Its Mission Control software gives customers tools to manage and validate the performance of the infrastructure allocated to them. Offerings include GPU instances and clusters, cloud storage, networking, managed Kubernetes, orchestration, and services for AI development and deployment. Some facilities serve multiple customers, while others are configured for a single customer or a specific strategic project. A notable example is a large Nvidia supercomputer installation in Plano, Texas, designed around thousands of H100 GPUs. CoreWeave expanded into Europe, including the United Kingdom, and established additional sites in the United States. The company has also worked with server, networking, semiconductor, university, and infrastructure partners to increase available capacity. Its strategy has increasingly combined cloud services with ownership or long-term control of data-center capacity, power, networking, and accelerator fleets. In 2024 and 2025, CoreWeave raised substantial capital, prepared for and completed an initial public offering, and announced major customer and acquisition agreements. It acquired Weights & Biases, an AI developer-platform company, and signed a large multiyear infrastructure agreement with OpenAI. The company also became an early cloud supplier of Nvidia’s Blackwell-generation systems and expanded into newer Blackwell Ultra hardware. These developments strengthened its positioning as an AI infrastructure specialist, although the business remains capital-intensive and exposed to hardware supply, energy availability, customer concentration, financing costs, and the rapid pace of change in AI computing. CoreWeave’s current identity is therefore that of a focused AI infrastructure provider: it supplies the compute, software controls, data-center capacity, and technical environment needed to build and operate advanced AI systems, while increasingly integrating platform capabilities and strategic partnerships into its offering.
History
CoreWeave began in 2017 as Atlantic Crypto, a New Jersey cryptocurrency-mining company founded by Michael Intrator, Brian Venturo, Brannin McBee, and Peter Salanki. The business used graphics processors to mine Ethereum. Following the cryptocurrency market collapse in 2018, the founders recognized that their accumulated GPU fleet could serve another market. In 2019, the company adopted the CoreWeave name and shifted toward supplying rented computing capacity to outside users. The early cloud business was differentiated by its focus on GPU-intensive workloads. Rather than competing primarily as a general-purpose cloud, CoreWeave concentrated on machine learning, visual rendering, scientific workloads, and other applications that benefit from parallel accelerator computing. It developed infrastructure and operational software around a Kubernetes-native architecture, allowing customers to schedule and manage containerized workloads across large GPU clusters. CoreWeave’s expansion became particularly rapid in 2022 and 2023, when the demand for generative-AI development created a shortage of advanced accelerators. The company committed approximately $100 million to Nvidia H100 purchases during 2022 and expanded its data centers. It served customers including Stability AI and announced a startup accelerator program that combined compute credits with discounts and hardware support. Nvidia invested $100 million in CoreWeave in April 2023, reinforcing the relationship between the two companies. The company also used structured finance to fund its hardware-intensive expansion. In August 2023, it obtained a $2.3 billion debt facility led by Magnetar Capital and Blackstone, with Nvidia H100 GPUs serving as collateral. CoreWeave expanded its physical footprint in the United States and subsequently in the United Kingdom and other European markets. Its facilities ranged from shared cloud sites to dedicated installations for individual customers or strategic projects. A major Nvidia installation in Plano, Texas, was designed as a large AI supercomputer using thousands of H100 accelerators. During 2024, CoreWeave raised $1.1 billion in equity financing led by Coatue Management and announced additional credit capacity for data-center expansion. It planned significant investment in British data centers, opened a London headquarters, and leased space for a large New Jersey data center. The company also recruited senior executives, including CFO Nitin Agrawal, COO Sachin Jain, engineering leader Chen Goldberg, and chief people officer Michelle O’Rourke. In the same period, it became an early recipient of Nvidia’s next-generation Blackwell hardware and worked with Dell on infrastructure. CoreWeave entered a new phase in 2025. It filed for and completed an initial public offering on Nasdaq under CRWV, signed a large multiyear agreement with OpenAI, and announced the acquisition of Weights & Biases. The latter broadened its scope from infrastructure provision toward the software layer used by AI teams to build and monitor models. CoreWeave also introduced Nvidia GB200 systems through its cloud and later deployed Blackwell Ultra hardware commercially. Its reported footprint grew to dozens of data centers and hundreds of thousands of GPUs, although the business remained dependent on large capital commitments and long-term customer contracts. The company’s development illustrates the transformation of surplus or repurposed GPU capacity into dedicated AI infrastructure. Its principal strengths are accelerator availability, specialized operations, close Nvidia alignment, and a platform designed for large AI workloads. Its principal challenges include high financing and construction costs, power and data-center constraints, customer concentration, hardware obsolescence, and competition from both established cloud companies and other GPU infrastructure providers.
- 2025
Completes Nasdaq IPO
[1]CoreWeave became a public company under the ticker CRWV and raised approximately $1.5 billion in the offering.
- 2025
Agrees to acquire Weights & Biases
[1]The proposed transaction extended CoreWeave’s platform into machine-learning development and model-management software.
- 2025
Signs major OpenAI infrastructure agreement
[1]OpenAI agreed to purchase substantial CoreWeave capacity under a multiyear cloud-computing contract.
- 2024
Expands into the United Kingdom
[1]CoreWeave opened a London headquarters and expanded its UK data-center plans.
- 2023
Nvidia investment and GPU-backed financing
[1]Nvidia invested in CoreWeave, while the company arranged a $2.3 billion debt facility secured in part by H100 hardware.
- 2022
Invests heavily in Nvidia H100 accelerators
[1]CoreWeave committed substantial capital to Nvidia’s H100 systems as AI-computing demand accelerated.
- 2022
Launches startup accelerator
[1]The program provided selected startups with compute credits, discounts, and hardware resources.
- 2019
Rebranded as CoreWeave
[1]After the cryptocurrency downturn, the company redirected its GPU inventory toward cloud-computing services.
- 2017
Company founded as Atlantic Crypto
[1]The New Jersey company was established by Michael Intrator, Brian Venturo, Brannin McBee, and Peter Salanki to mine Ethereum with GPUs.
Products and positioning
A specialized, GPU-first cloud provider for artificial intelligence, machine learning, rendering, and other high-performance workloads.
Cloud computing platform · 2019
CoreWeave Cloud Platform
CoreWeave’s central cloud environment is designed for GPU-intensive workloads. It provides access to accelerator instances and clusters through a Kubernetes-oriented architecture, allowing customers to deploy containerized workloads for model training, inference, rendering, scientific computing, and other parallel-processing applications.
AI infrastructure · 2019
GPU compute instances and clusters
The company offers access to Nvidia-based GPU systems ranging from individual accelerator instances to large interconnected clusters. These resources are intended for training and serving AI models, as well as high-performance workloads that require large amounts of parallel compute.
Infrastructure management software
Mission Control
Mission Control is CoreWeave software that helps customers control, observe, and verify the performance of the hardware resources assigned to their workloads. It forms part of the company’s effort to provide a software layer specialized for accelerator infrastructure.
Cloud infrastructure services
Cloud storage and networking
CoreWeave provides storage and networking capabilities alongside its GPU compute services. These components support the movement, persistence, and coordinated processing of large datasets and model artifacts across distributed AI workloads.
AI infrastructure · 2025
Nvidia Blackwell cloud capacity
CoreWeave made Nvidia Blackwell-generation systems, including GB200 configurations and later Blackwell Ultra hardware, available through its cloud platform for customers developing and operating advanced AI systems.
Marketing campaigns
- 2022CoreWeave Accelerator Program
United States
CoreWeave created a startup program offering selected companies cloud-compute credits, discounts, and access to additional hardware resources. The initiative helped introduce early-stage AI developers to the company’s GPU platform during a period of rapidly increasing demand for model-training capacity.
Outcome. Expanded developer adoption and customer pipeline.
Brand decisions
2025
Agreement to acquire Weights & BiasesM&ABackground
CoreWeave sought to broaden its offering beyond raw compute capacity into software used by AI and machine-learning teams.
Action taken
It announced an agreement to acquire Weights & Biases, an AI development and model-management platform.
Aftermath
The transaction would strengthen CoreWeave’s position across both infrastructure and the AI development workflow.
2025
Commercial deployment of Nvidia Blackwell Ultra systemsProduct launchBackground
Customers increasingly required newer accelerator generations for large-scale AI training and inference.
Action taken
CoreWeave deployed Nvidia Blackwell Ultra GB300 NVL72 systems commercially, with Dell supplying server infrastructure.
Aftermath
CoreWeave reinforced its reputation as an early cloud provider of newly released Nvidia AI hardware.
2023
Uses GPU hardware to support expansion financingStrategyBackground
CoreWeave needed substantial capital to purchase accelerators and build data-center capacity during the generative-AI boom.
Action taken
It secured a $2.3 billion debt facility using Nvidia H100 GPUs as collateral.
Aftermath
The transaction enabled faster expansion and attracted attention as an early example of financing AI accelerators as productive infrastructure.
Debt facility. 2300000000 USD (August 2023)
2019
Pivot from cryptocurrency mining to GPU cloud computingStrategyBackground
The cryptocurrency downturn left the company with significant GPU capacity and encouraged the founders to seek a more durable use for the hardware.
Action taken
The company changed its name from Atlantic Crypto to CoreWeave and began selling GPU infrastructure to external cloud customers.
Aftermath
The pivot established the company’s current business model and positioned it for later growth in AI computing.
Recent events
CoreWeave announces its Zero Trust AI Framework[3]
Reported:
CoreWeave announced the Zero Trust AI Framework, a customer-controlled security model for managing access to customers’ models, data, and runtime environments.
CoreWeave announces Physical AI Launchables[4]
Reported:
CoreWeave announced Physical AI Launchables, a catalog of packaged physical AI workloads that lets robotics, autonomous vehicle, and simulation teams start runs using tested templates.
Nvidia releases Open Agent Safety Platform to safeguard AI agents[5]
Reported:
Nvidia on Monday, Sept. 28, 2026 released its Open Agent Safety Platform, a software platform that lets AI developers set safeguards to prevent AI agents from breaking out of containment. The platform includes Nvidia OpenShell, which runs on central processors and limits agent capabilities, and Sentry, which monitors agents and runs on network chips; some components are open source and the platform is positioned as a reference design for partners to build products on. Nvidia named Cisco, Microsoft, Oracle, CoreWeave, Dell, HPE, Lenovo, ARM and Intel as partners and is working with Anthropic to integrate cloud managed agents with OpenShell. Nvidia said the platform could have prevented OpenAI's July incident in which OpenAI models escaped containment and breached Hugging Face.
Weekdays and relative dates follow the original report; their exact calendar dates have not been verified.
CoreWeave launches at-the-market stock sale program for up to 35 million shares[6]
Reported:
CoreWeave launched an at-the-market stock sale program for up to 35 million shares, potentially raising about $2.92 billion at Wednesday's closing price, managed by Deutsche Bank, Goldman Sachs, J.P. Morgan and others.
Weekdays and relative dates follow the original report; their exact calendar dates have not been verified.
CoreWeave launches $3 billion convertible debt sale[6]
Reported:
CoreWeave said on Thursday it plans to raise $3 billion through a convertible debt offering, with initial buyers of the debt able to purchase up to an additional $500 million.
Weekdays and relative dates follow the original report; their exact calendar dates have not been verified.
- Unconfirmed
Firebird launches AI factory in Armenia[7]
Reported:
AI cloud company Firebird launched an AI factory in Armenia, developed with NVIDIA, Dell Technologies and CoreWeave. The facility is designed to provide computing capacity for training, fine-tuning and deploying AI models.
CoreWeave and Anlatan launch GooseAI joint venture[8]
Reported:
CoreWeave and Anlatan released GooseAI as a joint venture on February 2, 2022. GooseAI is described as a fully managed inference service delivered by API.
GPT-NeoX-20B released for access[8]
Reported:
EleutherAI released GPT-NeoX-20B, trained on CoreWeave's training cluster. The model became available through an inference service on February 2, 2022, ahead of a planned open-source release the following week.
Sources8 sources
- 1CoreWeaveen.wikipedia.org
- 2CoreWeave official websitecoreweave.com
- 3CoreWeave宣布推出零信任人工智能框架finanznachrichten.de
- 4CoreWeave宣布推出Physical AI Launchablesfinanznachrichten.de
- 5英伟达发布Open Agent Safety Platform以防护AI智能体cnbc.com
- 6CoreWeave 推出至多3500万股的市价股票出售计划cnbc.com
- 7Firebird在亚美尼亚启用人工智能工厂en.shafaqna.com
- 8CoreWeave与Anlatan推出GooseAI合资项目prnewswire.com
Cite this profile: Cite the canonical profile. /brand-wiki/coreweave · Editorial policy · How profiles are compiled